1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
3 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
4 certain complicated string types. (reported by Gary King)
5 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
6 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
7 * fixed bug: DOCUMENTATION on structures no longer signals an
8 error if no structure type of the right name exists. (reported by
11 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
12 * new platform: experimental support for the Windows operating
13 system has been added. (thanks to Alastair Bridgewater)
14 * fixed several bugs in and robustified the PPC FFI (including
15 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
16 * bug fix: allow non-simple string symbol names (reported by Paul
18 * bug fix: interrupt handling on NetBSD (thanks to Richard M
20 * bug fix: saving a core corrupted callbacks on x86/x86-64
21 * bug fix: closed a loophole in metacircularity detection and
22 grounding in the PCL implementation of CLOS.
23 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
24 (reported by tomppa on #lisp)
25 * bug fix: clear the direction flag on Lisp -> C transitions, as
26 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
27 (reported by Andrej Grozin and Hendrik Maryns)
28 * optimization: major improvements to GC efficiency on GENCGC platforms
29 * optimization: faster implementation of EQUAL
30 * optimization: emit more efficient opcodes for some common
31 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
32 * optimization: several other minor code-generation improvements on x86-64
33 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
35 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
36 * minor incompatible change: (SETF CLASS-NAME) and (SETF
37 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
38 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
39 by ANSI as a generic function, and for consistency (SETF
40 GENERIC-FUNCTION-NAME) is treated likewise.
41 * fixed bug #233.b: make constraint propagation notice when a variable
42 value is changed after it is referenced but before it is used
43 * fixed bug #296: no more arbitrary behaviour from filenames with
44 odd characters as --load arguments. (reported by Adam Warner)
45 * fixed bug #390: :CHARACTER-SET pathname components now work as
46 expected. (reported by Tim Daly Jr)
47 * fixed bug #391: complicated :TYPE intersections in slot
48 definitions no longer cause an error in PCL internals.
49 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
50 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
51 returning the number of octets which would be written to the
52 file-stream. (thanks to Robert J. Macomber)
53 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
54 arguments to the internal stream functions. (thanks to David
56 * bug fix: stack exhaustion is now handled on the x86/SunOS
57 platform. (thanks to Daisuke Homma)
58 * bug fix: more accurate ROOM results on GENCGC platforms
59 * optimization: improved type inference for arithmetic-for
60 index variables in LOOP
61 * optimization: faster floating-point SQRT on x86-64
62 * fixed some bugs revealed by Paul Dietz' test suite:
63 ** DOCUMENTATION returns NIL instead of "" for method combinations
64 that don't have a docstring
66 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
67 * minor incompatible change: (SETF CLASS-NAME) and (SETF
68 GENERIC-FUNCTION-NAME) are no longer generic functions, and
69 therefore are not customizeable by user code (as seems to be at
70 least permitted and maybe required by AMOP). As a consolation,
71 however, the SBCL implementation of these functions now calls
72 REINITIALIZE-INSTANCE as specified by AMOP.
73 * bug fix: slot-definition documentation is propagated as per ANSI
74 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
75 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
76 argument, as do eslotd initialization methods. (from Pascal
77 Costanza's "Closer" project)
78 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
79 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
81 * bug fix: it is now possible to have more than one subclass of
82 STANDARD-GENERIC-FUNCTION without causing stack overflow.
83 (reported by Bruno Haible, Pascal Costanza and others)
84 * bug fix: the dependent update protocol now works for generic
85 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
87 * bug fix: condition-class instances corresponding to
88 DEFINE-CONDITION forms are now created eagerly. (reported by
89 Kalle Olavi Niemitalo on comp.lang.lisp)
90 * bug fix: floating point printing is more accurate in some
91 circumstances. (thanks to Simon Alexander)
92 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
93 merged with *DEFAULT-PATHNAME-DEFAULTS*.
94 * bug fix: callbacks on OS X now preserve stack-alignment.
