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 * bug fix: closed a loophole in metacircularity detection and
13 grounding in the PCL implementation of CLOS.
14 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
15 (reported by tomppa on #lisp)
16 * bug fix: clear the direction flag on Lisp -> C transitions, as
17 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
18 (reported by Andrej Grozin and Hendrik Maryns)
19 * optimization: major improvements to GC efficiency on GENCGC platforms
20 * optimization: faster implementation of EQUAL
21 * optimization: emit more efficient opcodes for some common
22 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
23 * optimization: several other minor code-generation improvements on x86-64
24 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
26 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
27 * minor incompatible change: (SETF CLASS-NAME) and (SETF
28 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
29 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
30 by ANSI as a generic function, and for consistency (SETF
31 GENERIC-FUNCTION-NAME) is treated likewise.
32 * fixed bug #233.b: make constraint propagation notice when a variable
33 value is changed after it is referenced but before it is used
34 * fixed bug #296: no more arbitrary behaviour from filenames with
35 odd characters as --load arguments. (reported by Adam Warner)
36 * fixed bug #390: :CHARACTER-SET pathname components now work as
37 expected. (reported by Tim Daly Jr)
38 * fixed bug #391: complicated :TYPE intersections in slot
39 definitions no longer cause an error in PCL internals.
40 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
41 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
42 returning the number of octets which would be written to the
43 file-stream. (thanks to Robert J. Macomber)
44 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
45 arguments to the internal stream functions. (thanks to David
47 * bug fix: stack exhaustion is now handled on the x86/SunOS
48 platform. (thanks to Daisuke Homma)
49 * bug fix: more accurate ROOM results on GENCGC platforms
50 * optimization: improved type inference for arithmetic-for
51 index variables in LOOP
52 * optimization: faster floating-point SQRT on x86-64
53 * fixed some bugs revealed by Paul Dietz' test suite:
54 ** DOCUMENTATION returns NIL instead of "" for method combinations
55 that don't have a docstring
57 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
58 * minor incompatible change: (SETF CLASS-NAME) and (SETF
59 GENERIC-FUNCTION-NAME) are no longer generic functions, and
60 therefore are not customizeable by user code (as seems to be at
61 least permitted and maybe required by AMOP). As a consolation,
62 however, the SBCL implementation of these functions now calls
63 REINITIALIZE-INSTANCE as specified by AMOP.
64 * bug fix: slot-definition documentation is propagated as per ANSI
65 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
66 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
67 argument, as do eslotd initialization methods. (from Pascal
68 Costanza's "Closer" project)
69 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
70 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
72 * bug fix: it is now possible to have more than one subclass of
73 STANDARD-GENERIC-FUNCTION without causing stack overflow.
74 (reported by Bruno Haible, Pascal Costanza and others)
75 * bug fix: the dependent update protocol now works for generic
76 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
78 * bug fix: condition-class instances corresponding to
79 DEFINE-CONDITION forms are now created eagerly. (reported by
80 Kalle Olavi Niemitalo on comp.lang.lisp)
81 * bug fix: floating point printing is more accurate in some
82 circumstances. (thanks to Simon Alexander)
83 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
84 merged with *DEFAULT-PATHNAME-DEFAULTS*.
85 * bug fix: callbacks on OS X now preserve stack-alignment.
86 (thanks to Cyrus Harmon)
87 * enhancement: the x86-64 disassembler is much better at
88 disassembling SSE instructions. (thanks to Lutz Euler)
89 * enhancement: CHECK-TYPE on a variable now allows type-inference
90 (thanks to Vincent Arkesteijn)
91 * optimization: improved performance of EUC-JP external format.
92 (thanks to NIIMI Satoshi)
93 * optimization: performance improvements to IO on file streams of
94 :ELEMENT-TYPE CHARACTER
95 * optimization: much faster memory allocation on x86-64
96 * optimization: faster garbage collections (latency and throughput)
98 * optimization: faster bignum subtraction and division on x86 and x86-64
99 (thanks to Lutz Euler)
100 * optimization: more accurate type-inference for loops that use a
101 floating point index variable or a negative step.
103 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
104 * bug fix: add a workaround to SBCL looping infinitely at startup on
105 Linux kernels with apparently buggy implementations of personality().
106 (thanks to Svein Ove Aas)
107 * bug fix: Unicode symbols are correctly printed in LDB backtraces
108 (thanks to David Lichteblau)
109 * bug fix: local bindings shadow global symbol macros for
110 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
111 * optimization: non-open coded uses of numeric comparison operators
112 (e.g. >) no longer cons when called with more than one parameter
113 on platforms supporting dynamic-extent allocation.
114 * enhancement: saving cores with foreign code loaded is now supported
115 on MIPS/Linux in addition to the previously supported platforms.
116 * bug fix: FIND-RESTART now tests for activity, not applicability when given
117 a restart object as identifier. (reported by Helmut Eller for CMUCL)
118 * bug fix: division by zero in sb-sprof when no samples were collected
119 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
120 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
121 (thanks to Alexander Kjeldaas)
122 * new feature: ignoring the return values of destructive functions
123 like NREVERSE signals a compile-time style-warning.
124 (thanks to Kevin Reid)
125 * new feature: the alignment of alien structure fields can be
126 explicitly specified. (thanks to Cyrus Harmon)
127 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
128 and dump core on SIGQUIT
130 ** incompatible change: threads do not inherit values of specials
131 from their parents (see manual)
132 ** bug fix: threads stacks belonging to dead threads are freed by the
133 next exiting thread, no need to gc to collect thread stacks anymore
134 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
135 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
136 ** bug fix: dynamic variable and thread start related gc lossage
137 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
139 * fixed some bugs revealed by Paul Dietz' test suite:
140 ** SUBTYPEP is slightly more accurate on heinously complicated
141 CONS types where some of the members have uncertain (in the
142 NIL, NIL sense) type relationships to each other.
144 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
145 no longer purified unless :PURIFY T is explicitly specified.
146 ** Non-purified cores are significantly smaller than before
148 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
149 * new feature: timers based on Zach Beane's excellent timer package
150 * added support for the following external formats: koi8-u,
151 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
152 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
153 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
154 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
155 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
156 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
157 * incompatible change: a threaded SBCL will no longer revert to
158 non-threaded mode on non-NPTL systems, but refuse to start entirely.
159 * bug fix: interrupts are disabled until startup is complete; no
160 more sigsegvs when receiving a signal to soon
161 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
163 * bug fix: add a workaround for the memory randomization features in
164 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
165 workaround will only be in effect on systems with the proc filesystem
167 * bug fix: printing objects of type HASH-TABLE signals a
168 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
170 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
171 have been running for over 50 days. (reported by Gilbert Baumann)
172 * bug fix: the logic for getting names of functions gets less
173 confused when confronded with alternate-metaclass
174 funcallable-instances. (reported by Cyrus Harmon)
175 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
176 consistent, even on internal alternate-metaclass objects.
177 * bug fix: flush closure information collected by physical
178 environment analysis prepass before the main pass. (bug reported
180 * bug fix: compiler pack phase does not modify a hashtable, which is
181 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
182 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
183 may be expanded. (reported by Alexander <malishev> on c.l.l)
184 * bug fix: no more style-warnings when compiling code using LAST
185 under high-SPEED low-SPACE settings. (reported by David Wallin
187 * fixed some incompatibilities between SBCL's MOP and the MOP
189 ** the METAOBJECT class is now implemented;
190 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
191 STANDARD-OBJECT, as required;
192 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
193 now compatible (as required by VALIDATE-SUPERCLASS); there
194 remains a consistency requirement over the presence or absence
195 of the FUNCTION class in the superclass list of finalized
196 classes; see the manual for more details;
197 ** the :AROUND method for COMPUTE-SLOTS on
198 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
199 requested slot ordering.
201 ** bug fix: parent thread now can be gc'ed even with a live
203 ** bug fix: binding a special with PROGV to no value is not
204 the same as that symbol not having been bound (thanks to
206 ** bug fix: binding specials is thread safe (thanks to
208 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
209 and SB-SPROF do not die with 'no handler for signal XX in
210 interrupt_handle_now(..)' anymore
211 ** bug fix: WITH-TIMEOUT works with multiple running threads
212 * fixed some bugs revealed by Paul Dietz' test suite:
213 ** ENSURE-GENERIC-FUNCTION should take a method class object for
214 the :method-class keyword argument.
216 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
217 * new port: the Solaris operating system on x86 processors is now
218 mostly supported, though some rough edges in the environment
219 remain. (thanks to Daisuke Homma)
220 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
221 space for its heap. (thanks to Thiemo Seufer)
222 * optimization: SBCL on MIPS platforms now supports dynamic-extent
223 closures. (thanks to Thiemo Seufer)
224 * minor incompatible change: eof selects abort in the debugger.
225 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
226 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
228 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
229 deprecated, and will go away in a future revision of SBCL.
230 * minor incompatible change: GC-ON and GC-OFF are no longer
231 implemented with a counter, it does not matter how many times gc
232 is switched on or off
233 * bug fix: discriminating functions for generic function classes
234 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
235 make invalid assumptions about method precedence order. (reported
237 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
238 platforms. (thanks to Thiemo Seufer)
239 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
240 * bug fix: correct pathname printing with printer escaping is on.
241 (thanks to Kevin Reid)
242 * bug fix: complex VOP definitions in "user-space" no longer trigger
243 package locks. (reported by Zach Beane)
244 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
245 for funcallable-instance functions, and is no different from
247 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
248 characters in the current readtable. (reported by Nicholas Neuss)
249 * bug fix: SUBTYPEP on various CONS types returns more a more
250 accurate acknowledgment of its certainty.
251 * optimizations: REMOVE-DUPLICATES now runs in linear time on
252 lists in some cases. This partially fixes bug 384.
253 * flush all standard streams before prompting in the REPL and the
255 * bug fix: signal handling and triggering gc do not conflict
256 directly anymore, in particular a high frequency sb-sprof does
257 not prevent gc from running
258 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
259 approximation for timezone and DST information between the
260 universal time epoch and the smallest negative 32-bit time_t.
261 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
262 year 1899 so long as the encoded time is non-negative.
263 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
264 the correct answer for non-constant rotate amounts when performing
265 an inline 32-bit rotation.
267 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
268 there is only one thread in the session
269 ** bug fix: memory leak for streams created in one thread and
270 written to in another
271 ** bug fix: lockup when compiled with gcc4
272 ** bug fix: race that allows the gc to be triggered when gc is
274 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
275 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
277 ** bug fix: finished threads are gc'ed properly
278 * fixed some bugs revealed by Paul Dietz' test suite:
279 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
280 has arguments with a different set of applicable methods from
281 the orignal arguments.
282 ** The type error thrown by MAP now has a correct expected-type
284 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
285 name a compiled function.
286 ** Three MISC tests where a large bignum was improperly coerced to
287 a float that couldn't represent that bignum during type
288 derivation were fixed.
289 ** SUBTYPEP can now handle types involving the negation of a
290 list-form FUNCTION type.
291 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
292 as (COMPLEX (AND RATIO (NOT FIXNUM))).
293 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
295 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
296 * New feature: Experimental support for bivalent streams: streams
297 opened with :element-type :default now allow character and binary
298 (unsigned-byte 8) I/O
299 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
300 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
302 * bug fix: no more highly sporadic "couldn't check whether ~S is
303 readable" when reading a stream and an interrupt hits in the middle
304 of a select system call
305 * compiler better recognizes complex arrays (reported by Eduardo
307 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
308 for variables on SBCL built with linkage-tables. (reported by Luis
310 * various error reporting improvements.
311 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
312 (Thanks to Thiemo Seufer)
313 * bug fix: nested reader invokations work correctly
314 * bug fix: it is possible to have simultaneous references to foreign
315 code and foreign data with the same name.
317 ** added x86-64 support
318 ** incompatible change: the threading api now works with thread
319 objects instead of thread ids
320 ** bug fix: threads are protected from signals and interruption when
321 starting up or going down
322 ** bug fix: a race where an exiting thread could lose its stack to gc
323 ** bug fix: don't halt on infinite error in threads if possible
324 ** fixed numerous gc deadlocks introduced in the pthread merge
325 ** bug fix: fixed thread safety issues in read and print
326 ** bug fix: debugger doesn't hang on session lock if interrupted at
327 an inappropriate moment
328 ** bug fix: run-program is now thread safe(r)
329 ** bug fix: inner with-recursive-lock no longer releases the mutex
330 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
331 * fixed some bugs revealed by Paul Dietz' test suite:
332 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
333 EXPECTED-TYPE slots filled.
334 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
336 ** MAKE-CONDITION accepts classes as type-designators.
337 ** COMPILE may never return NIL.
338 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
339 range before calling Unix time functions
341 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
342 * numerous signal handling fixes to increase stability
343 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
344 * minor incompatible change: we now correctly canonize default
345 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
346 as per the MOP, rather than the historical (INITARG INITFUNCTION
347 INITFORM). (reported by Bruno Haible)
348 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
349 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
350 for more information.
351 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
352 pathname is a directory pathname.
353 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
354 * dynamic space size on PPC has been increased to 768Mb. (thanks to
356 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
357 :METACLASS argument in addition to a class name. (reported by
358 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
359 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
360 (reported by Bruno Haible; patch from Gerd Moellmann)
361 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
363 * bug fix: more cleanups to the floating point exception handling on
364 x86-64 (thanks to James Knight)
365 * bug fix: the compiler does not try to constant fold calls of
366 COERCE and BIT-* functions when they return freshly constructed
367 objects. (reported by Nikodemus Siivola and Paul Dietz)
368 * optimization: improved the MIPS versions of generic (in the
369 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
370 * optimization: direct conversion of (unsigned-byte 32) to floats on
371 the PowerPC platform.
372 * optimization: structure instances with raw slots now use less
373 memory, and probably show better memory locality. (thanks to
375 * optimization: DYNAMIC-EXTENT declarations for lists and closures
376 are treated as requests for stack allocation on the x86-64,
377 Alpha-32, PPC and SPARC platforms.
378 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
379 the wrong file descriptor; implementation of SOCKET-OPEN-P.
380 (thanks to Tony Martinez)
382 ** gcing a dead thread can no longer lead to lockups
383 ** threads block signals until they are set up properly
384 ** errno is no longer shared by threads
385 ** interrupt-thread restores the eflags register on x86
386 ** fixed some lockups due to gc/thread interaction
387 * fixed some bugs revealed by Paul Dietz' test suite:
388 ** invalid dotted lists no longer raise a read error when
390 ** signal an error if a symbol that names a declaration is used
391 as the name of a type, or vice versa
392 ** allow using the (declare (typespec var*)) abbreviation for
393 (declare (type typespec var*)) with all type specifiers
394 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
395 ** always return NIL from PPRINT-POP when OBJECT is NIL
396 ** don't signal errors when pretty-printing malformed LABELS,
397 FLET or MACROLET forms
398 ** declarations in a DOLIST body are also in scope for the
400 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
402 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
405 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
406 * fixed cross-compiler leakages that prevented building a 32-bit
407 target with a 64-bit host compiler.
408 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
409 opened with :IF-EXISTS :OVERWRITE.
410 * fixed bug 281: error for an invalid qualifier in a short-form method
411 combination method is not signalled until the faulty method is called.
412 * bug fix: iteration variable type inferrer failed to deal with open
413 intervals. (reported by Alan Shields)
414 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
415 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
416 or more runtime options were provided to the sbcl binary.
417 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
418 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
419 * optimization: there's now a fast-path for fixnum arguments in the
420 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
421 * optimization: the code generated on x86-64 is more compact thanks
422 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
423 * bug fix: floating-point exception handling now partly works on
424 x86-64. (Thanks to James Knight)
425 * improvement to the MIPS backend from Thiemo Seufer: C-style
426 64-bit long long arguments and return values to alien functions
428 * fixed some bugs revealed by Paul Dietz' test suite:
429 ** the type-error signalled from WARN has a filled-in DATUM slot.
430 ** the type-error required when a stream is not associated with
431 a file has the stream as its datum.
432 ** type-errors on single-floats on x86-64 no longer have
433 :INVALID-OBJECT as the datum
434 ** the type-errors signalled for invalid function names now have
435 a correct expected type
436 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
437 for structures defined with a :TYPE. Documentation strings for
438 typed structures are no longer immediately discarded
439 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
440 broadcast streams and synonym streams. FILE-LENGTH now also works
441 on broadcast streams.
443 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
444 * incompatible change: the --noprogrammer option, deprecated since
445 version 0.7.5, has been removed. Please use the equivalent
446 --disable-debugger option instead.
447 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
449 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
450 inoperational for a while now) has been completely removed.
451 * null lexical environments are now printed as #<NULL-LEXENV>,
452 significantly reducing the amount of clutter in typical backtraces.
453 * documentation on weak pointers, finalization, and after GC hooks
454 has been added to the manual.
455 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
456 as other specialized array types, is much faster. SUBSEQ and
457 COPY-SEQ on such arrays have also been sped up.
458 * optimization: EQL is now more efficient when at least other argument
459 is known to be of type (OR FIXNUM (NOT NUMBER)).
460 * fixed bug: the runtime is now less vulnerable to changes in the
461 size of the SBCL object on OS X, and virtual memory is reserved for
462 all spaces that need to be at a fixed address.
463 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
464 * fixed bug: finalizers and after GC hooks that cause consing are now
466 * fixed bug: compiler error messages and summaries are now printed to
467 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
468 * fixed inference of the upper bound of an iteration variable.
469 (reported by Rajat Datta).