95 (thanks to Cyrus Harmon)
96 * enhancement: the x86-64 disassembler is much better at
97 disassembling SSE instructions. (thanks to Lutz Euler)
98 * enhancement: CHECK-TYPE on a variable now allows type-inference
99 (thanks to Vincent Arkesteijn)
100 * optimization: improved performance of EUC-JP external format.
101 (thanks to NIIMI Satoshi)
102 * optimization: performance improvements to IO on file streams of
103 :ELEMENT-TYPE CHARACTER
104 * optimization: much faster memory allocation on x86-64
105 * optimization: faster garbage collections (latency and throughput)
107 * optimization: faster bignum subtraction and division on x86 and x86-64
108 (thanks to Lutz Euler)
109 * optimization: more accurate type-inference for loops that use a
110 floating point index variable or a negative step.
112 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
113 * bug fix: add a workaround to SBCL looping infinitely at startup on
114 Linux kernels with apparently buggy implementations of personality().
115 (thanks to Svein Ove Aas)
116 * bug fix: Unicode symbols are correctly printed in LDB backtraces
117 (thanks to David Lichteblau)
118 * bug fix: local bindings shadow global symbol macros for
119 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
120 * optimization: non-open coded uses of numeric comparison operators
121 (e.g. >) no longer cons when called with more than one parameter
122 on platforms supporting dynamic-extent allocation.
123 * enhancement: saving cores with foreign code loaded is now supported
124 on MIPS/Linux in addition to the previously supported platforms.
125 * bug fix: FIND-RESTART now tests for activity, not applicability when given
126 a restart object as identifier. (reported by Helmut Eller for CMUCL)
127 * bug fix: division by zero in sb-sprof when no samples were collected
128 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
129 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
130 (thanks to Alexander Kjeldaas)
131 * new feature: ignoring the return values of destructive functions
132 like NREVERSE signals a compile-time style-warning.
133 (thanks to Kevin Reid)
134 * new feature: the alignment of alien structure fields can be
135 explicitly specified. (thanks to Cyrus Harmon)
136 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
137 and dump core on SIGQUIT
139 ** incompatible change: threads do not inherit values of specials
140 from their parents (see manual)
141 ** bug fix: threads stacks belonging to dead threads are freed by the
142 next exiting thread, no need to gc to collect thread stacks anymore
143 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
144 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
145 ** bug fix: dynamic variable and thread start related gc lossage
146 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
148 * fixed some bugs revealed by Paul Dietz' test suite:
149 ** SUBTYPEP is slightly more accurate on heinously complicated
150 CONS types where some of the members have uncertain (in the
151 NIL, NIL sense) type relationships to each other.
153 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
154 no longer purified unless :PURIFY T is explicitly specified.
155 ** Non-purified cores are significantly smaller than before
157 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
158 * new feature: timers based on Zach Beane's excellent timer package
159 * added support for the following external formats: koi8-u,
160 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
161 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
162 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
163 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
164 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
165 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
166 * incompatible change: a threaded SBCL will no longer revert to
167 non-threaded mode on non-NPTL systems, but refuse to start entirely.
168 * bug fix: interrupts are disabled until startup is complete; no
169 more sigsegvs when receiving a signal to soon
170 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
172 * bug fix: add a workaround for the memory randomization features in
173 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
174 workaround will only be in effect on systems with the proc filesystem
176 * bug fix: printing objects of type HASH-TABLE signals a
177 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
179 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
180 have been running for over 50 days. (reported by Gilbert Baumann)
181 * bug fix: the logic for getting names of functions gets less
182 confused when confronded with alternate-metaclass
183 funcallable-instances. (reported by Cyrus Harmon)
184 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
185 consistent, even on internal alternate-metaclass objects.
186 * bug fix: flush closure information collected by physical
187 environment analysis prepass before the main pass. (bug reported
189 * bug fix: compiler pack phase does not modify a hashtable, which is
190 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
191 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
192 may be expanded. (reported by Alexander <malishev> on c.l.l)
193 * bug fix: no more style-warnings when compiling code using LAST
194 under high-SPEED low-SPACE settings. (reported by David Wallin
196 * fixed some incompatibilities between SBCL's MOP and the MOP
198 ** the METAOBJECT class is now implemented;
199 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
200 STANDARD-OBJECT, as required;
201 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
202 now compatible (as required by VALIDATE-SUPERCLASS); there
203 remains a consistency requirement over the presence or absence
204 of the FUNCTION class in the superclass list of finalized
205 classes; see the manual for more details;
206 ** the :AROUND method for COMPUTE-SLOTS on
207 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
208 requested slot ordering.
210 ** bug fix: parent thread now can be gc'ed even with a live
212 ** bug fix: binding a special with PROGV to no value is not
213 the same as that symbol not having been bound (thanks to
215 ** bug fix: binding specials is thread safe (thanks to
217 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
218 and SB-SPROF do not die with 'no handler for signal XX in
219 interrupt_handle_now(..)' anymore
220 ** bug fix: WITH-TIMEOUT works with multiple running threads
221 * fixed some bugs revealed by Paul Dietz' test suite:
222 ** ENSURE-GENERIC-FUNCTION should take a method class object for
223 the :method-class keyword argument.
225 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
226 * new port: the Solaris operating system on x86 processors is now
227 mostly supported, though some rough edges in the environment
228 remain. (thanks to Daisuke Homma)
229 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
230 space for its heap. (thanks to Thiemo Seufer)
231 * optimization: SBCL on MIPS platforms now supports dynamic-extent
232 closures. (thanks to Thiemo Seufer)
233 * minor incompatible change: eof selects abort in the debugger.
234 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
235 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
237 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
238 deprecated, and will go away in a future revision of SBCL.
239 * minor incompatible change: GC-ON and GC-OFF are no longer
240 implemented with a counter, it does not matter how many times gc
241 is switched on or off
242 * bug fix: discriminating functions for generic function classes
243 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
244 make invalid assumptions about method precedence order. (reported
246 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
247 platforms. (thanks to Thiemo Seufer)
248 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
249 * bug fix: correct pathname printing with printer escaping is on.
250 (thanks to Kevin Reid)
251 * bug fix: complex VOP definitions in "user-space" no longer trigger
252 package locks. (reported by Zach Beane)
253 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
254 for funcallable-instance functions, and is no different from
256 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
257 characters in the current readtable. (reported by Nicholas Neuss)
258 * bug fix: SUBTYPEP on various CONS types returns more a more
259 accurate acknowledgment of its certainty.
260 * optimizations: REMOVE-DUPLICATES now runs in linear time on
261 lists in some cases. This partially fixes bug 384.
262 * flush all standard streams before prompting in the REPL and the
264 * bug fix: signal handling and triggering gc do not conflict
265 directly anymore, in particular a high frequency sb-sprof does
266 not prevent gc from running
267 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
268 approximation for timezone and DST information between the
269 universal time epoch and the smallest negative 32-bit time_t.
270 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
271 year 1899 so long as the encoded time is non-negative.
272 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
273 the correct answer for non-constant rotate amounts when performing
274 an inline 32-bit rotation.
276 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
277 there is only one thread in the session
278 ** bug fix: memory leak for streams created in one thread and
279 written to in another
280 ** bug fix: lockup when compiled with gcc4
281 ** bug fix: race that allows the gc to be triggered when gc is
283 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
284 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
286 ** bug fix: finished threads are gc'ed properly
287 * fixed some bugs revealed by Paul Dietz' test suite:
288 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
289 has arguments with a different set of applicable methods from
290 the orignal arguments.
291 ** The type error thrown by MAP now has a correct expected-type
293 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
294 name a compiled function.
295 ** Three MISC tests where a large bignum was improperly coerced to
296 a float that couldn't represent that bignum during type
297 derivation were fixed.
298 ** SUBTYPEP can now handle types involving the negation of a
299 list-form FUNCTION type.
300 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
301 as (COMPLEX (AND RATIO (NOT FIXNUM))).