470 * fixed bug 211e: calling local functions with duplicated constant
471 keyword argument no longer causes a bogus style warning about an
473 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
474 ftype declaration to be disregarded. (reported by Dave Roberts)
475 * fixed bug 373: caused by erronous compilation of references to alien
476 variables in the runtime on ppc/darwin.
477 * fixed bug 376: CONJUGATE type deriver.
478 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
479 (reported by Baughn on #lisp)
480 * a cross-compiler bug on non-x86 platforms has been identified and
481 fixed. (thanks to Bruno Haible)
482 * improvements to the MIPS runtime code for increased stability.
483 (thanks to Thiemo Seufer)
484 * increased the maximimum compact environment size to allow
485 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
486 * partial workaround for bug 135: don't name the function we're
487 calling for hairy cases of EVAL, so as not to accumulate one environment
488 entry per call to EVAL. (reported by Kevin Reid)
489 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
490 * optimization: type testing for non-vector arrays should be faster.
491 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
493 * bug fix: setting 31st element of a bit vector to zero did not work
495 * bug fix: redefining a class definition which failed due to a
496 previous accessor / function clash now works (but see BUGS entry
497 #380 for more problems in this area). (thanks to Zach Beane)
498 * the long form of DEFINE-METHOD-COMBINATION disables method group
499 checking when given a single method group with pattern *. (thanks
501 * on x86 compiler supports stack allocation of results of simple
502 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
503 * support for building and running on Mac OS X 10.4 "Tiger" has been added
504 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
505 * fixed some bugs related to Unicode integration:
506 ** the restarts for recovering from input and output encoding
507 errors only appear when there is in fact such an error to
509 * fixed some bugs revealed by Paul Dietz' test suite:
510 ** MISC.549 and similar: late transformation of unsafe type
511 assertions into derived types caused unexpected code
513 ** SCALE-FLOAT type deriver is less wrong.
514 ** type derivers for EXP, LOG and similar functions compute result
515 types for complex arguments better.
516 ** (MISC.563) CONJUGATE type deriver works for very restricted
518 ** out-of-line type testers for character strings are available.
519 ** EQUAL compiler transform understands specialness of objects
521 ** accessing double-floats stored on the stack now works on x86-64.
522 ** debugger internals could sometimes create invalid lispobjs,
523 resulting in GC crashes.
524 ** MISC.548: type check weakening can convert required type into
526 ** initialization forms for bindings are not in scope of free special
529 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
530 * incompatible change: thread support for non-NPTL systems has
531 been removed - locking is buggy and unreliable. A threaded
532 SBCL build will now warn at startup and refuse to create
533 new threads, unless futex support is detected in the kernel
534 * incompatible change: the top level REPL now has only an ABORT
535 restart associated with it, not TOPLEVEL and ABORT as it used to.
536 TOP and TOPLEVEL are now available as debugger commands for
537 returning to the top level.
538 * incompatible change: forms evaluated in the REPL now use the
539 global optimization policy.
540 * incompatible change: user- and system-initialization files are
541 no longer processed with LOAD, but by READ and EVAL; hence the
542 global optimization policy, startup package, readtable, etc,
544 * The .fasl file format number has been incremented because of
545 various incompatible changes.
546 * internal entry point details and argument counts no longer appear
547 in backtraces unless explicitly requested by setting
548 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
549 * built-in and standard functions no longer have names like "top
550 level local call to FOO".
551 * fixed bug 32: functions defined in non-null lexical environments
552 now have more legible printed representation
553 * fixed bug 33: functions defined in non-null lexical environemnts
554 are now more amenable to inspection by INSPECT.
555 * workaround for bug 354: XEPs no longer appear in backtraces unless
556 explicitly requested.
557 * fixed bug: receiving the signal which results from attempting to
558 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
559 SIGBUS on other *BSDs) on architectures where the C stack is also
560 the Lisp stack (x86 and x86-64) from foreign code no longer leads
561 to debugger confusion or wild instability. (reported by Cheuksan
563 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
564 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
565 (reported by Lutz Euler)
566 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
567 compiled code. (reported by Paul Dietz)
568 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
569 specialized parameter to a method to something that is not TYPEP
570 the specializer is now possible.
571 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
572 face of package deletion.
573 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
574 pathname. (reported by David Tolpin on c.l.l)
575 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
576 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
577 STRING-STREAM and FILE-STREAM, respectively.
578 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
579 than previously; in addition, its test suite is now run on build.
580 (thanks to Luke Gorrie)
581 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
582 * optimization: added a immediate representation for single-floats
584 * fixed some bugs revealed by Paul Dietz' test suite:
585 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
586 ** attempting to create a package with a colliding nickname causes
587 correctable errors to be signalled.
588 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
589 ** several x86-64 backend bugs related to sign-extension of immediate
592 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
593 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
594 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
596 * fixed loading of multiply forward-referenced layouts.
597 (thanks to Cheuksan Wang)
598 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
599 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
600 (see COMPILE-FILE documentation for details of :PRINT option)
601 * fixed bugs 19 and 317: fixed-format floating point printing is
602 more accurate. This also fixes a bug reported by Adam Warner
603 related to the ~@F format directive.
604 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
606 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
607 dispatch table if the from-char is a dispatch macro character.
608 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
609 by the wordsize no longer ignore the last word. (reported by Lutz
611 * fixed bug in type checking of dynamic-extent variables. (reported
613 * optimization: sequence traversal functions use their freedom to
614 coerce function designators to functions.
615 * optimization: code with many calls to CLOS methods specialized on
616 CLOS classes has had redundant type checks removed.
617 * optimization: added declarations to speed up operations that access
618 the internal character database (for example STRING-UPCASE)
619 * optimization: comparison operations between floats and sufficiently small
620 fixnums no longer create extra rationals
621 * fixed some bugs related to Unicode integration:
622 ** portions of multibyte characters at the end of buffers for
623 character-based file input are correctly transferred to the
624 start of the buffer at the next read.
625 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
626 passing it through to OPEN.
627 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
628 argument given to internal calls to OPEN.
629 ** the built-in comment readers (introduced by character sequences
630 ";" and "#|") are more forgiving to encoding errors; they will
631 STYLE-WARN and then attempt to resync the stream at a character
632 boundary. (thanks to Teemu Kalvas)
633 * fixed some bugs in the x86-64 port:
634 ** Negative short int return values from c-calls are sign-extended
636 ** The stack is aligned to 16-bytes for c-calls, as required by
637 the ABI. (thanks to Cheuksan Wang)
638 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
639 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
640 for enums. (thanks to Vincent Arkesteijn)
641 ** Multiple small optimizations and bugfixes for floating point
643 * fixed some bugs revealed by Paul Dietz' test suite:
644 ** Space, Tab, Linefeed, Return and Page have the invalid
645 secondary constituent character trait.
646 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
648 ** WITH-INPUT-FROM-STRING should only update the index place on
650 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
652 ** Bit-array operations (BIT-AND and similar) worked incorrectly
653 with one-dimensional arrays with fill pointers.
654 ** TYPE-OF failed on a complex with an integer realpart and a
656 ** compiler failure during type inference for the code of form
657 (IF (EQL X (THE ...)) ...) (MISC.535).
659 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
660 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
661 platform. The port supports 61-bit fixnums, large memory spaces
662 and reloading shared object files.
663 * enhancement: saving cores with foreign code loaded is now
664 supported on ppc/Darwin in addition to the previously supported
666 * enhancement: the statistical profiler now walks deeper into the
667 call stack for more meaningful call-graphs and accrued time
668 reports (x86/x86-64 only). It also now reports time spent in
670 * enhancement: it is now possible to trace most individual methods
671 of a generic function in addition to tracing the generic function
673 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
675 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
676 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
677 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
678 produces an error. (thanks to Vincent Arkesteijn)
679 * bug fix: NAMESTRING on pathnames with :WILD components in their
680 directories works correctly. (thanks to Artem V. Andreev)
681 * fixed bug 125: compiler preserves identity of closures. (reported
683 * bug fixed: functions with &REST arguments sometimes failed with
684 "Undefined function" when compiled with (DEBUG 3). (reported by
686 * bug fix: overflow during compiling of setting element of a bit
687 vector with constant index and value. (reported by Timmy Douglas)
688 * build fix: fixed the dependence on *LOAD-PATHNAME* and
689 *COMPILE-FILE-PATHNAME* being absolute pathnames.
690 * on x86 compiler partially supports stack allocation of dynamic-extent
692 * GO and RETURN-FROM do not check the extent of their exit points
693 when compiled with SAFETY 0.
694 * fixed some bugs related to Unicode integration:
695 ** encoding and decoding errors are now much more robustly
696 handled; it should now be possible to recover even from invalid
697 input or output to the terminal. (thanks to Teemu Kalvas)
698 ** provided a first cut at implementing STRING-TO-OCTETS and
699 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
700 ** altered the SB-MD5 contributed module slightly, changing the
701 interface just enough for it to be supportable for builds where
702 lisp characters are not eight bits.
703 * fixed some bugs revealed by Paul Dietz' test suite:
704 ** the FORMATTER-generated functions for ~V[ conditionals require
705 the correct number of arguments.
706 ** READ-FROM-STRING returns the mandated second value when applied
707 to displaced strings.
708 ** the #\Rubout and #\Backspace characters are treated as invalid
709 constituent characters by the tokenizer.
711 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
712 * new feature: reloading changed shared object files with
713 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
714 * new feature: references to foreign variables and functions
715 can now be compiled and loaded before the corresponding shared
716 object file is loaded, as long as the foreign definitions are
717 available at runtime.
718 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
719 supported. (thanks to Dan Debertin)
720 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
721 just keywords. (thanks to Vincent Arkesteijn)
722 * fixed bug #331: structure-class instances corresponding to
723 DEFSTRUCT forms are now created eagerly.
724 * fixed bug #345: backtraces from calls to undefined functions work
725 on x86 as well. Related bug #61 is now also partially fixed on x86
726 (backtraces from throws to unknown catch tags.)
727 * bug fix: lambda-list parsing is now stricter vrt. order and number
728 of lambda-list keywords.
729 * bug fix: as specified by AMOP, an error is signalled if a
730 class-option appears multiple times in a DEFCLASS form. (reported
732 * bug fix: RANDOM can be compiled when the compiler derives the type
733 of its numeric argument as a disjoint set of small integers.
734 (reported by Paul Dietz)
735 * bug fix: starting a core saved with shared objects loaded when
736 those objects are not available no longer causes threaded SBCL to
737 hang. (reported by Sean Ross)
738 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
740 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
741 argument, which must always be NIL. (reported by Kalle Niemitalo)
742 * bug fix: printing 1.0d+23 no longer results in an error.
743 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
744 * bug fix: structure slot setters preserve evaluation order. (thanks
746 * fixed some bugs related to Unicode integration:
747 ** RUN-PROGRAM can allow its child to take input from a Lisp
748 stream. (reported by Stefan Scholl)
749 ** ASDF-INSTALL successfully downloads .tar.gz files in all
750 locales. (reported by Ken Causey)
751 * fixed some bugs revealed by Paul Dietz' test suite:
752 ** INCF, DECF and REMF evaluate their place form as specified in
754 ** FORMATTER expands ~{ ~} iteration directives with V or #
755 parameters correctly.
756 ** FORMATTER deals with the ~@[ ~] conditional directive where the
757 consequent uses no arguments correctly.
758 ** the system has a partial understanding of the (COMPLEX RATIO)
760 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
763 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
764 * new feature: a build-time option (controlled by the :SB-UNICODE
765 keyword feature, enabled by default) for building the system with
766 support for the entire 21-bit character space defined by the
768 * new feature: the system now has rudimentary external-format
769 support; the primary user-visible change at this time is that
770 characters with the high bit set (such as lower-case-e-acute) will
771 print correctly to a terminal in a UTF-8 environment.
772 * minor incompatible change: BASE-CHAR no longer names a class;
773 however, CHARACTER continues to do so, as required by ANSI.
774 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
775 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
776 has been moved to the SB-EXT package (temporarily re-exported from
778 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
779 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
781 * fixed bug #351: better error-handlind and reporting for malformed
783 * fixed bug #350: bignum-printing is now more memory-efficient,
784 allowing printing of very large bignums, eg. (expt 2 10000000).
785 (reported by Bruno Haible)
786 * fixed bug #302: better primitive-type selection for intersection
788 * fixed bug #308: non-graphic characters now all have names, as
789 required. (reported by Bruno Haible)
790 * bug fix: redefining a class with different superclasses now correctly
791 removes it from the direct-subclasses of its previous superclasses.
792 (reported by David Morse)
793 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
794 new value now works. (reported by Bruno Haible)
795 * bug fix: correct canonicalization of multiple non-standard slot
796 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
797 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
798 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
799 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
800 now exists, an signals an error.
801 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
802 during macroexpansion for non-iteration variables as well. (reported
803 by Bruno Haible for CMUCL)
804 * bug fix: Cyclic structures and unprintable objects in compiler
805 messages no longer cause errors. (reported by Bruno Haible)
806 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
807 and READ-FROM-STRING all now return a primary value of NIL if
808 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
809 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
810 (reported by Bruno Haible for CMUCL)
811 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
812 do not designate a condition. (reported by Bruno Haible for
814 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
815 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
816 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
817 permitting resolution in favour of any of the conflicting symbols.
818 (reported by Bruno Haible for CMUCL)
819 * bug fix: EQUAL compiler optimizations is less aggressive on
820 strings which can potentially compare true despite having distinct
821 specialized array element types.
822 * bug fix: unit enumerations can be defined without dividing by
823 zero. (thanks to Vincent Arkesteijn)
824 * FORMAT compile-time argument count checking has been enhanced.
825 (report from Bruno Haible for CMUCL)
826 * a partial workaround for the bug 262: the compiler does not try to
827 inline-expand a local function doing RETURN-FROM from a deleted
828 BLOCK. (thanks to Peter Denno for the bug report and to David
829 Wragg for the simple test case)
830 * fixed some bugs revealed by Paul Dietz' test suite:
831 ** RENAME-PACKAGE allows all package designators as new package
833 ** constraint propagation and lambda variable substitution are
834 more cautious in dealing with partially deleted code.
835 ** compiler sometimes lost reoptimization passes.
836 ** CERROR, given a condition as condition designator, passes its
837 remaining arguments to the continue format control without
839 ** Case-altering FORMAT directives work correctly on non-ASCII
841 ** The REAL type specifier handles bounds outside the floating
842 point ranges without signalling FLOATING-POINT-OVERFLOW.
843 ** Functions with IR1-transformations can create intercomponent
844 references to global functions.
845 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
847 ** FORMAT ~R treats a nil value for its first parameter correctly.
849 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
850 * enhancement: saving cores with foreign code loaded is now
851 supported on x86/NetBSD and sparc/Linux in addition to the previously
853 * bug fix: on some platforms repeated installations caused multiple
854 copies of HTML documentation to be installed -- should not happen
855 any more. (reported by Stefan Scholl)
856 * bug fix: parsing self-recursive alien record types multiple times
857 no longer causes infinite recursion. (reported by Thomas F. Burdick,
858 original patch by Helmut Eller for CMUCL)
859 * bug fix: stack-exhaustion detection works now on NetBSD as well.
860 (thanks to Richard Kreuter)
861 * bug fix: defining classes whose accessors are methods on existing
862 generic functions in other (locked) packages no longer signals
863 bogus package lock violations. (reported by François-René Rideau)
864 * bug fix: special variables as DEFMETHOD parameters no longer have
865 associated bogus type declarations. (reported by David Wragg and
867 * bug fix: read-write consistency on streams of element-type
868 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
869 * bug fix: redefiniton of the only method of a generic function with
870 no DEFGENERIC no longer emits a full WARNING. In addition,
871 redefinition of generic functions with no DEFGENERIC to an
872 incompatible lambda list now signals an error. (thanks to Zach
874 * bug fix: DEFGENERIC now works even when there's a function of the
875 same name in an enclosing lexical environment. (thanks to Zach
877 * fixed compiler failure, caused by instrumenting code during
878 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
879 * optimization: added loop analysis and improved register allocation
880 * fixed some bugs revealed by Paul Dietz' test suite:
881 ** POSITION on displaced vectors with non-zero displacement
882 returns the right answer.
883 ** (SIMPLE-STRING) is a valid type specifier for sequence
885 ** *PRINT-LEVEL* handling for slotless structures is pedantically
887 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
888 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
890 ** The justification version of the FORMAT ~< directive treats
891 non-zero minpad parameter correctly.
893 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
894 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
895 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
896 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
897 the supported interface.
898 * new feature: Single-stepping of code compiled with DEBUG 2 or
899 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
900 * new feature: saving cores with foreign code loaded is now
901 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
902 Timothy Moore's work for CMUCL)
903 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
904 parameters to * instead of NIL if no initform is supplied.
905 (reported by Johan Bockgaard)
906 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
907 correctly when FUNCALL appears as the car of the form. Note:
908 despite this FUNCALL forms are not currently subject to
909 compiler-macro expansion. (port of Raymond Toy's fix for the
910 same from CMUCL, reported by Johan Bockgaard)
911 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
912 (thanks for Teemu Kalvas)
913 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
914 case instead of terminating the iteration (thanks for Julian
915 Squires, Sean Champ and Raymond Toy)
916 * bug fix: incorrect expansion of defgeneric that caused a style
917 warning. (thanks for Zach Beane)
918 * bug fix: slot accessor effective method computation works properly
919 for classes with multiple non-standard applicable methods on
920 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
921 * on x86 compiler supports stack allocation of results of LIST and
922 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
923 CMUCL implementation by Gerd Moellmann)
924 * fixed some bugs revealed by Paul Dietz' test suite:
925 ** FORMAT strings with both the ~<~:;~> form of the justification
926 directive and pretty-printing directives cause an error.
928 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
929 * incompatible change: the internal functions
930 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
931 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
932 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
933 instead of the old functions.
934 * new feature: on platforms where "dladdr" is available foreign
935 function names now appear in backtraces. (based on Helmut Eller's
937 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
938 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
940 * documentation: ASDF manual is now distributed with SBCL.
941 * bug fix: SBCL can now load its contributed modules with REQUIRE
942 even if the system-provided entries have been removed from
943 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
945 * bug fix: backtraces involving undefined functions or assembly
946 routines are more informative. (thanks to Brian Downing)
947 * bug fix: mutually referent alien structures now work correctly.
948 (reported by Rick Taube)
949 * bug fix: structures defined by WITH-ALIEN can be referred to
950 within other definitions in the same WITH-ALIEN.
951 * bug fix: division operators (MOD, TRUNCATE and the like) with
952 constant zero divisors and integer dividends no longer generate
954 * bug fix: provide default methods for INPUT-STREAM-P and
955 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
956 * bug fix: improve the detection and resolution of MOP metacycles
957 (where effective method computation depends on the generic function
958 having its effective method computed). (reported by Bruno Haible)
959 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
960 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
961 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
962 * bug fix: pathnames with relative directory components can be
963 represented relative to default pathnames.
964 * optimization: in taking the GCD of bignums, reduce the two bignums
965 to approximately the same size (using Euclid's algorithm) before
966 applying the more sophisticated binary GCD. (thanks to Juho
968 * optimization: COUNT on bitvectors now operates word-at-a-time.
969 * optimization: ASH with a positive, but not necessarily constant,
970 (leftwards) shift, when in a modular context, is compiled to a
972 * fixed some bugs revealed by Paul Dietz' test suite:
973 ** FORMAT variable parameters ("~V<char>") are defaulted properly
974 if the corresponding argument is NIL.
975 ** FORMAT directives accepting a minpad parameter treat negative
977 ** literal commas as character parameters to FORMAT directives are
979 ** literal spaces directly after ~<Newline> directives within a
980 format-logical-block (~:< ~@:>) do not induce :FILL-style
981 conditional newlines.
982 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
983 as specified: it no longer includes conditional newlines).
984 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
986 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
987 :PER-LINE-PREFIX argument does not evaluate to a string.
989 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
990 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
991 the manual for details; for now, package locks can be disabled by
992 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
993 if no major problems are found then it is likely that they will be
994 compiled in unconditionally.
995 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
996 unsupported operators on all platforms. To load a shared library
997 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
998 object file, link it into a shared library outside of SBCL and
999 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1001 * bug workaround: ROOM T and the GC were not working together
1002 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1003 reporting facility aren't preserved. That reporting has been
1004 disabled, so now until and unless someone figures out how to make it
1005 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1006 * minor incompatible change: as threatened around sbcl-0.8.0, the
1007 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1008 The symbols are also exported from SB-PCL for backwards
1009 compatibility, but more so than before SB-PCL should be treated as
1010 an implementation-internal package.
1011 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1013 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1014 It is known as of this release not to work on the Alpha, however.
1015 * fixed bug #167: errors signalled due to illegal syntax in method
1016 bodies are now more legible.
1017 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1018 designators and can hence be used with TYPEP.
1019 * fixed bug #333: CHECK-TYPE now ensures that the type error
1020 signalled, if any, has the right object to be accessed by
1021 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1022 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1023 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1025 * fixed bug #334: programmatic addition of slots using specialized
1026 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1027 and :ALLOCATION :CLASS slots.
1028 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1029 float exponents. (rereported by Peter Seibel)
1030 * fixed a bug: #\Space (and other whitespace characters) are no
1031 longer considered to be macro characters in standard syntax by
1032 GET-MACRO-CHARACTER.
1033 * fixed bug: initialization of condition class metaobjects no longer
1034 causes an instance of the condition to be created. (reported by
1036 * fixed bug: it is now possible to have slots such that
1037 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1038 description is neither :INSTANCE nor :CLASS.
1039 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1040 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1041 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1042 non-standard SLOT-DEFINITION-ALLOCATION values.
1043 * fixed bugs: various race conditions handling exiting threads.
1044 CL-PPCRE's thread torture test now passes.
1045 * fixed arguably-a-bug: GC time proportional to number of threads in
1046 system even when most of them are idle
1047 * optimization: improved performance of BIT and SBIT on bit-vectors.
1048 * .fasl file incompatibility: The fasl file version number has
1049 been incremented because of changes associated with package locks.
1051 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1052 * minor incompatible change: the system no longer provides
1053 optimization hints (conditions which are TYPEP
1054 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1055 mode; these hints are still emitted when higher SPEED optimization
1057 * new contrib module: a sampling profiler (profiling by statistical
1058 sampling, rather than by instrumenting functions) is available as
1059 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1060 * the behaviour of the standard function ED is now customizeable by
1061 third parties through a hook variable: see ED's documentation
1062 string for information on the protocol.
1063 * the compiler no longer emits efficiency notes for (FUNCALL X)
1064 when the type of X is uncertain under default optimization
1066 * fixed bug 276: mutating a binding of a specialized parameter to a
1067 method to something that is not TYPEP the specializer is now
1069 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1070 exhibit the required behaviour on the x86 platform. (thanks to
1071 Peter van Eynde, Eric Marsden and Bruno Haible)
1072 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1073 even for difficult arguments. (reported by Peter Graves)
1074 * fixed bug 141a: the backquote printer now descends quoted
1076 * fixed another bug in backquote printing: no more destructive
1077 modification of the form's list structure. (reported by Brian
1079 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1080 move between its address being taken and the call to
1081 interrupt_thread, fixing a crashing race condition.
1082 * the SB-POSIX contrib implementation has been adjusted so that it
1083 no longer exhibits ridiculously poor performance when constructing
1084 instances corresponding to C structs.
1086 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1087 * minor incompatible change: the sb-grovel contrib now treats C
1088 structures as alien (in the sense of SB-ALIEN) objects rather than
1089 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1090 has implications for memory management of client code
1091 (sb-grovel-returned objects must now be manually managed) and for
1092 type safety (alien objects now have full types).
1093 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1094 used to control emission of compiler diagnostics, rather than the
1095 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1096 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1097 quality should be considered deprecated.
1098 * install.sh now installs the user manual as well
1099 * (not quite a new documentable feature, but worth considering in
1100 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1101 beginnings of a semantically meaningful condition hierarchy is
1102 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1104 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1105 designator as the defaults argument.
1106 * fixed bug: Displaced arrays whose displaced-to array has become
1107 too small now cause ARRAY-DIMENSION to signal an error, providing
1108 for safer bounds-checking. (reported by Bruno Haible)
1109 * fixed bug: DEFCLASS slot definitions with identical :READER and
1110 :WRITER names now signal a reasonable error. (reported by Thomas
1112 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1114 * fixed bug: Invalid slot specification errors now print correctly.
1115 (thanks to Zach Beane)
1116 * fixed bug 320: Shared to local slot value transfers in class
1117 redefinitions now happen corrently from superclasses as
1118 well. (reported by Bruno Haible)
1119 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1121 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1122 improper lists correctly. (reported by Bruno Haible)
1123 * fixed bug 313: source-transform for <fun-name> was erroneously
1124 applied to a call of a value of a variable with name <fun-name>.
1125 (reported by Antonio Menezes Leitao)
1126 * fixed bug 307: The obsolete instance protocol ensures that
1127 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1128 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1129 remove dead unknown-values globs from the middle of the stack.
1130 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1131 easier to resume long computations after using *BREAK-ON-SIGNALS*
1132 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1133 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1134 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1135 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1136 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1137 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1139 * fixed a (fairly theoretical) bug in string printing: if
1140 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1141 does not have array-element-type equal to the most general string
1143 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1144 function name. (thanks to Nikodemus Siivola)
1145 * fixed bug: compile-time format string checker failed on
1146 non-closed ~{. (reported by Thomas F Burdick)
1147 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1148 don't warn on legal loop code involving more than one aggregate
1149 boolean. (thanks to Nikodemus Siivola)
1150 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1151 in keyword position to MAKE-PACKAGE was not being reported as
1153 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1154 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1156 * fixed bugs 280 and 312: the checking for multiple definitions in a
1157 file is less likely to become confused by uses of inline
1159 * fixed bug: the #S reader macro performs the keyword coercion
1160 specified for slot names. (reported by Kalle Niemitalo)
1161 * fixed bug: lambda lists may contain symbols whose names start with
1162 & but are not lambda-list-keywords; their occurrence triggers a
1164 * fixed bug 321: define-method-combination argument lambda lists do
1165 not cause invalid code to be generated when &optional or &aux
1166 variables have default values. (reported by Bruno Haible)
1167 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1168 have CLOS classes; this appears also to have fixed reported
1169 instability in stack exhaustion detection.
1170 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1171 MUFFLE-WARNING when no associated restart is present is now
1173 * optimization: rearranged the expansion of various defining macros
1174 so that each expands into only one top-level form in a
1175 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1177 * optimization: used a previously little-used slot in symbols to
1178 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1180 * fixed some bugs revealed by Paul Dietz' test suite:
1181 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1183 ** Fixed an optimization bug related to inheritance of initforms
1184 from local to shared slots.
1185 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1186 ** CAST optimizer forgot to flush argument derived type.
1187 ** print/read consistency on floats is now orders of magnitude
1188 more likely. (thanks also to Bruno Haible for a similar report
1190 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1191 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1192 for objects of type REAL. Make it so.
1193 ** SXHASH is specified (infelicitously) to respect similarity,
1194 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1195 it so. (thanks to Markus Ziegler)
1196 ** on the Alpha, the compiler succeeds in compiling functions
1197 returning a known number of arguments greater than 63.
1198 ** fixed handling of invalid NIL arguments in keyword position
1200 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1202 ** provided a readably-printable representation for RANDOM-STATE
1204 ** ensured that pathnames, where they have a namestring, always
1205 print using #P"..." syntax.
1207 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1208 * Support for the forthcoming 2.0 version of the NetBSD kernel
1209 running on x86 hardware has been added. (thanks to Perry
1210 E. Metzger most immediately, and others for their past work)
1211 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1212 current release version; SBCL's OpenBSD support had been broken
1213 since about the time of OpenBSD's switch to ELF binary format.)
1214 * [placeholder for DX summary]
1215 ** user code with &REST lists declared dynamic-extent, under high
1216 speed or space and low safety and debug optimization policy.
1217 * The manual has been converted to Texinfo format and the debugger
1218 chapter from the cmucl manual has been added.
1219 * A facility has been added to extract documentation strings from
1220 sbcl and store them as Texinfo-formatted snippets for inclusion in
1221 the manual (via Texinfo's @include directive)
1222 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1223 the test case to Dave Roberts)
1224 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1225 pointers. (reported by Sean Ross)
1226 * bug fix: PROFILE output is printed nicely even for large numerical
1227 values. (thanks to Zach Beane)
1228 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1229 greater than 32 handle EOF correctly.
1230 * bug fix: on X86 an immediate argument of the IMUL instruction is
1231 correctly printed in disassembly. (thanks to Lutz Euler)
1232 * bug fix: class slots in redefined classes preserve their old
1233 values. (thanks to Bruno Haible and Nikodemus Siivola)
1234 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1235 succeeds. (reported by Marco Baringer)
1236 * fixed some bugs revealed by Paul Dietz' test suite:
1237 ** READ-SEQUENCE now works on ECHO-STREAMs.
1238 ** RATIONALIZE works more according to its specification. (thanks
1241 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1242 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1243 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1244 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1245 should matter to you only if you rebind the printer control
1246 variables and then find you want different bindings in the
1247 debugger than in the ordinary execution of your program.)
1248 * The runtime build system has been tweaked to support building
1249 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1250 assembler and linker. (thanks to Nikodemus Siivola)
1251 * Unbound, undefined, undeclared variables now trigger full
1252 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1253 more often programmer error than deliberate exploitation of undefined
1255 * optimization: the hash algorithm for strings has changed to one
1256 that is less vulnerable to spurious collisions. (thanks to Juho
1258 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1259 needless bounds checking. (thanks to Juho Snellman)
1260 * optimization: implemented multiplication as a modular
1261 (UNSIGNED-BYTE 32) operation on the PPC backend.
1262 * fixed some bugs revealed by Paul Dietz' test suite:
1263 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1265 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1266 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1267 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1269 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1271 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1273 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1274 their output stream on EOF from read.
1275 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1276 have been read to end-of-file.
1277 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1279 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1280 description of determination of which consecutive characters
1282 ** Printing the "Space" character with escaping on now yields "#\\ ",
1283 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1284 ** Reading floating-point numbers with *READ-BASE* set to a number
1285 less than 10 works correctly.
1286 ** Reading floating-point numbers with *READ-BASE* set to a number
1287 more than 10 works correctly.
1288 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1289 the readtable currently in effect.
1291 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1292 * minor incompatible change: parsing of namestrings on a physical
1293 (Unix) host has changed; numbers after the final #\. in a
1294 namestring are no longer interpreted as a version field. This is
1295 intented to be largely invisible to the user, except that the
1296 meaning of the namestring "*.*.*" has changed: it now refers to a
1297 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1298 should usually be replaced by
1299 (make-pathname :name :wild :type :wild :version :wild)
1300 with the added benefit that this is more likely to be portable.
1301 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1302 option to OPEN now signals an error if the file being opened
1303 exists; this may have an impact on existing code.
1304 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1305 signals received, on the PowerPC platforms (both Linux and
1306 Darwin). (thanks to Pierre Mai for pointing out the location of
1308 * several fixes on OS X: The system now builds and runs cleanly on
1309 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1310 (thanks to Brian Mastenbrook)
1311 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1312 second-resolution: integer multiples of 1/3600 between -24 and 24.
1313 (thanks to Vincent Arkesteijn)
1314 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1315 type when called with 1 argument; PEEK-CHAR checked type of
1316 PEEK-TYPE only after having read first character from a
1317 stream. (reported by Peter Graves)
1318 * bug fix: the garbage collector now has much better locality
1319 behaviour, and in particular no longer treats all memory as being
1320 exhausted when many small objects point to each other in a deeply
1322 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1323 recognized as being TYPEP their class.
1324 * bug fix: the PUSHNEW documentation string has been corrected.
1325 (thanks to Vincent Arkesteijn)
1326 * bug fix: defaulting of the value for the last of an atypically
1327 large number of multiple values being bound was not being
1328 performed correctly on the Alpha or PPC platforms
1329 * optimization: implemented multiplication as a modular
1330 (UNSIGNED-BYTE 32) operation on the x86 backend.
1331 * optimization: SEARCH on simple-base-strings can now be open-coded.
1332 (see also contrib/compiler-extras.lisp for inspiration for
1333 teaching the compiler about the Boyer-Moore algorithm).
1334 * value, returned by MAX (and MIN) called with several EQUALP, but
1335 not EQL, arguments now does not depend on compiler settings.
1336 * fixed some bugs revealed by Paul Dietz' test suite:
1337 ** in stack analysis liveness information is propagated from
1338 non-local entry points.
1339 ** pathwise CAST removing failed when the CAST node did not start
1341 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1342 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1344 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1345 host is already defined.
1346 ** RENAME-FILE works on streams instead of signalling an internal
1348 ** PEEK-CHAR uses the current readtable when determining whether
1349 or not a character is whitespace.
1350 ** MERGE-PATHNAMES handles the case when the pathname does not
1351 specify a name while the default-pathname specifies a version.
1352 ** Pathnames now stand a better chance of respecting print/read
1354 ** Attempting to use standardized file system operators with a
1355 pathname with invalid :DIRECTORY components signals a
1357 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1359 ** DIRECTORY on logical pathnames is more correct.
1360 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1361 signal a TYPE-ERROR if their argument is not a stream.
1362 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1363 designator argument does not designate a stream.
1364 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1365 examining the synonym.
1366 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1368 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1369 element-type larger than ([UN]SIGNED-BYTE 32).
1371 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1372 * When built with the :SB-FUTEX feature, threaded builds now take
1373 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1374 for faster/more reliable mutex and condition variable support.
1375 * Incompatible change (but one you probably shouldn't have been using
1376 anyway): the interface and code for arbitrating between multiple
1377 threads in the same user session has been redesigned.
1378 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1379 error when called without an explicit environment argument.
1380 (thanks to Nikodemus Siivola)
1381 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1382 confused about their position. (thanks to Adam Warner and Gerd
1384 * bug fix: Pretty printing backquoted forms with unquotations in the
1385 argument list position of various code constructs such as LAMBDA
1386 now works correctly. (reported by Paul Dietz)
1387 * bug fix: Pretty printing unquotations no longer loses all
1388 stream position information.
1389 * optimization: performance of string output streams is now less
1390 poor for multiple small sequence writes.
1391 * optimization: performance of CSUBTYPEP in the presence of complex
1392 expressions involving CONS and NOT many times has been improved.
1393 (reported by Paul Dietz)
1394 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1396 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1397 streams for the REPL, for the debugger, and so on. (thanks to
1399 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1400 * fixed some bugs revealed by Paul Dietz' test suite:
1401 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1402 is ignored if it is too small, rather than propagating through
1403 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1404 ** extremely complex negations of CONS types were not being
1405 sufficiently canonicalized, leading to inconsistencies in
1407 ** VALUES tranformer lost derived type.