302 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
304 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
305 * New feature: Experimental support for bivalent streams: streams
306 opened with :element-type :default now allow character and binary
307 (unsigned-byte 8) I/O
308 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
309 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
311 * bug fix: no more highly sporadic "couldn't check whether ~S is
312 readable" when reading a stream and an interrupt hits in the middle
313 of a select system call
314 * compiler better recognizes complex arrays (reported by Eduardo
316 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
317 for variables on SBCL built with linkage-tables. (reported by Luis
319 * various error reporting improvements.
320 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
321 (Thanks to Thiemo Seufer)
322 * bug fix: nested reader invokations work correctly
323 * bug fix: it is possible to have simultaneous references to foreign
324 code and foreign data with the same name.
326 ** added x86-64 support
327 ** incompatible change: the threading api now works with thread
328 objects instead of thread ids
329 ** bug fix: threads are protected from signals and interruption when
330 starting up or going down
331 ** bug fix: a race where an exiting thread could lose its stack to gc
332 ** bug fix: don't halt on infinite error in threads if possible
333 ** fixed numerous gc deadlocks introduced in the pthread merge
334 ** bug fix: fixed thread safety issues in read and print
335 ** bug fix: debugger doesn't hang on session lock if interrupted at
336 an inappropriate moment
337 ** bug fix: run-program is now thread safe(r)
338 ** bug fix: inner with-recursive-lock no longer releases the mutex
339 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
340 * fixed some bugs revealed by Paul Dietz' test suite:
341 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
342 EXPECTED-TYPE slots filled.
343 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
345 ** MAKE-CONDITION accepts classes as type-designators.
346 ** COMPILE may never return NIL.
347 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
348 range before calling Unix time functions
350 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
351 * numerous signal handling fixes to increase stability
352 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
353 * minor incompatible change: we now correctly canonize default
354 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
355 as per the MOP, rather than the historical (INITARG INITFUNCTION
356 INITFORM). (reported by Bruno Haible)
357 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
358 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
359 for more information.
360 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
361 pathname is a directory pathname.
362 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
363 * dynamic space size on PPC has been increased to 768Mb. (thanks to
365 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
366 :METACLASS argument in addition to a class name. (reported by
367 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
368 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
369 (reported by Bruno Haible; patch from Gerd Moellmann)
370 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
372 * bug fix: more cleanups to the floating point exception handling on
373 x86-64 (thanks to James Knight)
374 * bug fix: the compiler does not try to constant fold calls of
375 COERCE and BIT-* functions when they return freshly constructed
376 objects. (reported by Nikodemus Siivola and Paul Dietz)
377 * optimization: improved the MIPS versions of generic (in the
378 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
379 * optimization: direct conversion of (unsigned-byte 32) to floats on
380 the PowerPC platform.
381 * optimization: structure instances with raw slots now use less
382 memory, and probably show better memory locality. (thanks to
384 * optimization: DYNAMIC-EXTENT declarations for lists and closures
385 are treated as requests for stack allocation on the x86-64,
386 Alpha-32, PPC and SPARC platforms.
387 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
388 the wrong file descriptor; implementation of SOCKET-OPEN-P.
389 (thanks to Tony Martinez)
391 ** gcing a dead thread can no longer lead to lockups
392 ** threads block signals until they are set up properly
393 ** errno is no longer shared by threads
394 ** interrupt-thread restores the eflags register on x86
395 ** fixed some lockups due to gc/thread interaction
396 * fixed some bugs revealed by Paul Dietz' test suite:
397 ** invalid dotted lists no longer raise a read error when
399 ** signal an error if a symbol that names a declaration is used
400 as the name of a type, or vice versa
401 ** allow using the (declare (typespec var*)) abbreviation for
402 (declare (type typespec var*)) with all type specifiers
403 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
404 ** always return NIL from PPRINT-POP when OBJECT is NIL
405 ** don't signal errors when pretty-printing malformed LABELS,
406 FLET or MACROLET forms
407 ** declarations in a DOLIST body are also in scope for the
409 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
411 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
414 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
415 * fixed cross-compiler leakages that prevented building a 32-bit
416 target with a 64-bit host compiler.