1409 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1410 * fixed a bootstrapping bug: the build process no longer assumes
1411 that the various BOOLE-related constants have the same value in
1412 host and target lisps. (noted by Paul Dietz' test suite on an
1413 SBCL binary built from CLISP)
1414 * The system can now be dynamically linked on the MIPS platform,
1415 which enables dynamic loading of foreign code from Lisp. (thanks
1416 to Ralf Baechle for discussions on the MIPS ABI)
1417 * The system now records debugging information for its own source
1418 files in a filesystem-position-independent manner, relative to
1419 the "SYS" logical host.
1420 * fixed a compiler bug: MV-LET convertion did not check references
1421 to the "max args" entry point. (reported by Brian Downing)
1422 * tweaked disassembly notes to be less confident about proclaiming
1423 some instruction as an LRA. (thanks to Brian Downing)
1424 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1425 now each have their own history, command character, and other
1426 characteristics. (thanks to David Lichteblau)
1427 * fixed some bugs revealed by Paul Dietz' test suite:
1428 ** compiler failure in compiling LOGAND expressions including a
1430 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1431 shift greater than 32.
1432 ** FLUSH-DEST did not mark blocks for type check regeneration.
1433 ** HANDLER-CASE failed to accept declarations in handler clauses
1434 in some circumstances.
1436 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1437 * New code in contrib/sb-introspect (still probably not entirely
1438 stable yet) provides some support for smart Lisp development
1439 environments like SLIME.
1440 * The conditions signalled for errors occurring when loading .fasl
1441 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1442 in a way which should help ASDF recover gracefully.
1443 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1444 stuff has been cleaned up. If you code contrib/ stuff, this might
1445 affect you, and you can look at contrib/README, contrib/STANDARDS,
1446 and/or the 0.8.4.27 diff to check.
1447 * In full calls the compiler now does not generate checks for declared
1448 argument types for all arguments.
1449 * various threading fixes
1450 ** and some experimental patches which didn't make it into
1451 the main tree for this release, but which are shipped in
1452 contrib/experimental-thread.patch as a possible fix for some
1453 failures (deadlock, spinning...) in GC-intensive multithreaded
1455 * fixed PPC build problem (source code incompatibility of different
1456 library versions): added offsetof() hackery which attempts to divine
1457 where glibc maintainers put uc_mcontext today
1458 * fixed bug 282: compiler does not trust type assertions while passing
1459 arguments to a full call.
1460 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1461 &OPTIONAL VALUES type parameter.
1462 * fix bug 214: algorithm for noting rejected templates is now more
1463 similar to that of template seletion. (also reported by rydis on
1465 * fixed bug 141b: printing backquoted information readably and prettily
1466 inserts a space where necessary.
1467 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1468 (CHARACTER 10) are now reported as errors, rather than propagated
1469 as unknown types. (reported by piso on #lisp)
1470 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1471 with values NIL and :ERROR. (thanks to Milan Zamazal)
1472 * fixed bug 191c: CLOS now does proper keyword argument checking as
1473 described in CLHS 7.6.5 and 7.6.5.1.
1474 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1475 longer raise an error; further, using a list as a for-as-arithmetic
1476 counter now raises a meaningful error.
1477 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1478 now understood by sequence creation functions such as MAKE-SEQUENCE
1480 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1481 a STRING-INPUT-STREAM.
1482 * compiler enhancement: SIGNUM is now better able to derive the type
1484 * type declarations inside WITH-SLOTS are checked. (reported by
1486 * fixed some bugs revealed by Paul Dietz' test suite:
1487 ** incorrect optimization of TRUNCATE for a positive first
1488 argument and negative second.
1489 ** compiler failure in let-convertion during flushing dead code.
1490 ** compiler failure while deriving type of TRUNCATE on an
1491 interval, containing 0.
1492 ** ASH of a negative bignum by a negative bignum count now returns
1494 ** intersection of CONS types now canonicalizes properly, fixing
1495 inconsistencies in SUBTYPEP.
1497 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1498 * incompatible change: The --disable-debugger command line
1499 option now clobbers the debugger at a more fundamental
1500 level, by redefining #'INVOKE-DEBUGGER instead of by
1501 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1502 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1503 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1504 enter the debugger REPL and then suffer recursive errors
1505 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1506 while under the new system, BREAK in --disable-debugger mode
1507 terminates the system just as an unhandled error would.
1508 * fixed compiler performance when processing loops with a step >1;
1509 * bug fix: DOCUMENTATION now retrieves generic function
1510 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1511 support has been systematized, and now supports the methods
1512 specified by ANSI, along with a default method and a method for
1513 slot documentation. (reported by Nathan Froyd)
1514 * bug fix: effective methods associated with a generic function are
1515 no longer cached over a change of that generic function's method
1516 combination. (reported by Andreas Fuchs)
1517 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1518 names a non-existent directory. (thanks to Andreas Fuchs)
1519 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1520 the PPC platform, lead to stack corruption; this has been fixed.
1521 (reported by Rainer Joswig)
1522 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1523 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1524 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1525 arithmetic in general, is now much more reliable on the PPC
1527 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1528 platform now returns the right answer.
1529 * optimization: restored some effective method precomputation in
1530 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1531 precomputation is now tunable.
1532 * optimization: compiler-internal data structure use has been
1533 reviewed, and changes have been made that should improve the
1534 performance of the compiler by about 20%.
1535 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1536 simple-base-strings and simple-bit-vectors is improved.
1537 * optimization: the optimization of 32-bit logical and arithmetic
1538 functions introduced in version 0.8.3 on the x86 has been
1539 implemented on the mips, ppc and sparc platforms; an
1540 implementation of the same facility, but for 64-bit arithmetic,
1541 has been added for the alpha.
1542 * microoptimization: the compiler is better able to make use of the
1543 x86 LEA instruction for multiplication by constants.
1544 * bug fix: in some situations compiler did not report usage of
1545 generic arithmetic in (SPEED 3) policy.
1546 * bug 145b fix: compiler used wrong type specifier while converting
1547 MEMBER-types to numeric.
1548 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1550 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1551 index argument now works properly on the Alpha platform.
1552 * bug fix: floating point exception treatment on the Alpha platform
1554 * bug fix: FILE-POSITION works much better on string input and
1555 output streams. (thanks to Nikodemus Siivola)
1556 * bug fix: many threading/garbage collection symptoms sorted.
1557 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1558 might be pseudo-atomic.
1559 * internal change: Stopping for GC is now done with signals not
1560 ptrace. GC is now done in whichever thread wanted it, instead of
1562 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1564 * bug fix: result form in DO is not contained in the implicit
1566 * incompatible change: ICR structure is changed; the value part of
1567 CONTINUATION is now called LVAR; corresponding functions are
1568 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1569 * added type deriver for ISQRT (thanks to Robert E. Brown).
1570 * bug fix: better support for loading from the command line when an
1571 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1573 * fixed some bugs revealed by Paul Dietz' test suite:
1574 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1575 ** ROUND and FROUND now give the right answer when given very
1576 small float arguments.
1577 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1579 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1580 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1581 ** LCM should return a non-negative integer.
1582 ** PARSE-INTEGER returned the index of a terminator instead of the
1583 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1584 ** PARSE-INTEGER returned an incorrect index being applied to a
1586 ** LCM with two arguments of 0 returns 0 rather than signalling
1588 ** unsigned addition of a 32-bit constant with the high bit set no
1589 longer causes an internal compiler error on the x86.
1590 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1591 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1592 with negative last argument.
1593 ** byte specifiers with zero size and position no longer cause
1594 an error during type derivation.
1595 ** bignum multiplication on the Alpha platform now returns the
1597 * porting: The system now builds on SuSE AMD64, although it still
1598 generates a 32-bit binary.
1599 * .fasl file incompatibility: The fasl file version number has
1600 been incremented (because of the changes to internal compiler
1601 data structures referred to above).
1603 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1604 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1605 more accurately, on the Darwin kernel running on PowerPC hardware.
1606 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1607 * Compiler code deletion notes now signal a condition of type
1608 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1609 an associated MUFFLE-WARNING restart.
1610 * The compiler now performs limited argument count validation of
1611 constant format strings in FORMAT, and where appropriate in ERROR,
1612 CERROR and WARN. (thanks to Gerd Moellmann)
1613 * New ASDF-INSTALL contrib can be used for automatic download and
1614 installation of third-party Lisp code from CCLAN or other sites
1616 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1617 which forces another thread to execute a function supplied by the
1619 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1620 now accept and act upon their :ELEMENT-TYPE keyword argument.
1621 (reported by Martin Atzmueller, Edi Weitz)
1622 * bug fix: FILE-POSITION now accepts position designators up to
1623 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1624 is the greater. (thanks to Patrik Nordebo)
1625 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1626 Kalle Olavi Niemitalo)
1627 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1628 printable. (reported by Eric Marsden)
1629 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1630 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1631 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1632 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1633 (thanks to Stig Sandoe)
1634 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1635 UNREAD-CHAR now correctly includes the unread character in the
1636 target sequence. (thanks to Gerd Moellmann)
1637 * bug fix in threaded builds: the system can now be suspended and
1638 resumed by shell job control with minimal disruption.
1639 * bug fixes in times and timezones >2038AD
1640 * better handling of "where is GNU make?" problem in build scripts
1641 (thanks to Nikodemus Siivola)
1642 * new optimization: inside a named function any reference to a
1643 function with the same name is considered to be a self-reference;
1644 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1645 optimization quality.
1646 * new optimization on x86: logical functions and + now have
1647 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1648 used when the result is truncated to 32 bits.
1649 * VALUES declaration is partially enabled.
1650 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1651 * fixed some bugs revealed by Paul Dietz' test suite:
1652 ** The system now obeys the constraint imposed by
1653 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1654 types form a lattice under type intersection.
1655 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1656 ** ASSOC now ignores NIL elements in an alist.
1657 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1658 and (1+ MOST-POSITIVE-FIXNUM) answers.
1659 ** The addition of a method with invalid qualifiers to a generic
1660 function does not cause an error to be signalled immediately;
1661 a warning is signalled, and the error is generated only on
1662 calling the generic function.
1663 * changed .fasl file version number, in part to add type codes for
1664 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1665 obscure ANSI requirements
1667 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1668 * fixed bug 148: failure to inline-expand a local function left
1669 garbage, confusing the compiler.
1670 * fixed bugs 3cd: structure slot readers perform type check if the
1671 slot can have an invalid value (i.e. it is either not initialized
1672 or can be written with a less specific slot writer).
1673 * bug fix: the compiler now traps array references to elements off
1674 the end of an array; previously, the bounds checking in some
1675 circumstances could go off-by-one.
1676 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1678 * type declarations for array element types now obey the description
1679 on the CLHS page "Declaration TYPE", as per discussions on
1680 sbcl-help around 2003-05-08. This means that a declaration
1681 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1682 declaration, all references to BAR will be asserted or assumed
1683 (with THE, so dependent on compiler policy) to involve objects of
1684 type FOO. Note that no such declaration is implied in
1685 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1686 * declared types of functions from the "Conditions"
1687 chapter. (reported by Paul Dietz)
1688 * bug fix: CERROR accepts a function as its first argument.
1689 * bug fix: NTH an NTHCDR accept a bignum as index
1690 arguments. (reported by Adam Warner)
1691 * optimization: character compare routines now optimize comparing
1692 against a constant character. (reported by Gilbert Baumann)
1693 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1694 argument now works properly on the MIPS platform.
1695 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1696 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1697 declared to have a functional type, does not cause an error
1699 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1700 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1701 * fixed bug: if last continuation of a deleted block has a
1702 destination, this destination should be deleted too. (reported by
1704 * fixed a bug in the bootstrap process: the host compiler's values
1705 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1706 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1707 test case from Patrik Nordebo)
1708 * improved the ability of the disassembler on the PPC platform to
1709 provide helpful disassembly notes.
1710 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1711 the class in more cases than previously.
1712 * bug fix: FILE-POSITION now understands :START and :END for
1713 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1714 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1715 8) data. (thanks to David Lichteblau)
1716 * bug fix: it is possible to add a method to a generic function
1717 without lambda list.
1718 * bug fix: reader failed to signal END-OF-FILE inside an
1719 object representation. (reported by Nikodemus Siivola)
1720 * fixed some bugs revealed by Paul Dietz' test suite:
1721 ** LAST and [N]BUTLAST should accept a bignum.
1722 ** condition slot accessors are methods.
1723 ** (VECTOR NIL) is a subtype of STRING.
1725 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1726 * minor incompatible change: some nonsensical specialized lambda
1727 lists (used in DEFMETHOD) which were previously ignored now signal
1729 * minor incompatible change: the system is now aware of the types of
1730 variables in the COMMON-LISP package, and will signal errors for
1731 most violations of these type constraints (where previously they
1732 were silently accepted).
1733 * minor incompatible change: COMPILE-FILE now uses the freedom
1734 afforded (ANSI 3.2.2.3) to use derived function types for
1735 functions defined in the same file. This also permits the system
1736 to warn on static type mismatches and function
1737 redefinition. (Currently it does not work with high DEBUG level.)
1738 * minor incompatible change: VALUES declaration is disabled.
1739 * When issuing notes, the compiler now signals a condition of type
1740 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1741 restart for use in user handlers. It is expected that the
1742 COMPILER-NOTE condition will eventually become a condition
1743 supertype to a hierarchy of note types, which will then be
1744 handleable in a similar fashion. However, at the moment, no such
1745 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1746 but it's an implementation detail, not a classification for the
1748 * Changes in type checking closed the following bugs:
1749 ** type checking of unused values (192b, 194d, 203);
1750 ** template selection based on unsafe type assertions (192c, 236);
1751 ** type checking in branches (194bc).
1752 * A short form of VALUES type specifier has ANSI meaning (it has
1753 increased the number of situations when SBCL cannot perform type
1755 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1756 keywords or constants is permissible.
1757 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1758 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1759 classes. (thanks to Antonio Martinez)
1760 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1761 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1762 argument to a pretty-print stream. Instead, it leaves any such
1763 operations to DESCRIBE-OBJECT methods.
1764 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1766 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1767 lambda lists. (thanks to David Lichteblau)
1768 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1769 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1770 (thanks to Gerd Moellmann)
1771 * fixed bug 246: increased compilation speed of long
1772 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1774 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1776 * DEFCONSTANT now throws a condition of type
1777 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1778 constant to a non-EQL value; CONTINUE and ABORT restarts
1779 respectively change and preserve the value.
1780 * fixed bug 63: The code walker, part of the implementation of CLOS,
1781 is now better at handling symbol macros.
1782 * bug fix: There is no longer an internal implementation type named
1783 CL:LENGTH. (reported by Raymond Toy)
1784 * bug fix: In macro-like defining macros/special operators the
1785 implicit block does not enclose the lambda list.
1786 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1787 enclosed in parentheses) are not suitable as type specifiers, and
1788 their use properly signals an error now.
1789 * bug fix: An argument count mismatch for a type specifier in code
1790 being compiled no longer causes an unhandled error at compile
1791 time, but signals a compile-time warning.
1792 * fixed simple vector readable printing
1793 * bug fix: DESCRIBE takes more care over whether the class
1794 precedence list slot of a class is bound before accessing it.
1795 (reported by Markus Krummenacker)
1796 * bug fix: FORMATTER can successfully compile pretty-printer format
1797 strings which use variants of the ~* directive inside.
1798 * bug fix: SEARCH now applies its TEST predicate to the elements of
1799 the arguments in the correct order. (thanks to Wolfhard Buss)
1800 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1801 in type inference. (thanks to Robert E. Brown)
1802 * bug fix: Reading in symbols with an explicit package name of ""
1803 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1804 * fixed some bugs revealed by Paul Dietz' test suite:
1805 ** NIL is now allowed as a structure slot name.
1806 ** Arbitrary numbers, not just REALs, are allowed in certain
1807 circumstances in LOOP for-as-arithmetic clauses.
1808 ** Multiple class redefinitions before slot access no longer
1809 causes a type error.
1810 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1811 association between the name and a class.
1812 ** Generic functions with non-standard method-combination and over
1813 five methods all of which return constants no longer return NIL
1814 after the first few invocations. (thanks to Gerd Moellmann)
1815 ** CALL-NEXT-METHOD with no arguments now passes the original
1816 values of the arguments, even in the presence of assignment.
1817 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1819 ** DISASSEMBLE works with closures and funcallable instances.
1820 ** ADD-METHOD now returns the generic function, not the new method.
1821 ** FIND-METHOD signals an error if the lengths of the specializers
1822 is incompatible with the generic function, even if the ERRORP
1824 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1825 which its argument is a member.
1826 ** DEFCLASS only redefines the class named by its class-name
1827 argument if that name is the proper name of the class;
1828 otherwise, it creates a new class.
1829 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1830 of the UNBOUND-SLOT condition to the name of the slot.
1831 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1832 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1833 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1834 SLOT-MAKUNBOUND in the specified fashion.
1836 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1837 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1838 cross-compilation host. As a consequence, we can now bootstrap our
1839 way up to SBCL starting with a bare gcc toolchain and human-readable
1840 source code (first the source to CLISP, then the source to SBCL).
1841 * A contributed module containing a partial implementation of the
1842 simple-streams interface has been included. (thanks to Rudi
1844 * A contributed module implementing the RFC1321 Message Digest
1845 Algorithm, known as MD5, has been included.
1846 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1847 no longer has any effect, as the code controlled by this feature
1848 has been deleted. (As far as we know, no-one has ever built using
1849 this feature, and its semantics were confused in any case).
1850 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1851 work on conditions (as required by the ANSI specification),
1852 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1853 expected behaviour on conditions. Users should note, however,
1854 that such behaviour is not required by the ANSI specification,
1855 and so use of this behaviour may render their code unportable.
1856 * fixed some bugs revealed by Paul Dietz' test suite:
1857 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1859 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1860 ** SXHASH on condition objects no longer returns NIL.
1861 ** :ALLOCATION :CLASS slots are better treated; their values are
1862 updated on class redefinition, and initforms inherited from
1863 superclasses are applied.
1864 ** REMOVE-METHOD returns its generic function argument even when
1865 no method was removed.
1866 ** SHARED-INITIALIZE now initializes the values of the requested
1867 slots, including those with :ALLOCATION :CLASS.
1868 ** ALLOCATE-INSTANCE now works on structure classes defined via
1869 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1871 ** SLOT-EXISTS-P now works on conditions, as well as structures
1873 ** MAKE-LOAD-FORM now has the required methods on
1874 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1875 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1876 keyword, and now implements the SLOT-NAMES argument.
1877 ** methods with &OPTIONAL arguments no longer allow too many
1878 arguments to be passed in the call without error.
1879 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1880 option is consistent with the required arguments of the generic
1881 function lambda list.
1882 * bug fix: REQUIRE accepts a string designator. (Thanks to
1884 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1885 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1886 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1887 * bug fix: The compiler now checks for duplicated variables in macro
1889 * bug fix: SETQ on globals returns the correct value.
1890 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1891 not a COMPILER-ERROR (followed by some other strange error on
1892 choosing the CONTINUE restart).
1893 * bug fix: make.sh and friends are now more consistent in the way that
1894 they look for GNU "make".
1896 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1897 * experimental native threads support (on x86 Linux >=2.4 only).
1898 This is not compiled in by default: you need to add :SB-THREAD to
1899 the target features. See the "Beyond ANSI" chapter of the manual
1901 * fix for longstanding nonANSIism: The old distinction between
1902 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1903 The return value from CL:FIND-CLASS is now a CLOS class, and
1904 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1905 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1906 * An interface to the MetaObject Protocol, as described in Kiczales,
1907 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1908 MIT Press, 1991, is available from the SB-MOP package.
1909 * incompatible change: the SB-PCL package should now be considered
1910 a private implementation detail, and no longer a semi-private MOP
1912 * minor incompatible change: due to rearrangement for threads, the
1913 control stack and binding stack are now allocated at arbitrary
1914 addresses instead of being hardcoded per-port. Users affected by
1915 this probably have to be doing advanced things with shared
1916 libraries, and will know who they are.
1917 * minor incompatible change: Previously, all --eval forms used were
1918 processed with READ before any of them were processed with EVAL.
1919 Now each --eval form is processed with both READ and EVAL before
1920 the next --eval form is processed. (Thus package operations like
1921 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1922 work as the user might reasonably expect.)
1923 * minor incompatible change: *STANDARD-INPUT* is now only an
1924 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1926 * minor incompatible change: Y-OR-N-P is now character-oriented,
1927 not line oriented. Also, YES-OR-NO-P now works without errors.
1928 (thanks to Antonio Martinez)
1929 * sb-aclrepl module improvements: an integrated inspector, added
1930 repl features, and a bug fix to :trace command.
1931 * Known functions, which cannot be open coded by the backend, are
1932 considered to be able to check types of their arguments. (fixing
1933 a bug report by Nathan J. Froyd)
1934 * fixed a bug in computing method discriminating functions: It is
1935 now possible to define methods specialized on classes which have
1936 forward-referenced superclasses. (thanks to Gerd Moellmann)
1937 * fixed evaluation order in optional entries (reported by Gilbert
1939 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1940 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1941 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1942 required-by-AMOP NAME argument, as well as CLASS and
1943 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1944 * fixed bug 20: DEFMETHOD can define methods using names that are
1945 not the proper names of classes to designate class specializers.
1946 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1947 Unix file descriptors, instead of blowing up. (thanks to Antonio
1949 * Garbage collection refactoring: user-visible change is that a
1950 call to the GC function during WITHOUT-GCING will not do garbage
1951 collection until the end of the WITHOUT-GCING. If you were doing
1952 this you were probably losing anyway.
1953 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1954 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1955 TYPEP the latter but not the former.
1956 * The compiler issues a full WARNING for calls to undefined functions
1957 with names from the CL package.
1958 * MAP-INTO for a vector destination is open coded. (reported by
1959 Brian Downing on c.l.l)
1960 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1961 documentation string.
1962 * fixed some bugs revealed by Paul Dietz' test suite:
1963 ** COPY-ALIST now signals an error if its argument is a dotted
1965 ** Condition slots are now accessed more correctly in the presence
1966 of multiple initargs for a given slot.
1967 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1968 exclude restarts of the same name associated with a different
1970 ** DEFCLASS of forward-referenced classes with another
1971 forward-referenced class in the superclasses list no longer
1973 ** Condition slots are now initialized once each, not multiple
1974 times. (thanks to Gerd Moellmann)
1975 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1977 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1978 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1979 arguments contain duplicated elements.
1980 ** RESTART-CASE understands local macros.
1981 ** RESTART-CASE associates exactly its own restarts with a condition.
1982 ** ENDP in safe mode checks its argument to be of type LIST.
1983 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1984 in question is unbound.
1985 ** Optimized MAKE-INSTANCE functions no longer cause internal
1986 assertion failures in the presence of duplicate initargs.
1987 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1988 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1989 propagated even in the ctor optimized implementation of
1991 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1993 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1994 * a better implementation of SXHASH on (simple) bit vectors,
1995 measured both in execution speed and in distribution of results
1996 over the positive fixnums, has been installed. Likewise, a better
1997 implementation of EQUAL for simple bit vectors is now available.
1998 * fixed CEILING optimization for a divisor of form 2^k.
1999 * fixed bug 240 (emitting extra style warnings "using the lexical
2000 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2001 by Antonio Martinez)
2002 * fixed SXHASH, giving different results for NIL depending on type
2003 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2004 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2005 variable. (found by Rolf Wester)
2006 * SBCL does not ignore type declarations for special
2007 variables. (reported by rif on c.l.l 2003-03-05)
2008 * some bug fixes in contrib/sb-aclrepl/
2009 * fixed some bugs revealed by Paul Dietz' test suite:
2010 ** a bug in the CONS type specifier, whereby the CAR and CDR
2011 types got intertwined, has been fixed;
2012 ** the type system is now able to reason about the interaction
2013 between INTEGER and RATIO types more completely;
2014 ** APPEND, [N]REVERSE and NRECONC check that those their
2015 arguments, which must be proper lists, are really so;
2016 ** An array specialized to be unable to hold elements has been
2017 implemented, as required -- yes, really -- by ANSI;
2018 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2019 on malformed property lists;
2021 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2022 * incompatible packaging change: in line with Unix convention,
2023 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2024 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2025 * REQUIRE and PROVIDE are now optionally capable of doing something
2026 useful. See the documentation string for REQUIRE.
2027 * infrastructure for a managed SBCL contrib system: contributed
2028 modules in this release include:
2029 ** the ASDF system definition facility;
2030 ** an interface to the BSD Sockets API;
2031 ** an ACL-like convenience interface to the repl;
2032 (thanks to Kevin Rosenberg)
2033 ** an implementation of ROTATE-BYTE, with efficient implementation
2035 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2036 gives a closer approximation to the right answer than previously.
2037 (thanks to Raymond Toy)
2038 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2039 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2040 optional environment arguments, as required by ANSI.
2041 * fixed bugs in other functions taking environment objects, allowing
2042 calls with an explicit NIL environment argument to be compiled
2044 * fixed bug 228: primary return values from
2045 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2046 COMPILE or FUNCTION.
2047 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2048 :TYPE will no longer signal errors on innocuous objects.
2049 * fixed bug 231b: SETQ is better at respecting type declarations in
2050 the lexical environment.
2051 * fixed a bug in DEFCLASS: classes named by symbols with no or
2052 unprintable packages can now be defined.
2053 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2054 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2055 * fixed compiler failure related to checking types of functions.
2056 (reported by Robert E. Brown)
2057 * the compiler is now much more consistent in its error-checking
2058 treatment of bounding index arguments to sequence functions: in
2059 (SAFETY 3) code, errors will be signalled in almost all cases if
2060 invalid sequence bounding indices are passed to functions defined
2061 by ANSI to operate on sequences.
2062 * fixed a bug in the build procedure: documentation of SBCL-specific
2063 packages is now preserved and available in the final Lisp image.
2064 * lifted FDEFINITION lookup out of loops in the implementation of
2065 many list operations. (thanks to Robert E. Brown)
2066 * fixed a bug in the reader: the #n# reader macro now works for
2067 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2068 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2069 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2070 E. Brown and Tony Martinez respectively)
2071 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2072 SLOT-BOUNDP: in particular, it is now able to optimize them much
2073 better, and is now not vulnerable to having packages renamed.
2074 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2075 always called when a slot is not present in an instance. (thanks
2077 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2078 longer signal an error on unreading a character following EOF on
2079 the previous constituent stream. (thanks to Tony Martinez)
2080 * fixed some bugs revealed by Paul Dietz' test suite:
2081 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2082 not just nonnegative fixnums;
2083 ** the logical bit-array operators such as BIT-AND now accept an
2084 explicit NIL for their "opt-arg" argument (to indicate a
2085 freshly-consed result bit-array);
2086 ** ELT now signals an error on an invalid sequence index in safe
2088 ** the type system is now cleverer about negations of numeric
2089 types, and consequently understands the BIGNUM and RATIO types
2091 ** the type system is now cleverer about the interaction between
2092 INTEGER and RATIO types: while bugs still remain, many more
2093 cases are accurately computed;
2094 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2095 if it is in the last clause;
2096 ** CONSTANTLY now correctly returns a side-effect-free function in
2098 ** DECLARE is no longer treated as a special-operator; in
2099 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2100 * incremented fasl file version number due to the change in the
2101 DEFSTRUCT-SLOT-DESCRIPTION structure.
2103 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2104 * minor incompatible change: code processed by the "interpreter" or
2105 EVAL now has a compilation optimization policy of (DEBUG 2)
2106 (changed from (DEBUG 1)) to improve debuggability of interactive
2107 development, and to allow the use of the debug RETURN command in
2109 * an experimental implementation of the RETURN command for the
2110 debugger has been included. (thanks to Frederik Kuivinen)
2111 * fixed bug 62: constraints were not propagated into a loop.
2112 * fixed bug in embedded calls of SORT (reported and investigated by
2114 * fixed some bugs revealed by Paul Dietz' test suite:
2115 ** printing and reading of arrays with some dimensions having
2116 length 0 (thanks to Gerd Moellmann);
2117 ** BOA constructor with &AUX argument without a default value does
2118 not cause a type error;
2119 ** CONSTANTP now returns true for all self-evaluating objects.
2121 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2122 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2123 accessors that are related by inheritance, as specified in the
2124 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2125 Valtteri Vuorikoski)
2126 * The compiler is now able to inline functions that were defined in
2127 a complex lexical environment (e.g. inside a MACROLET).
2128 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2130 * Support for the upcoming FreeBSD-5.0 release has been included.
2131 (thanks to Dag-Erling Smorgrav)
2132 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2133 effect when it is not in a toplevel context.
2134 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2135 stand a better chance of being correct. (thanks to Gerd
2137 * fixed bug in COERCE, which now signals an error on coercing a
2138 rational to a bounded real type which excludes the expected
2140 * The compiler is now able to derive types more accurately from the
2141 COERCE and COMPILE functions.
2142 * fixed bug 223: functional binding is considered to be constant
2143 only for symbols in the CL package.
2144 * fixed bug 231: SETQ did not check the type of a variable being set
2145 (reported by Robert E. Brown)
2146 * A new optimization for MAKE-INSTANCE has been included, fixing
2147 various bugs (including relating to :ALLOCATION :CLASS slots and
2148 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2150 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2151 ** As required by ANSI, LOOP now disallows anonymous collection
2152 clauses such as COLLECT I in conjunction with aggregate boolean
2153 clauses such as THEREIS (= I 1);
2154 ** LOOP now signals an error when any variable is reused in the
2155 same loop (including the potentially useful construct analogous
2156 to WITH A = 1 WITH A = (1+ A);
2157 ** IT is only a special loop symbol within the first clause of a
2158 conditional loop clause;
2159 ** LOOP with a typed iteration variable over a hashtable now
2160 signals a type error iff it should.
2161 * fixed some other bugs revealed by Paul Dietz' test suite:
2162 ** FILE-STREAM now names the class previously known as FD-STREAM;
2163 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2164 argument) no longer signals an error;
2165 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2166 of the slot symbol, rather than using the current package
2167 ((:CONC-NAME "") continues to intern the slot's name in the
2169 * incremented fasl file version number, because of the incompatible
2170 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2171 of the new implementation of DEFINE-COMPILER-MACRO.
2173 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2174 * Support for building SBCL for MIPS platforms running in
2175 little-endian mode has now been checked in, and basic
2176 functionality on said platforms verified.
2177 * minor incompatible change: PCL now records the pathname of a file
2178 in which methods and the like are defined, rather than its
2180 * minor incompatible change: TRUENAME now considers the truename of
2181 a file naming a directory to be the pathname with :DIRECTORY
2182 component indicating that directory.
2183 * minor incompatible change: a NAMED clause in the extended form of
2184 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2185 reason for the previous behaviour is unclear.
2186 * more systematization and improvement of CLOS and MOP conformance
2187 in PCL (thanks to Gerd Moellman and Pierre Mai):
2188 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2190 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2191 primary methods with no specializers;
2192 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2194 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2195 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2196 FORWARD-REFERENCED-CLASSes; error reporting on
2197 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2199 ** SXHASH on CLOS instances now uses a slot internal to the
2200 instance to return different numbers on distinct instances,
2201 while preserving the same return value through invocations of
2203 ** DEFMETHOD signals errors when methods with longer incongruent
2204 lambda lists are added to generic functions;
2205 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2206 CLASS, as specified in AMOP;
2207 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2208 on the order returned by the primary method for classes of
2209 class STANDARD-CLASS;
2210 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2211 * fixed some bugs shown by Paul Dietz' test suite:
2212 ** DOLIST puts its body in TAGBODY;
2213 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2215 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2216 value producing form;
2217 ** if more variables are given to PROGV than values, extra
2218 variables are bound and made to have no value;
2219 ** NSUBSTITUTE on list arguments gets the right answer with
2221 ** ELT signals an error of type TYPE-ERROR when the index argument
2222 is not a valid sequence index;
2223 ** LOOP signals (at macroexpansion time) an error of type
2224 PROGRAM-ERROR when duplicate variable names are found;
2225 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2226 ** FUNCALL of special-operators now cause an error of type
2228 ** PSETQ now works as required in the presence of side-effecting
2229 symbol-macro places;
2230 ** NCONC accepts any object as its last argument;
2231 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2233 ** loop-for-as-package does not require a package to be explicitely
2235 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2237 * fixed bug 166: compiler preserves "there is a way to go"
2238 invariant when deleting code.
2239 * fixed bug 172: macro lambda lists with required arguments after
2240 &REST arguments now cause an error to be signalled. (thanks to
2242 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2243 bug. (thanks to Matthew Danish)
2244 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2246 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2247 arguments are lexically rebound. (thanks to Gerd Moellmann and
2249 * fixed bug 194: error messages are now more informative when there
2250 is no primary method applicable in a call to a generic
2251 function. (thanks to Gerd Moellmann)
2252 * fixed bug in command line argument checking (thanks to Julian
2254 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2255 (thanks to Matthew Danish)
2256 * incremented fasl file version number, because of the
2257 SXHASH-related changes in the layout of CLOS data structures
2259 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2260 * minor incompatible change: The runtime (the Unix executable named
2261 "sbcl") is now much pickier about the .core files it will load.
2262 Essentially it now requires .core files to descend from the same
2263 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2264 as the runtime does. (The intent is to prevent the crashes which
2265 can occur, and which can even be reported as mysterious failures,
2266 when people patch the sources or change the build parameters
2267 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2268 sbcl and .core files.)
2269 * fixed bug: VALUES-LIST is no longer optimized away.
2270 * fixed bug 142: The FFI conversion of C string values to Lisp
2271 string values no longer conses excessively. (thanks to Nathan
2272 Froyd porting Raymond Toy's fix to CMU CL)
2273 * began to systematize and improve MOP conformance in PCL (thanks to
2274 Nathan Froyd, Gerd Moellman and Pierre Mai):
2275 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2277 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2278 ** FINALIZE-INHERITANCE is now called on class finalization;
2279 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2280 argument precedence order.
2281 * fixed bug 202: The compiler no longer fails on functions whose
2282 derived types contradict their declared type.
2283 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2284 so it can be non-toplevel.
2285 * The fasl file version number has changed (because of the new
2286 implementation of DEFMACRO).
2287 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2288 safe code, that any length requirement by their type-specifier
2289 argument is valid. The exceptions to this are described in bug
2291 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2292 functions now check the types of their inputs as required by ANSI.
2293 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2294 introduced symbol is DECLAREd to be SPECIAL.
2295 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2296 * fixed bug: :COUNT argument to sequence functions may be negative
2297 * fixed bug: body of DO-SYMBOLS may contain declarations
2298 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2299 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2300 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2301 symbol macro only once
2302 * fixed printing of call frame when argument list is unavailable
2303 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2304 * compiler no longer signals WARNING on unknown keyword
2307 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2308 * A beta-quality port to the mips architecture running Linux,
2309 based on the old CMUCL backend, has been made. It has been tested
2310 on a big-endian kernel, and works sufficiently well to be able to
2311 rebuild itself; it has not been tested in little-endian mode.