417 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
418 opened with :IF-EXISTS :OVERWRITE.
419 * fixed bug 281: error for an invalid qualifier in a short-form method
420 combination method is not signalled until the faulty method is called.
421 * bug fix: iteration variable type inferrer failed to deal with open
422 intervals. (reported by Alan Shields)
423 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
424 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
425 or more runtime options were provided to the sbcl binary.
426 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
427 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
428 * optimization: there's now a fast-path for fixnum arguments in the
429 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
430 * optimization: the code generated on x86-64 is more compact thanks
431 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
432 * bug fix: floating-point exception handling now partly works on
433 x86-64. (Thanks to James Knight)
434 * improvement to the MIPS backend from Thiemo Seufer: C-style
435 64-bit long long arguments and return values to alien functions
437 * fixed some bugs revealed by Paul Dietz' test suite:
438 ** the type-error signalled from WARN has a filled-in DATUM slot.
439 ** the type-error required when a stream is not associated with
440 a file has the stream as its datum.
441 ** type-errors on single-floats on x86-64 no longer have
442 :INVALID-OBJECT as the datum
443 ** the type-errors signalled for invalid function names now have
444 a correct expected type
445 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
446 for structures defined with a :TYPE. Documentation strings for
447 typed structures are no longer immediately discarded
448 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
449 broadcast streams and synonym streams. FILE-LENGTH now also works
450 on broadcast streams.
452 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
453 * incompatible change: the --noprogrammer option, deprecated since
454 version 0.7.5, has been removed. Please use the equivalent
455 --disable-debugger option instead.
456 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
458 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
459 inoperational for a while now) has been completely removed.
460 * null lexical environments are now printed as #<NULL-LEXENV>,
461 significantly reducing the amount of clutter in typical backtraces.
462 * documentation on weak pointers, finalization, and after GC hooks
463 has been added to the manual.
464 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
465 as other specialized array types, is much faster. SUBSEQ and
466 COPY-SEQ on such arrays have also been sped up.
467 * optimization: EQL is now more efficient when at least other argument
468 is known to be of type (OR FIXNUM (NOT NUMBER)).
469 * fixed bug: the runtime is now less vulnerable to changes in the
470 size of the SBCL object on OS X, and virtual memory is reserved for
471 all spaces that need to be at a fixed address.
472 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
473 * fixed bug: finalizers and after GC hooks that cause consing are now
475 * fixed bug: compiler error messages and summaries are now printed to
476 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
477 * fixed inference of the upper bound of an iteration variable.
478 (reported by Rajat Datta).
479 * fixed bug 211e: calling local functions with duplicated constant
480 keyword argument no longer causes a bogus style warning about an
482 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
483 ftype declaration to be disregarded. (reported by Dave Roberts)
484 * fixed bug 373: caused by erronous compilation of references to alien
485 variables in the runtime on ppc/darwin.
486 * fixed bug 376: CONJUGATE type deriver.
487 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
488 (reported by Baughn on #lisp)
489 * a cross-compiler bug on non-x86 platforms has been identified and
490 fixed. (thanks to Bruno Haible)
491 * improvements to the MIPS runtime code for increased stability.
492 (thanks to Thiemo Seufer)
493 * increased the maximimum compact environment size to allow
494 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
495 * partial workaround for bug 135: don't name the function we're
496 calling for hairy cases of EVAL, so as not to accumulate one environment
497 entry per call to EVAL. (reported by Kevin Reid)
498 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
499 * optimization: type testing for non-vector arrays should be faster.
500 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
502 * bug fix: setting 31st element of a bit vector to zero did not work
504 * bug fix: redefining a class definition which failed due to a
505 previous accessor / function clash now works (but see BUGS entry
506 #380 for more problems in this area). (thanks to Zach Beane)
507 * the long form of DEFINE-METHOD-COMBINATION disables method group
508 checking when given a single method group with pattern *. (thanks
510 * on x86 compiler supports stack allocation of results of simple
511 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
512 * support for building and running on Mac OS X 10.4 "Tiger" has been added
513 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
514 * fixed some bugs related to Unicode integration:
515 ** the restarts for recovering from input and output encoding
516 errors only appear when there is in fact such an error to
518 * fixed some bugs revealed by Paul Dietz' test suite:
519 ** MISC.549 and similar: late transformation of unsafe type
520 assertions into derived types caused unexpected code
522 ** SCALE-FLOAT type deriver is less wrong.