2312 * fixed an inconsistency between gencgc.c and purify.c which made
2313 dumping/loading .core files unreliable
2314 * fixed bug 120a: The compiler now deals correctly with IFs where
2315 the consequent is the same as the alternative, instead of
2316 misderiving the return type. (thanks to Alexey Dejneka)
2317 * fixed bug 113: Logical pathnames are now dumpable (the logical
2318 host is resolved at load-time, throwing an error if it is not
2320 * fixed bug 174: FORMAT's error message is slightly clearer when a
2321 non-printing character is used in a format directive.
2322 * fixed several bugs in compiler checking of type declarations, i.e.
2323 violations of the Python "declarations are assertions" principle
2324 (thanks to Alexey Dejneka)
2325 * fixed several bugs in PCL's error checking (thanks to Gerd
2327 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2329 * fixed bug in compilation of functions as first class values
2330 (thanks to Antonio Martinez-Shotton)
2331 * The compiler's handling TYPE-ERRORs which it can prove will
2332 inevitably happen at runtime has been cleaned up and corrected
2333 in several ways. (thanks to Alexey Dejneka)
2334 * improved argument type checking for various basic arithmetic
2335 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2336 lavished on them in the past that they can be compiled in many
2337 ways in different special cases
2338 * fixed bug 181: compiler checks validity of user supplied type
2340 * cleaned up code flushing in optimization: Function calls which
2341 should signal errors for safety purposes (e.g. which ANSI says
2342 should signal errors when their arguments are of incorrect type)
2343 are no longer optimized away.
2344 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2345 * incremented fasl file version number, because changes in the
2346 implementation of sequence functions like COERCE caused
2347 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2348 in old inline expansions) to become undefined. (Actually these
2349 changes were later undone, so we might very well be binary
2350 compatible with 0.7.7 after all, but leaving the version number
2351 incremented seemed like the simplest and most conservative
2354 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2355 * An alpha-quality port to the parisc architecture running Linux,
2356 based on the old CMUCL backend, has been made. This, even more so
2357 than the other backends, should be considered still a work in
2358 progress; known problems include that the Linux kernel in 64-bit
2359 mode does not propagate the correct sigcontext structure to
2360 userspace, and consequently SBCL on a parisc64 kernel will not
2362 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2363 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2364 suppressed.) Also now that the compiler is looking at declarations
2365 in the environment, it checks optimization declarations as well,
2366 and suppresses inlining when (> DEBUG SPEED).
2367 * More fixes have been made to treatment of floating point exception
2368 treatment and other Unix signals. In particular, floating point
2369 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2370 * The detection and handling of control stack exhaustion (infinite
2371 or very deeply nested recursion) has changed. Stack exhaustion
2372 detection is now done by write-protecting pages at the OS level
2373 and applies at all optimization settings; when found, a
2374 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2375 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2376 be caught using IGNORE-ERRORS.
2377 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2378 correctly when transforming recursive LABELS functions to LETs.
2379 (thanks to Alexey Dejneka)
2380 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2381 that are names of constants or global variables.
2382 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2383 alien routines with docstrings.
2384 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2385 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2387 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2388 (thanks to Eric Marsden)
2389 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2390 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2391 * Bug fix: The SPARC backend can now compile functions involving
2392 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2393 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2394 to a functions expecting a PATHNAME-DESIGNATOR.
2395 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2396 lambda lists. (thanks to Alexey Dejneka)
2397 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2398 functions so that EXPORTing the name of the class doesn't cause
2399 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2400 to Antonio Martinez for reporting this)
2401 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2402 constant-folded without failing an assertion. (thanks to Einar
2403 Floystad Dorum for reporting this)
2404 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2405 types are now dealt with more correctly by the compiler.
2406 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2407 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2408 behaviour analogous to RENAME-FILE. This puts its behaviour more
2409 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2411 * The fasl file version number has changed again. (because of the
2412 bug fix involving the names of PCL MAKE-INSTANCE functions)
2414 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2415 * bug fix: Floating point exceptions are treated much more
2416 consistently on the x86/Linux and PPC/Linux platforms.
2417 * Array initialization with :INITIAL-ELEMENT is now much faster for
2418 cases when the compiler cannot open code the array creation, but
2419 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2420 array accesses have also seen a speed increase.
2421 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2422 specified. (This was at the root of some bad interactions between
2423 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2425 * bug fix: Internal error arguments for undefined functions are now
2426 computed correctly on the PPC/Linux platform.
2427 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2428 Raymond Toy's patch for CMU CL)
2429 * Support for the Solaris 9 operating environment has been included
2430 (thanks to Daniel Merritt)
2431 * A very ugly but hopefully complete draft of the missing FFI chapter
2432 of the manual has been created by reformatting the corresponding
2433 CMU CL manual chapter into (currently very ugly and incoherent)
2434 DocBook and bringing it up to date for SBCL behavior. Thus, the
2435 manual is now essentially complete, at least by my extreme
2436 once-and-only-once standards, whereby it's acceptable to refer to
2437 the doc strings of SB-EXT functions as the primary documentation.
2438 * The fasl file version number has changed again, due to cleanup of
2439 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2440 is no longer a static symbol.)
2442 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2443 * SBCL now builds with OpenMCL (version 0.12) as the
2444 cross-compilation host; also, more progress has been made toward
2445 bootstrapping under CLISP.
2446 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2448 * bug 158 fixed: The compiler can now deal with integer loop
2449 increments different from 1; fixing this turned out also to fix
2451 * bug 169 fixed: no more bogus warnings about using lexical bindings
2452 despite the presence of perfectly good SPECIAL declarations (thanks
2453 to David Lichteblau)
2454 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2455 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2456 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2458 * bug 180 fixed: Method combination specifications no longer ignore
2459 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2460 * bug fix: Structure type predicate functions now check their argument
2461 count as they should.
2462 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2463 correctly. (thanks to Pierre Mai)
2464 * minor incompatible change: The --noprogrammer option is deprecated
2465 in favor of the new --disable-debugger option, which is very similar.
2466 (The major difference is that it takes effect at a slightly different
2467 time at startup, causing handling of errors in --sysinit and
2468 --userinit files will be affected differently.) The
2469 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2470 been added to allow this functionality to be controlled from ordinary
2471 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2472 maintainers, who might want to run non-interactive scripts to
2473 build SBCL cores which will later be used interactively.)
2474 * minor incompatible change: The LOAD function no longer, when given
2475 a wild pathname to load, loads all files matching that pathname.
2476 Instead, an error of type FILE-ERROR is signalled.
2478 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2479 * bug 147 fixed: The compiler preserves its block link/count
2480 invariants more correctly now so that it doesn't crash. (thanks
2482 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2484 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2485 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2486 no-value-for-this-character as NIL (as specified by ANSI).
2487 * HOST-NAMESTRING on physical pathnames now returns a string that is
2488 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2489 (thanks to Christophe Rhodes)
2490 * The Alpha port handles icache flushing more correctly. (thanks to
2492 * More progress has been made toward bootstrapping under CLISP. (thanks
2493 to Christophe Rhodes)
2494 * The fasl file format has changed again, because dynamic loading
2495 on OpenBSD (which has non-ELF object files) motivated some cleanups
2496 in the way that foreign symbols are transformed and passed around.
2497 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2498 is no longer treated as whitespace by the reader, but instead as
2499 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2500 |A
\7fB|, instead of A as it used to.
2502 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2503 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2504 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2505 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2506 this as of 0.7.1.45, but was left out of the previous news section
2507 (thanks to Dan Barlow)
2508 * SBCL now runs on the Solaris operating system on SPARC architectures
2509 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2510 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2511 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2512 * SPARC backend cleanups, allowing builds of cores optimized for V8
2513 and V9 SPARCS, and also emission of code targeted to a particular
2514 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2516 * SBCL is closer to bootstrapping under CLISP, thanks to various
2517 fixes by Christophe Rhodes.
2518 * The fasl file format has changed again, to allow the compiler's
2519 INFO database to support symbol macros.
2520 * The user manual (in doc/) is formatted into HTML more nicely.
2521 (thanks to coreythomas)
2522 * The system is smarter about SUBTYPEP relationships, especially
2523 those involving NOT types (including types such as ATOM which are
2524 represented internally using NOT types). Thus SUBTYPEP is less
2525 likely to return (VALUES NIL NIL) in general, and in particular
2526 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2528 * The fasl file format has changed again, because the internal
2529 representation of types now includes a new slot to support the new
2530 SUBTYPEP-of-NOT-types logic.
2531 * (not a change in the main branch of SBCL, but a related prototype
2532 which can hopefully be merged into the main branch of SBCL in the
2533 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2534 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2535 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2536 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2537 dispatch macro character combinations. (thanks to Alexey Dejneka)
2538 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2539 that superficially look like logical namestrings correctly.
2540 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2542 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2543 default supplied-p) for &optional and &key arguments. (thanks to
2546 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2547 * incompatible change: The compiler is now less aggressive about
2548 tail call optimization, doing it only when (> SPACE DEBUG) or
2549 (> SPEED DEBUG). (This is an incompatible change because there are
2550 programs which relied on the old CMU-CL-style behavior to optimize
2551 away their unbounded recursion which will now die of stack overflow.)
2552 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2553 for non-GENCGC systems has been increased to 20M (since that
2554 seems much closer to the likely performance optimum for modern
2555 systems than the old 4M value was)
2556 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2557 and *DEBUG-PRINT-LEVEL*
2558 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2559 of CMU CL's support for SPARC, and various endianness and other
2560 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2561 * new syntactic sugar for the Unix command line: --load foo.bar is now
2562 an alternate notation for --eval '(load "foo.bar")'.
2564 ** The system now detects stack overflow and handles it gracefully,
2565 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2566 optimization settings. (This is a good thing in general, and
2567 its introduction in this version should be particularly timely
2568 for anyone whose code fails because of suppression of tail
2570 ** The system now hunts for the C variable "environ" in a more
2571 devious way, to avoid segfaults when the C library version differs
2572 between compile time and run time. (thanks to Christophe Rhodes)
2573 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2574 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2575 ** The compiler no longer issues bogus style warnings for undefined
2576 classes in the same source file as the DEFCLASSes which defined
2577 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2579 ** fixes in CONDITION class precedence list for undefined function
2580 errors (thanks to Alexei Dejneka)
2581 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2582 correctly. (thanks to Dan Barlow)
2583 ** portability fixes aiming at bootstrapping under CLISP (thanks
2584 to Dave McDonald and Christophe Rhodes)
2585 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2586 ** fixes in type translation and and type inference (thanks to
2588 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2589 ** various fixes in the new ports (thanks to Dan Barlow)
2590 * several changes related to debugging:
2591 ** suppression of tail recursion, as noted above
2592 ** stack overflow detection, as noted above
2593 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2594 is now the default. (For some time encapsulation has been more
2595 reliable than the breakpoint-based :ENCAPSULATE NIL
2596 implementation, at least on X86 systems; and I just noticed that
2597 encapsulation also seems closer to the spirit of the ANSI
2600 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2602 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2603 up properly again. (There was a packaging bug in 0.7.0 which
2604 left their definitions in SB-SYS::LOAD-FOREIGN and
2605 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2606 vital for most things which interface to C-level interfaces,
2607 like extensions working with sockets or databases or
2608 Perl-compatible regexes or whatever, and the need to fix
2609 this bug is the main reason that 0.7.1 was released so
2611 ** DEFGENERIC is now choosier about the methods it redefines, so that
2612 reLOADing a previously-LOADed file containing DEFGENERICs does
2613 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2614 cycle now works as it should. (thanks to Alexey Dejneka)
2615 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2616 fixed by Christophe Rhodes. (He actually submitted this patch
2617 months ago, and I delayed until after 0.7.0.)
2618 ** Bug 111 (internal compiler confusion about runtime checks on
2619 FUNCTION types) was fixed by Alexey Dejneka.
2620 * Some internal cleanups (getting rid of variables which aren't
2621 needed now that the byte interpreter is gone) caused the fasl
2622 file format number to change again.
2624 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2625 * major incompatible change: The default fasl file extension, i.e. the
2626 default extension for files produced by COMPILE-FILE, has changed
2627 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2629 ** There are many changes in the implementation of the compiler.
2630 SBCL is now essentially a compiler-only implementation of ANSI
2631 Common Lisp. EVAL still "interprets" a few special cases, but
2632 almost all the interesting cases are handled by creating
2633 a LAMBDA expression, calling COMPILE on it, then calling
2634 FUNCALL on the result.
2635 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2636 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2637 Since the code is newer, there might still be some new bugs
2638 (though not as many as before Martin Atzmueller's fixes:-). But
2639 the new code is substantially simpler and clearer, and hopefully
2640 any remaining bugs will be simpler, less fundamental, and more
2641 fixable then the bugs in the old code.
2642 ** The revised compiler is still a little unsteady on its feet.
2644 *** The debugging information it produces (particularly the names
2645 of FUNCTION objects) is sometimes much less useful than what
2646 the old compiler produced.
2647 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2648 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2649 within a MACROLET) has been temporarily weakened.
2650 ** There are new compiler optimizations for various functions:
2651 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2652 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2653 *** the math functions TRUNCATE, FLOOR, and CEILING
2654 *** the function-of-all-trades COERCE
2655 Mostly these should be transparent, but there's one
2656 potentially-annoying problem (bug 117): when the compiler
2657 inline-expands a function and does type analysis on the result,
2658 it can create control paths which have type mismatches, and
2659 when it can't prove that those control paths aren't taken,
2660 it will issue WARNINGs about the type mismatches. This is
2661 a particular problem in practice for the new sequence functions.
2662 It's not clear how this should be fixed, and for now, a
2663 workaround is given in the entry for 117 in the BUGS file.
2664 ** (Because of the interaction between the two previous items --
2665 occasional inlining problems and new inline expansions -- some
2666 of the new sequence function optimizations won't really kick in
2667 completely until debugging information, and then inlining, are
2668 straightened out in some future version.)
2669 * minor incompatible changes:
2670 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2671 DIRECTORY no longer implicitly promotes NIL slots of its
2672 pathname argument to :WILD. In particular, when you ask for the
2673 contents of a directory (which you used to be able to do without
2674 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2675 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2676 ** changes in behavior that ANSI explicitly defines to be
2677 implementation dependent:
2678 *** The new compiler-only implementation still conforms with ANSI,
2679 but acts a little different than before. Besides the obvious
2680 changes in performance tradeoffs (that the cost per form passed
2681 to EVAL has gone up, and the cost per form executed by EVAL
2682 has gone down), the behavior of the system changes a little
2683 because there are no longer any interpreted function objects.
2684 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2685 e.g. doing COMPILE on the output of interactive DEFUN is
2687 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2689 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2690 has changed from (:CL) to NIL.
2691 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2692 e.g. #\Newline and #\Tab, don't have names specified in the
2693 ANSI Common Lisp standard, is now based on their ASCII symbolic
2694 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2695 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2696 are no longer used for output.
2697 ** changes in internal implementation constants:
2698 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2699 4 million. (If your application spends a lot of time GCing and
2700 you have a lot of RAM, you might want to experiment with
2701 increasing it even more.)
2702 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2703 However, almost all old code should still continue to work without
2704 immediate update, as SB-C-CALL is now a (deprecated) nickname
2706 ** Old operator names in the style DEF-FOO are now deprecated in
2707 favor of new corresponding names DEFINE-FOO, for consistency with
2708 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2709 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2710 internal symbols, but a few supported extensions like
2711 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2712 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2713 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2714 etc. as you get deeper into recursive calls to the debugger
2715 command loop, instead of the old "5]", "5]]", "5]]]"
2716 sequence. (I was motivated to do this when squabbles between
2717 ILISP and SBCL left me very deeply nested in the debugger. In the
2718 short term, this change will probably provoke more ILISP/SBCL
2719 squabbles, but hopefully it will be an improvement in the long run.)
2720 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2721 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2722 function, since declaiming return types involving aliens is
2723 (1) annoyingly messy to do by hand and (2) vital to efficient
2724 compilation of code which calls such functions.
2725 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2726 longer reexported by the SB-EXT package. They're solely useful
2727 for alien code, so it seems more logical that you should get
2728 them from the SB-ALIEN package, not in SB-EXT.
2729 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2730 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2731 features. Instead, the code that they used to control is always
2732 built into the system.
2733 * many other bug fixes
2734 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2735 advantage of the new EVAL-WHEN stuff and to clean them up in
2736 general, and they are now more ANSI-compliant in a number of
2737 ways. Martin Atzmueller is responsible for a lot of this.
2738 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2740 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2741 *** correct ERROR type for various file operations
2742 *** some fixes for Lisp streams
2743 *** DEFMETHOD syntax checking
2744 *** changing old weird representation of debug information as
2745 strings (which, among their other deficiencies, don't transform
2746 correctly when you rename packages, and don't change their
2747 print representation when you change things like *PACKAGE*
2748 and *PRINT-LENGTH*) to symbols and lists of symbols
2749 He also made several improvements and fixed several bugs in DESCRIBE.
2750 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2752 *** misbehavior of WRITE-STRING/WRITE-LINE
2753 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2754 and several other LOOP problems as well
2755 *** DIRECTORY when similar filenames are present
2756 *** DEFGENERIC with :METHOD options
2757 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2758 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2759 *** argument ordering in FIND with :TEST option
2760 *** mishandled package designator argument in APROPOS-LIST
2761 *** various problems in the backquote readmacro
2762 *** a bug in APROPOS
2763 *** probably some others that I'm not describing very well here,
2764 since the CVS log documents them by reference to sbcl-devel
2765 messages, and the SourceForge archives aren't working well.:-(
2766 ** Dan Barlow improved the Alpha port (and is making progress on the
2767 PPC port, for those of you who think different).
2768 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2769 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2770 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2771 continues to work on the SPARC port (for those of you in a position
2772 to look down upon our little PC-compatible boxes from a great height).