523 ** type derivers for EXP, LOG and similar functions compute result
524 types for complex arguments better.
525 ** (MISC.563) CONJUGATE type deriver works for very restricted
527 ** out-of-line type testers for character strings are available.
528 ** EQUAL compiler transform understands specialness of objects
530 ** accessing double-floats stored on the stack now works on x86-64.
531 ** debugger internals could sometimes create invalid lispobjs,
532 resulting in GC crashes.
533 ** MISC.548: type check weakening can convert required type into
535 ** initialization forms for bindings are not in scope of free special
538 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
539 * incompatible change: thread support for non-NPTL systems has
540 been removed - locking is buggy and unreliable. A threaded
541 SBCL build will now warn at startup and refuse to create
542 new threads, unless futex support is detected in the kernel
543 * incompatible change: the top level REPL now has only an ABORT
544 restart associated with it, not TOPLEVEL and ABORT as it used to.
545 TOP and TOPLEVEL are now available as debugger commands for
546 returning to the top level.
547 * incompatible change: forms evaluated in the REPL now use the
548 global optimization policy.
549 * incompatible change: user- and system-initialization files are
550 no longer processed with LOAD, but by READ and EVAL; hence the
551 global optimization policy, startup package, readtable, etc,
553 * The .fasl file format number has been incremented because of
554 various incompatible changes.
555 * internal entry point details and argument counts no longer appear
556 in backtraces unless explicitly requested by setting
557 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
558 * built-in and standard functions no longer have names like "top
559 level local call to FOO".
560 * fixed bug 32: functions defined in non-null lexical environments
561 now have more legible printed representation
562 * fixed bug 33: functions defined in non-null lexical environemnts
563 are now more amenable to inspection by INSPECT.
564 * workaround for bug 354: XEPs no longer appear in backtraces unless
565 explicitly requested.
566 * fixed bug: receiving the signal which results from attempting to
567 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
568 SIGBUS on other *BSDs) on architectures where the C stack is also
569 the Lisp stack (x86 and x86-64) from foreign code no longer leads
570 to debugger confusion or wild instability. (reported by Cheuksan
572 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
573 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
574 (reported by Lutz Euler)
575 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
576 compiled code. (reported by Paul Dietz)
577 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
578 specialized parameter to a method to something that is not TYPEP
579 the specializer is now possible.
580 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
581 face of package deletion.
582 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
583 pathname. (reported by David Tolpin on c.l.l)
584 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
585 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
586 STRING-STREAM and FILE-STREAM, respectively.
587 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
588 than previously; in addition, its test suite is now run on build.
589 (thanks to Luke Gorrie)
590 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
591 * optimization: added a immediate representation for single-floats
593 * fixed some bugs revealed by Paul Dietz' test suite:
594 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
595 ** attempting to create a package with a colliding nickname causes
596 correctable errors to be signalled.
597 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
598 ** several x86-64 backend bugs related to sign-extension of immediate
601 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
602 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
603 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
605 * fixed loading of multiply forward-referenced layouts.
606 (thanks to Cheuksan Wang)
607 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
608 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
609 (see COMPILE-FILE documentation for details of :PRINT option)
610 * fixed bugs 19 and 317: fixed-format floating point printing is
611 more accurate. This also fixes a bug reported by Adam Warner
612 related to the ~@F format directive.
613 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
615 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
616 dispatch table if the from-char is a dispatch macro character.
617 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
618 by the wordsize no longer ignore the last word. (reported by Lutz
620 * fixed bug in type checking of dynamic-extent variables. (reported
622 * optimization: sequence traversal functions use their freedom to
623 coerce function designators to functions.