2773 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2774 and uses that copy, rather than the current dynamic value, when
2775 it's trying to decide whether to truncate output. Thus e.g.
2776 (let ((*print-lines* 50))
2777 (pprint-logical-block (stream nil)
2779 (let ((*print-lines* 8))
2780 (print (aref possiblybigthings i) stream)))))
2781 should now truncate the logical block only at 50 lines, instead of
2782 often truncating it at 8 lines, as it did before.
2783 * The doc/cmucl/ directory, containing old CMU CL documentation
2784 from the time of the fork, is no longer part of the base system.
2785 SourceForge has shut down its anonymous FTP service, and with it
2786 my original plan for distributing the old CMU CL documentation
2787 there. For now, if you need these files you can download an old
2788 SBCL source release and extract them from it.
2789 * The fasl file version number changed again, for dozens of reasons,
2790 some of which are apparent above.
2792 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2793 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2794 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2795 patch, so that DISASSEMBLE output is much nicer.
2796 * The code in the SB-PROFILE package now seems reasonably stable.
2797 I still haven't decided what the final interface should look like
2798 (I'd like PROFILE to interact cleanly with TRACE, since both
2799 facilities use function encapsulation) but if you have a need
2800 for profiling now, you can probably use it successfully with
2801 the current CMU-CL-style interface.
2802 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2803 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2804 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2805 link now returns the dangling link itself, and for similar
2806 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2807 link itself. (In these cases the old code signalled an error and
2808 looped endlessly, respectively.) Thus, DIRECTORY now works even
2809 in the presence of dangling and cyclic symbolic links.
2810 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2811 is now a supported extension again, since the consensus on
2812 sbcl-devel was that it can be useful for ordinary development
2813 work, not just for debugging SBCL itself.
2814 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2815 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2816 that currently-defined functions might be redefined later with
2817 different return types.
2818 * Hash tables can be printed readably, as inspired by CMU CL code
2819 of Eric Marsden and SBCL code of Martin Atzmueller.
2820 * better error handling in CLOS method combination, thanks to
2821 Martin Atzmueller porting Pierre Mai's CMU CL patches
2822 * more overflow fixes for >16Mbyte I/O buffers
2823 * A bug in READ has been fixed, so that now a single Ctrl-D
2824 character suffices to cause end-of-file on character streams.
2825 In particular, now you only need one Ctrl-D at the command
2826 line (not two) to exit SBCL.
2827 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2829 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2830 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2831 does the right thing.
2832 * The implementation of some type tests, especially for CONDITION
2833 types, is now tidier and maybe faster, due to CMU CL code
2834 originally by Douglas Crosher, ported by Martin Atzmueller.
2835 * Some math functions have been fixed, and there are new
2836 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2837 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2838 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2839 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2840 likely to be built into sbcl-0.7.0.)
2841 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2842 CL, ported to SBCL by Martin Atzmueller)
2843 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2844 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2845 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2846 * There's a new slam.sh hack to shorten the edit/compile/debug
2847 cycle for low-level changes to SBCL itself, and a new
2848 :SB-AFTER-XC-CORE target feature to control the generation of
2849 the after-xc.core file needed by slam.sh.
2850 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2851 COMPILE-FILE is no longer supported, so that now every function
2852 gets an entry point, so that block compilation looks a little
2853 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2854 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2855 returns the number of bytes consed since the system started,
2856 rather than the number consed since the first time the function
2857 was called. (The new definition parallels ANSI functions like
2858 CL:GET-INTERNAL-RUN-TIME.)
2859 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2860 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2861 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2862 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2863 * Fasl file version numbers are now independent of the target CPU,
2864 since historically most system changes which required version
2865 number changes have affected all CPUs equally. Similarly,
2866 the byte fasl file version is now equal to the ordinary
2869 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2870 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2871 is no longer recognized. I apologize for this, because it was
2872 listed in SB-EXT as a supported extension, but I found that
2873 its existing behavior was poorly specified, as well as incorrectly
2874 specified, and it looked like too much of a mess to straighten it
2875 out. I have enough on my hands trying to get ANSI stuff to work..
2876 * many patches ported from CMU CL by Martin Atzmueller, with
2877 half a dozen bug fixes in pretty-printing and the debugger, and
2878 half a dozen others elsewhere
2879 * fixed bug 13: Floating point infinities are now supported again.
2880 They might still be a little bit flaky, but thanks to bug reports
2881 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2882 as flaky as they were.
2883 * The --noprogrammer command line option is now supported. (Its
2884 behavior is slightly different in detail from what the old man
2885 page claimed it would do, but it's still appropriate under the
2886 same circumstances that the man page talks about.)
2887 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2888 are now supported, and enabled by default. Thus, the compiler can
2889 handle many floating point and complex operations much less
2890 inefficiently. (Thus e.g. you can implement a complex FFT
2892 * The compiler now detects type mismatches between DECLAIM FTYPE
2893 and DEFUN better, and implements CHECK-TYPE more correctly, and
2894 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2895 thanks to patches from Martin Atzmueller.
2896 * various fixes to make the cross-compiler more portable to
2897 ANSI-conforming-but-different cross-compilation hosts (notably
2898 Lispworks for Windows, following bug reports from Arthur Lemmens)
2899 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2900 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2901 fixed thanks to Pierre Mai's CMU CL patches.
2902 * fixes to keep the system from overflowing internal counters when
2903 it tries to use i/o buffers larger than 16M bytes
2904 * fixed bug 45a: Various internal functions required to support
2905 complex special functions have been merged from CMU CL sources.
2906 (When I was first setting up SBCL, I misunderstood a compile-time
2907 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2908 * improved support for type intersection and union, fixing bug 12
2909 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2910 more obscure bugs as well
2911 * some steps toward byte-compiling non-performance-critical
2912 parts of the system, courtesy of patches from Martin Atzmueller
2913 * Christophe Rhodes has made some debian packages of sbcl at
2914 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2915 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2916 stable, but are nonetheless usable. When he's ready, I'd be happy
2917 to add them to the SourceForge "File Releases" section. (And if
2918 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2919 * new fasl file format version number (because of changes in
2920 internal representation of (OR ..) types to accommodate the new
2921 support for (AND ..) types, among other things)
2923 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2924 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2926 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2928 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2929 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2930 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2931 * DESCRIBE now works on structure objects again.
2932 * Most function call argument type mismatches are now handled as
2933 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2934 know whether the function will be redefined before the call is
2935 executed. (The compiler could flag local calls with full WARNINGs,
2936 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2937 it doesn't keep track of enough information to know whether calls
2938 are local in this sense.)
2939 * Compiler output is now more verbose, with messages truncated
2940 later than before. (There should be some supported way for users
2941 to override the default verbosity, but I haven't decided how to
2942 provide it yet, so this behavior is still controlled by the internal
2943 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2944 src/compiler/ir1util.lisp.)
2945 * Fasl file format version numbers have increased again, because
2946 support for the Gray streams extension changes the layout of the
2947 system's STREAM objects.
2948 * The Gray subclassable streams extension now works, thanks to a
2949 patch from Martin Atzmueller.
2950 * The full LOAD-FOREIGN extension (not just the primitive
2951 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2952 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2953 but like most other programs, it defaults to copying the Unix
2954 environment from the original process instead of starting the
2955 new process in an empty environment.
2956 * Extensions which manipulate the Unix environment now support
2957 an :ENVIRONMENT keyword option which doesn't smash case or
2958 do other bad things. The CMU-CL-style :ENV option is retained
2959 for porting convenience.
2960 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2961 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2963 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2965 * A patch from Martin Atzmueller seems to have solved the SIGINT
2966 problem, and as far as we know, signal-handling now works cleanly.
2967 (If you find any new bugs, please report them!)
2968 * The system no longer defaults Lisp source file names to types
2969 ".l", ".cl", or ".lsp", but only to ".lisp".
2970 * The compiler no longer uses special default file extensions for
2971 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2972 seems to expect a single default extension for all compiled code,
2973 and there's no compelling reason to try to stretch the standard
2974 to allow two different extensions.) Instead, byte-compiled files
2975 default to the same extension as native-compiled files.
2976 * Fasl file format version numbers have increased again, because
2977 a rearrangement of internal implementation packages made some
2978 dumped symbols in old fasl files unreadable in new cores.
2979 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2980 many fewer weird special cases.
2981 * Bug #17 (differing COMPILE-FILE behavior between logical and
2982 physical pathnames) has been fixed, and some related misbehavior too,
2983 thanks to a patch from Martin Atzmueller.
2984 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2985 by Tim Moore, ported to SBCL by Martin Atzmueller.
2986 * Martin Atzmueller fixed several filesystem-related problems,
2987 including bug #36, in part by porting CMU CL patches, which were
2988 written in part by Paul Werkowski.
2989 * More compiler warnings in src/runtime/ are gone, thanks to
2990 more patches from Martin Atzmueller.
2991 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2994 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2996 * DESCRIBE now works on CONDITION objects.
2997 * The debugger now handles errors which arise when trying to print
2998 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3000 * The build system now uses an additional file, customize-target-features.lisp,
3001 to allow local modifications to the target *FEATURES* list. (The point of
3002 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3003 debugging features enabled, without having to worry about propagating it
3004 into everyone's system when I do a "cvs update".) When no
3005 customize-target-features.lisp file exists, the target *FEATURES* list
3006 should be constructed the same way as before.
3007 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3008 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3009 warn about behavior which is undefined under the ANSI spec. Note: This
3010 is specified by ANSI, but it's not very popular with programmers.
3011 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3012 macro in the SBCL sources for an example of a workaround which you
3013 might use to make portable ANSI-standard code which does what you want.
3014 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3015 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3016 when it's used as a non-toplevel form. (This is required in order
3017 to implement the DEFCONSTANT-EQX macro.)
3018 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3019 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3020 funny things, due to bugs in the implementation of EVAL-WHEN
3021 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3022 non-toplevel DEFCONSTANTs are uncommon.)
3023 * The core file version number and fasl file version number have been
3024 incremented, because the old noncompliant DEFCONSTANT behavior involved
3025 calling functions which no longer exist, and because I also took the
3026 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3027 * fixed bug 1 (error handling before read-eval-print loop starts), and
3028 redid debugger restarts and related debugger commands somewhat while
3030 ** The QUIT debugger command is gone, since it did something
3031 rather different than the SB-EXT:QUIT command, and since it never
3032 worked properly outside the main toplevel read/eval/print loop.
3033 Invoking the new TOPLEVEL restart provides the same functionality.
3034 ** The GO debugger command is also gone, since you can just invoke
3035 the CONTINUE restart directly instead.
3036 ** The TOP debugger command is also gone, since it's redundant with the
3037 FRAME 0 command, and since it interfered with abbreviations for the
3039 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3041 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3042 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3043 some porting work by Martin Atzmueller)
3044 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3045 with a boatload of patches which he ported to SBCL. Now that those
3047 ** The system tries to make sure that its low-priority messages
3048 are prefixed by semicolons, to help people who like to use
3049 syntax highlighting in their ILISP buffer. (This patch
3050 was originally due to Raymond Toy.)
3051 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3052 patches originally written by Raymond Toy.
3053 ** The compiler understands coercion between single-value and
3054 multiple-VALUES type expressions better, getting rid of some very
3055 weird behavior, thanks to patches originally by Robert MacLachlan
3056 and Douglas Crosher.
3057 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3058 lambda lists, thanks to a patch originally by Pierre Mai.
3059 ** The system no longer bogusly warns about "abbreviated type
3061 ** The compiler gets less confused by inlining and RETURN-FROM,
3062 thanks to some patches originally by Tim Moore.
3063 ** The system no longer hangs when dumping circular lists to fasl
3064 files, thanks to a patch originally from Douglas Crosher.
3065 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3066 undefined function error.
3067 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3068 in structure slot initforms) for now, documented workaround instead:-|
3069 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3070 * fixed bug 5: added stubs for various Gray stream functions called
3071 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3072 installed, at least appropriate type errors are generated
3073 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3074 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3075 consistently in DEFMETHOD forms.
3076 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3077 it doesn't seem to affect SBCL after all
3078 * The C runtime system now builds with better optimization and many
3079 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3081 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3083 * The system is now under CVS at SourceForge (instead of the
3084 CVS repository on my home machine).
3085 * The new signal handling code has been tweaked to treat register
3086 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3087 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3088 downstream, e.g. in debug-int.lisp, has implicit dependencies
3089 on the unsignedness of integer representation of machine words,
3090 and that caused the system to bomb out with infinite regress
3091 when trying to recover from type errors involving signed values,
3092 e.g. (BUTLAST '(1 2 3) -1).)
3093 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3094 (This was one of the bugs Peter Van Eynde reported back in July.)
3095 * The system now uses code inspired by Colin Walters' O(N)
3096 implementation of MAP (from the cmucl-imp@cons.org mailing
3097 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3098 inline the MAP operation, and there is more than one
3099 sequence argument to the MAP call (so that it can't just
3100 do ETYPECASE once and for all based on the type of the
3101 single sequence argument). (The old non-inline implementation
3102 of the general M-argument sequence-of-length-N case required
3103 O(M*N*N) time when any of the sequence arguments were LISTs.)
3104 * The QUIT :UNIX-CODE keyword argument has been renamed to
3105 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3106 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3108 * Raymond Wiker's patches to port dynamic loading from Linux to
3109 FreeBSD have been added.
3110 * The BUGS file is now more nearly up to date, thanks in large part
3111 to Martin Atzmueller's review of it.
3112 * The debugger now flushes standard output streams before it begins
3113 its output ("debugger invoked" and so forth).
3114 * The core version number and fasl file version number have both
3115 been incremented, because of incompatible changes in the layout
3117 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3118 used to not be called for a saved Lisp image.)
3119 * Martin Atzmueller's version of a patch to fix a compiler crash,
3120 as posted on sbcl-devel 13 September 2000, has been installed.
3121 * Instead of installing Martin Atzmueller's patch for the
3122 compiler transform for SUBSEQ, I deleted the compiler transform,
3123 and transforms for some similar consing operations.
3124 * A bug in signal handling which kept TRACE from working on OpenBSD
3126 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3127 away by constant folding
3128 * The system now defines its address space constants in one place
3129 (in the Lisp sources), and propagates them automatically elsewhere
3130 (through GENESIS and the sbcl.h file). Therefore, patching the
3131 address map is less unnecessarily tedious and error-prone. The
3132 Lisp names of address space constants have also been systematized.
3133 * CVS tags like dollar-Header-dollar have been removed from
3134 the sources, because they have never saved me trouble and
3135 they've been source of trouble working with patches and other
3136 diff-related operations.
3137 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3138 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3140 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3142 * The system has been ported to OpenBSD.
3143 * The system now compiles with a simple "sh make.sh" on the systems
3144 that it's supported on. I.e., now you no longer need to tweak
3145 text in the target-features.lisp-expr and symlinks in src/runtime/
3146 by hand, the make.sh takes care of it for you.
3147 * The system is no longer so grossly inefficient when compiling code
3148 involving vectors implemented as general (not simple) vectors (VECTOR T),
3149 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3150 which dares to use the various sequence functions on non-simple
3151 vectors, takes less of a performance hit.
3152 * There is now a primitive type predicate VECTOR-T-P
3153 to test for the (VECTOR T) type, so that e.g.
3154 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3155 can now be compiled with some semblance of efficiency. (The old code
3156 turned the type declaration into a full call to %TYPEP at runtime!)
3157 * AREF on (VECTOR T) is still not fast, since it's still compiled
3158 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3159 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3160 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3161 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3162 without making full calls to SUBTYPEP at runtime.
3163 (Some analogous efficiency issues for non-simple vectors specialized to
3164 element types other than T, or for non-simple multidimensional arrays,
3165 have not been addressed. They could almost certainly be handled the
3166 same way if anyone is motivated to do so.)
3167 * The changes in array handling break binary compatibility, so
3168 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3169 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3170 instead of (VALUES T T).
3171 * By following the instructions that Dan Barlow posted to sbcl-devel
3172 on 2 July 2000, I was able to enable primitive dynamic object
3173 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3174 functionality is not implemented (since it calls ld to resolve
3175 library references automatically, requiring RUN-PROGRAM for its
3176 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3177 try to resolve library references) is now supported.
3178 * The system now flushes the standard output streams when it terminates,
3179 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3180 them at several other probably-convenient times, e.g. in each pass of
3181 the toplevel read-eval-print loop, and after evaluating a form given
3182 as an "--eval" command-line option. (These changes were motivated by a
3183 discussion of stream flushing issues on cmucl-imp in August 2000.)
3184 * The source transform for TYPEP of array types no longer assumes
3185 that an array whose element type is a not-yet-defined type
3186 is implemented as an array of T, but instead punts, so that the
3187 type will be interpreted at runtime.
3188 * There is now some support for cross-compiling in make.sh: each of
3189 the phases of make.sh has its own script. (This should be transparent
3190 to people doing ordinary, non-cross-compile builds.)
3191 * Since my laptop doesn't have hundreds of megabytes of memory like
3192 my desktop machine, I became more motivated to do some items on
3193 my to-do list in order to reduce the size of the system a little:
3194 ** Arrange for various needed-only-at-cold-init things to be
3195 uninterned after cold init. To support this, those things have
3196 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3197 symbols with such names are now uninterned after cold init).
3198 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3199 fasl files for cold load.
3200 ** Remove the old compiler structure pooling code (which used to
3201 be conditional on the target feature :SB-ALLOC) completely.
3202 ** Redo the representation of some data in cold init to be more compact.
3203 (I also looked into supporting byte compiled code at bootstrap time,
3204 which would probably reduce the size of the system a lot, but that
3205 looked too complicated, so I punted for now.)
3206 * The maximum signal nesting depth in the src/runtime/ support code has
3207 been reduced from 4096 to 256. (I don't know any reason for the very
3208 large old value. If the new smaller value turns out to break something,
3209 I'll probably just bump it back up.)
3210 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3212 * Many, many bugs reported by Peter Van Eynde have been added to
3213 the BUGS list; some have even been fixed.
3214 * While enabling dynamic object file loading, I tried to make the
3215 code easier to understand, renaming various functions and variables
3216 with less ambiguous names, and changing some function calling
3217 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3218 * While trying to figure out how to do the OpenBSD port, I tried to
3219 clean up some of the code in src/runtime/. In particular, I dropped
3220 support for non-POSIX signal handling, added various comments,
3221 tweaked the code to reduce the number of compilation warnings, and
3222 renamed some files to increase consistency.
3223 * To support the new automatic configuration functionality in make.sh,
3224 the source file target-features.lisp-expr has been replaced with the
3225 source file base-target-features.lisp-expr and the machine-generated
3226 file local-target-features.lisp-expr.
3227 * fixed a stupid quoting error in make.sh so that using CMU CL
3228 "lisp -batch" as cross-compilation host works again
3230 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3232 * DESCRIBE no longer tries to call itself recursively to describe
3233 bound/fbound values, so that it no longer fails on symbols which are
3234 bound to themselves (like keywords, T, and NIL).
3235 * DESCRIBE now works on generic functions.
3236 * The printer now prints less-screwed-up representations of closures
3237 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3238 * A private symbol is used instead of the :EMPTY keyword previously
3239 used to mark empty slots in hash tables. Thus
3240 (DEFVAR *HT* (MAKE-HASH-TABLE))
3241 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3242 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3243 now does what ANSI says that it should. (You can still get
3244 similar noncompliant behavior if bang on the hash table
3245 implementation with all the symbols you get back from
3246 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3247 This breaks binary compatibility, since tests for equality to
3248 :EMPTY are wired into things like the macroexpansion of
3249 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3251 * There's now a minimal placeholder implementation for CL:STEP,
3252 as required by ANSI.
3253 * An obscure bug in the interaction of the normal compiler, the byte
3254 compiler, inlining, and structure predicates has been patched
3255 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3256 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3258 * Missing ordinary arguments in a macro call are now detected even
3259 when the macro lambda list contains &KEY or &REST.
3260 * The debugger no longer complains about encountering the top of the
3261 stack when you type "FRAME 0" to explicitly instruct it to go to
3262 the top of the stack. And it now prints the frame you request even
3263 if it's the current frame (instead of saying "You are here.").
3264 * As specified by ANSI, the system now always prints keywords
3265 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3267 * The default initial SIZE of HASH-TABLEs is now smaller.
3268 * Type information from CLOS class dispatch is now propagated
3269 into DEFMETHOD bodies, so that e.g.
3270 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3272 is now basically equivalent to
3273 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3274 (DECLARE (TYPE SINGLE-FLOAT X))
3276 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3277 operation, without having to do run-time type dispatch.
3278 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3279 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3280 * Since I don't seem to be making big file reorganizations very often
3281 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3282 is overflowing my ability to conveniently back them up), I've finally
3283 checked the system into CVS. (The CVS repository is on my home system,
3284 not at SourceForge -- putting it on SourceForge might come later.)
3285 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3286 high-level GC-NOTIFY-FOO functions send their output. (There's
3287 still very little control of where low-level verbose GC functions
3288 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3289 less than it used to -- the GC-NOTIFY-FOO functions are now under
3290 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3291 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3292 in only one place in the source code, and propagates it automatically
3293 everywhere that it's needed. Thus e.g. when I bump the version from
3294 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3295 * The C source files now include boilerplate legalese and documentation
3296 at the head of each file (just as the Lisp source files already did).
3297 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3298 to his page will be replaced with a link to his new CLiki service.
3300 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3302 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3303 * The build process now looks for GNU make under the default name "gmake",
3304 instead of "make" as it used to. If GNU make is not available as "gmake"
3305 on your system, you can change this default behavior by setting the
3306 GNUMAKE environment variable.
3307 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3308 can build without error under CMU CL.
3310 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3312 * There is now a partial SBCL user manual (with some new text and some
3313 text cribbed from the CMU CL manual).
3314 * The beginnings of a profiler have been added (starting with the
3315 CMU CL profiler and simplifying and cleaning up). Eventually the
3316 main interface should be through the TRACE macro, but for now,
3317 it's still accessed through vaguely CMU-CL-style functions and macros
3318 exported from the package SB-PROFILE.
3319 * Some problems left over from porting CMU CL to the new
3320 cross-compilation bootstrap process have been cleaned up:
3321 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3322 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3323 part when getting rid of the compiler *BACKEND* stuff.)
3324 ** The value of *NULL-TYPE* was screwed up, because it was
3325 being initialized before the type system knew the final
3326 definition of the 'NULL type. This screwed up several key
3327 optimizations in the compiler, causing inefficiency in all sorts
3328 of places. (I found it because I wanted to understand why
3329 GET-INTERNAL-RUN-TIME was consing.)
3330 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3331 PROCLAIM FTYPE information. Unfortunately this broke binary
3332 compatibility again, since now the forms output by DEFGENERIC
3333 to refer to functions which didn't exist in 0.6.3.
3334 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3335 can use the new (as of 0.6.3) transform for SOME into MAP into
3337 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3338 values to (INTEGER 0 1000000), so that the time code will no longer
3339 occasionally get blown up by Linux returning 1000000 microseconds
3340 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3341 its output conform to the ANSI spec. (Alas, this makes its output
3342 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3343 * A full call to MAP NIL with a single sequence argument no longer conses.
3344 * fixes to problems pointed out by Martin Atzmueller:
3345 * The manual page no longer talks about multiprocessing as though
3346 it were currently supported.
3347 * The ILISP support patches have been removed from the distribution,
3348 because as of version 5.10.1, ILISP now supports SBCL without us
3349 having to maintain patches.
3350 * added a modified version of Raymond Toy's recent CMU CL patch for
3351 EQUALP comparison of HASH-TABLE
3353 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3355 * The system still can't cross-compile itself with
3356 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3357 entails), but at least it can compile more of itself that way
3358 than it used to be able to, and various buglets which were uncovered
3359 by trying to cross-compile itself that way have now been fixed.
3360 * This release breaks binary compatibility again. This time
3361 at least I've incremented the FASL file format version to 2, so that the
3362 problem can be detected reliably instead of just causing weird errors.
3363 * various new style warnings:
3364 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3365 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3366 ** using the lexical binding of a variable named in the *FOO* style
3367 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3368 as specified by ANSI.
3369 * *RANDOM-STATE* is no longer automatically initialized from
3370 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3371 default behavior of the system is to repeat its behavior every time
3372 it's run. If you'd like to change this behavior, you can always
3373 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3374 old convention there was no comparably easy way to get the system to
3375 repeat its behavior every time it was run.
3376 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3377 FTYPE declarations has been removed, in favor of their ANSI
3378 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3379 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3380 the types of their sequence arguments can be determined at compile time.
3381 This is done through a new open code expansion for MAP which eliminates
3382 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3383 argument types can be determined at compile time.
3384 * The optimizer now transforms COERCE into an identity operation when it
3385 can prove that the coerced object is already of the correct type. (This
3386 can be a win for machine generated code, including the output of other
3387 optimization transforms, such as the MAP transform above.)
3388 * Credit information has been moved from source file headers into CREDITS.
3389 * Source file headers have been made more standard.
3390 * The CASE macro now compiles without complaining even when it has
3393 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3395 * (Note that the way that the PCL macroexpansions were rewritten
3396 to accommodate the change in DEFGENERIC below breaks binary
3397 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3398 not run under sbcl-0.6.2. Once we get out of alpha releases,
3399 i.e. hit release 1.0.0, we'll probably try to maintain binary
3400 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3401 and sbcl-1.4.4. Until then, however, it might be fairly common
3402 for maintenance releases to break binary compatibility.)
3403 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3405 * The reporting of unhandled signals has been changed to print some
3406 explanatory text as well as the report form. (Previously only
3407 the report form was printed.)
3408 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3409 it defines, so that the compiler no longer issues undefined function
3410 warnings for compiled-but-not-yet-loaded generic functions.
3411 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3412 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3413 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3414 section of the ANSI spec are supported.
3415 * The "" nickname for the "KEYWORD" package has been removed.
3416 The reader still handles symbol tokens which begin with a package marker
3417 as keywords, but it doesn't expose its mechanism for doing so in the
3418 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3419 * The system now issues STYLE-WARNINGs for contradictory TYPE
3420 proclamations. (Warnings for contradictory FTYPE proclamations would
3421 be nice too, but those can't be done usefully unless the type system
3422 is made smarter about FUNCTION types.)
3423 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3424 systematized, so that "*target-*.lisp is supposed to exist only on the
3425 target and imply that there's a related file which exists on the
3426 host, and *host-*.lisp is supposed to exist only on the host and imply
3427 that there's a related file which exists on the target. This involves a
3428 lot of renaming. Hopefully the acute confusion caused by the renaming
3429 will be justified by the reduction in chronic confusion..
3430 ** runtime-type.lisp -> early-target-type.lisp
3431 ** target-type.lisp -> late-target-type.lisp
3432 ** early-host-format.lisp -> early-format.lisp
3433 ** late-host-format.lisp -> late-format.lisp
3434 ** host-error.lisp -> misc-error.lisp
3435 ** early-error.lisp -> early-target-error.lisp
3436 ** late-error.lisp -> late-target-error.lisp
3437 ** host-defboot.lisp -> early-defboot.lisp
3438 ** code/misc.lisp -> code/target-misc.lisp
3439 ** code/host-misc.lisp -> code/misc.lisp
3440 ** code/numbers.lisp -> code/target-numbers.lisp
3441 ** code/early-numbers.lisp -> numbers.lisp
3442 ** early-host-type.lisp -> early-type.lisp
3443 ** late-host-type.lisp -> late-type.lisp
3444 ** host-typep.lisp -> typep.lisp
3445 ** load.lisp -> target-load.lisp
3446 ** host-load.lisp -> load.lisp
3447 ** host-disassem.lisp -> disassem.lisp
3448 ** host-insts.lisp -> insts.lisp
3449 ** byte-comp.lisp -> target-byte-comp.lisp
3450 ** host-byte-comp.lisp -> byte-comp.lisp
3451 ** host-signal.lisp -> signal.lisp
3452 ** host-defstruct.lisp -> defstruct.lisp
3453 ** late-target-type.lisp -> deftypes-for-target.lisp
3454 Furthermore, several other previously target-only files foo.lisp (e.g.
3455 hash-table.lisp and random.lisp) have been split into a target-and-host
3456 foo.lisp file and a target-only target-foo.lisp file, with their key type
3457 definitions in the target-and-host part, so that the cross-compiler will
3458 know more about target types.
3459 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3460 gone away. In their place are various *BACKEND-FOO* variables
3461 corresponding to the slots of the old structure.
3462 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3463 into the target SBCL has been fixed.
3464 * The chill.lisp system for loading cold code into a running SBCL
3466 * Support for the CMU CL "scavenger hook" extension has been removed.
3467 (It was undocumented and unused in the CMU CL sources that SBCL was
3468 derived from, and stale in sbcl-0.6.1.)
3469 * Various errors in the cross-compiler type system were detected
3470 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3471 (enabling various consistency checks). Many of them were fixed,
3472 but some hard problems remain, so the compiler is back to
3473 running without *TYPE-SYSTEM-INITIALIZED* for now.
3474 * As part of the cross-compiler type system cleanup, I implemented
3475 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3476 * I have started adding UNCROSS calls throughout the type system
3477 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3478 on cross-compiler input files will be able to go away, and various
3480 * CONSTANTP now returns true for quoted forms (as explicitly required
3483 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3485 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3486 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3487 (MAKE-LIST -1). In the longer term, it ought to become true
3488 that declarations are assertions even at SAFETY 1. For now, it's not
3489 quite true even at SAFETY 3, but it's at least more nearly true..
3490 (Note that this change seems to increases the size of the system by
3491 O(5%) and to decrease the speed of the compiler by 20% or more.)
3492 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3493 for the problem of printing dozens of lines of distracting information
3494 about low-level system machinery as part of the top stack frame
3495 on entry to the debugger when an undefined function was called.
3496 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3497 is not reset to COMMON-LISP-USER.
3498 * Compilation of stuff related to dyncount.lisp has been made conditional
3499 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3500 smaller. The various dyncount-related symbols have been moved into
3501 a new "SB-DYNCOUNT" package.
3502 * tty-inspect.lisp has been renamed to inspect.lisp.
3503 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3504 feature has gone away. (When we eventually port to other flavors of
3505 libc and/or Unix, we'll try to make the differences between flavors
3506 invisible at the user level.)
3507 * Various other *FEATURES* tags, and/or their associated conditionals,
3508 have been removed if obsolescent, or given better documentation, or
3509 sometimes given more-mnemonic names.
3511 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3513 * tidied up "make.sh" script
3514 * tidied up system directory structure
3515 * better "clean.sh" behavior
3516 * added doc/FOR-CMUCL-DEVELOPERS
3517 * many many small tweaks to output format, e.g. removing possibly-confusing
3518 trailing #\. character in DESCRIBE-INSTANCE
3519 * (EQUALP #\A 'A) no longer signals an error.
3520 * new hashing code, including EQUALP hashing
3521 * tidied up Lisp initialization and toplevel
3522 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3523 * command line argument processing
3524 * added POSIX-GETENV function to deal with Unix-ish environment variables
3525 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3526 terminating SBCL on EOF
3527 * non-verbose GC by default
3528 * There is no more "sbcl" shell script; the sbcl file is now the C
3529 runtime executable (just like CMU CL).
3530 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3532 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3533 symbol and package information as Lisp native symbol and package objects
3534 instead of strings naming symbols and strings naming packages. This way,
3535 whenever packages are renamed (as in warm init), debug information is
3536 transformed along with everything else.
3537 * tweaked the optimization policy declarations which control the building
3538 of SBCL itself. Now, among other things, the system no longer saves
3539 source location debugging information. (This helps two problems at once
3540 by reducing SBCL size and by keeping SBCL from trying to look for its
3541 sources -- which may not exist -- when reporting errors.)
3542 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3543 debugging and testing purposes
3544 * cleaned up printing, making the printer call PRINT-OBJECT for
3545 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3546 methods, giving nearly-ANSI behavior
3547 * converted almost all special variables to use *FOO* naming convention
3548 * deleted PARSE-TIME functionality, since it can be done portably
3549 * moved some files out of cold init into warm init
3550 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3552 * regularized formatting of source files
3553 * added an install.sh script
3554 * fixed ridiculous memory usage of cross-compiler by making
3555 compiler/alloc.lisp not try to do pooling unless it can hook
3556 itself into the GC of the cross-compilation host. Now the system
3557 builds nicely on my old laptop.
3558 * added :SB-ALLOC in target-features.lisp-expr
3559 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3560 was not implemented)
3561 * re-did condition handling and note reporting in the compiler. Notes
3562 are no longer handled by signalling conditions. Style warnings
3563 and warnings are handled more correctly and reported in such a way
3564 that it's easy to find one or the other in your output (so that you
3565 can e.g. figure out which of many problems caused COMPILE-FILE to
3567 * changed the severity of several compiler warnings from full WARNING
3568 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3569 compiler note reporting so that it doesn't use the condition system
3570 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3571 COMPILE-FILE command)
3572 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3573 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3574 no longer supported, since their implementation was deeply intertwingled
3575 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3576 * removed bogus "support" for compiler macros named (SETF FOO), and
3577 removed the compiler macro for SETF INFO (but only after making a fool
3578 of myself on the cmucl-imp mailing list by posting a bogus patch for
3579 DEFINE-COMPILER-MACRO..)
3580 * Compiled files containing forms which have side effects on the Lisp
3581 reader (such as DEFPACKAGE forms) are now handled more correctly.
3582 (Compiler queuing of top level lambdas has been suppressed by setting
3583 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3584 * deleted various currently-unused source files, e.g. gengc.lisp. They
3585 may be added back at some point e.g. when porting to other architectures,
3586 but until they are it's distracting to distribute them and to try to
3588 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3589 there were so many of them they're just distractions, and UNCROSS is
3590 known to be able to handle the current sources
3591 * moved PROFILE functionality into TRACE, so that it will be clear
3592 how the wrapping and unwrapping of functions when you profile them
3593 interacts with the wrapping and unwrapping of functions when you
3594 trace them. (Actually, the functionality isn't there yet, but at least
3595 the interface specification is there. Hopefully, the functionality will
3596 arrive with some future maintenance release.)
3597 * removed host-oops.lisp
3598 * changed signature of QUIT function to allow UNIX-CODE argument
3599 * fixed READ-SEQUENCE bug
3600 * tweaked verbose GC output so that it looks more like the progress
3601 output that ANSI specifies for functions like LOAD
3602 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3603 * added <http://sbcl.sourceforge.com> to the banner information printed by