624 * optimization: code with many calls to CLOS methods specialized on
625 CLOS classes has had redundant type checks removed.
626 * optimization: added declarations to speed up operations that access
627 the internal character database (for example STRING-UPCASE)
628 * optimization: comparison operations between floats and sufficiently small
629 fixnums no longer create extra rationals
630 * fixed some bugs related to Unicode integration:
631 ** portions of multibyte characters at the end of buffers for
632 character-based file input are correctly transferred to the
633 start of the buffer at the next read.
634 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
635 passing it through to OPEN.
636 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
637 argument given to internal calls to OPEN.
638 ** the built-in comment readers (introduced by character sequences
639 ";" and "#|") are more forgiving to encoding errors; they will
640 STYLE-WARN and then attempt to resync the stream at a character
641 boundary. (thanks to Teemu Kalvas)
642 * fixed some bugs in the x86-64 port:
643 ** Negative short int return values from c-calls are sign-extended
645 ** The stack is aligned to 16-bytes for c-calls, as required by
646 the ABI. (thanks to Cheuksan Wang)
647 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
648 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
649 for enums. (thanks to Vincent Arkesteijn)
650 ** Multiple small optimizations and bugfixes for floating point
652 * fixed some bugs revealed by Paul Dietz' test suite:
653 ** Space, Tab, Linefeed, Return and Page have the invalid
654 secondary constituent character trait.
655 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
657 ** WITH-INPUT-FROM-STRING should only update the index place on
659 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
661 ** Bit-array operations (BIT-AND and similar) worked incorrectly
662 with one-dimensional arrays with fill pointers.
663 ** TYPE-OF failed on a complex with an integer realpart and a
665 ** compiler failure during type inference for the code of form
666 (IF (EQL X (THE ...)) ...) (MISC.535).
668 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
669 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
670 platform. The port supports 61-bit fixnums, large memory spaces
671 and reloading shared object files.
672 * enhancement: saving cores with foreign code loaded is now
673 supported on ppc/Darwin in addition to the previously supported
675 * enhancement: the statistical profiler now walks deeper into the
676 call stack for more meaningful call-graphs and accrued time
677 reports (x86/x86-64 only). It also now reports time spent in
679 * enhancement: it is now possible to trace most individual methods
680 of a generic function in addition to tracing the generic function
682 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
684 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
685 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
686 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
687 produces an error. (thanks to Vincent Arkesteijn)
688 * bug fix: NAMESTRING on pathnames with :WILD components in their
689 directories works correctly. (thanks to Artem V. Andreev)
690 * fixed bug 125: compiler preserves identity of closures. (reported
692 * bug fixed: functions with &REST arguments sometimes failed with
693 "Undefined function" when compiled with (DEBUG 3). (reported by
695 * bug fix: overflow during compiling of setting element of a bit
696 vector with constant index and value. (reported by Timmy Douglas)
697 * build fix: fixed the dependence on *LOAD-PATHNAME* and
698 *COMPILE-FILE-PATHNAME* being absolute pathnames.
699 * on x86 compiler partially supports stack allocation of dynamic-extent
701 * GO and RETURN-FROM do not check the extent of their exit points
702 when compiled with SAFETY 0.
703 * fixed some bugs related to Unicode integration:
704 ** encoding and decoding errors are now much more robustly
705 handled; it should now be possible to recover even from invalid
706 input or output to the terminal. (thanks to Teemu Kalvas)
707 ** provided a first cut at implementing STRING-TO-OCTETS and
708 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
709 ** altered the SB-MD5 contributed module slightly, changing the
710 interface just enough for it to be supportable for builds where
711 lisp characters are not eight bits.
712 * fixed some bugs revealed by Paul Dietz' test suite:
713 ** the FORMATTER-generated functions for ~V[ conditionals require
714 the correct number of arguments.
715 ** READ-FROM-STRING returns the mandated second value when applied
716 to displaced strings.
717 ** the #\Rubout and #\Backspace characters are treated as invalid
718 constituent characters by the tokenizer.
720 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
721 * new feature: reloading changed shared object files with
722 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
723 * new feature: references to foreign variables and functions
724 can now be compiled and loaded before the corresponding shared
725 object file is loaded, as long as the foreign definitions are
726 available at runtime.
727 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
728 supported. (thanks to Dan Debertin)
729 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
730 just keywords. (thanks to Vincent Arkesteijn)
731 * fixed bug #331: structure-class instances corresponding to
732 DEFSTRUCT forms are now created eagerly.
733 * fixed bug #345: backtraces from calls to undefined functions work
734 on x86 as well. Related bug #61 is now also partially fixed on x86
735 (backtraces from throws to unknown catch tags.)
736 * bug fix: lambda-list parsing is now stricter vrt. order and number
737 of lambda-list keywords.
738 * bug fix: as specified by AMOP, an error is signalled if a
739 class-option appears multiple times in a DEFCLASS form. (reported
741 * bug fix: RANDOM can be compiled when the compiler derives the type
742 of its numeric argument as a disjoint set of small integers.
743 (reported by Paul Dietz)
744 * bug fix: starting a core saved with shared objects loaded when
745 those objects are not available no longer causes threaded SBCL to
746 hang. (reported by Sean Ross)
747 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
749 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
750 argument, which must always be NIL. (reported by Kalle Niemitalo)
751 * bug fix: printing 1.0d+23 no longer results in an error.
752 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
753 * bug fix: structure slot setters preserve evaluation order. (thanks
755 * fixed some bugs related to Unicode integration:
756 ** RUN-PROGRAM can allow its child to take input from a Lisp
757 stream. (reported by Stefan Scholl)
758 ** ASDF-INSTALL successfully downloads .tar.gz files in all
759 locales. (reported by Ken Causey)
760 * fixed some bugs revealed by Paul Dietz' test suite:
761 ** INCF, DECF and REMF evaluate their place form as specified in
763 ** FORMATTER expands ~{ ~} iteration directives with V or #
764 parameters correctly.
765 ** FORMATTER deals with the ~@[ ~] conditional directive where the
766 consequent uses no arguments correctly.
767 ** the system has a partial understanding of the (COMPLEX RATIO)
769 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
772 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
773 * new feature: a build-time option (controlled by the :SB-UNICODE
774 keyword feature, enabled by default) for building the system with
775 support for the entire 21-bit character space defined by the
777 * new feature: the system now has rudimentary external-format
778 support; the primary user-visible change at this time is that
779 characters with the high bit set (such as lower-case-e-acute) will
780 print correctly to a terminal in a UTF-8 environment.
781 * minor incompatible change: BASE-CHAR no longer names a class;
782 however, CHARACTER continues to do so, as required by ANSI.
783 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
784 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
785 has been moved to the SB-EXT package (temporarily re-exported from
787 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
788 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
790 * fixed bug #351: better error-handlind and reporting for malformed
792 * fixed bug #350: bignum-printing is now more memory-efficient,
793 allowing printing of very large bignums, eg. (expt 2 10000000).
794 (reported by Bruno Haible)
795 * fixed bug #302: better primitive-type selection for intersection
797 * fixed bug #308: non-graphic characters now all have names, as
798 required. (reported by Bruno Haible)
799 * bug fix: redefining a class with different superclasses now correctly
800 removes it from the direct-subclasses of its previous superclasses.
801 (reported by David Morse)
802 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
803 new value now works. (reported by Bruno Haible)
804 * bug fix: correct canonicalization of multiple non-standard slot
805 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
806 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
807 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
808 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
809 now exists, an signals an error.
810 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
811 during macroexpansion for non-iteration variables as well. (reported
812 by Bruno Haible for CMUCL)
813 * bug fix: Cyclic structures and unprintable objects in compiler
814 messages no longer cause errors. (reported by Bruno Haible)
815 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
816 and READ-FROM-STRING all now return a primary value of NIL if
817 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
818 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
819 (reported by Bruno Haible for CMUCL)
820 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
821 do not designate a condition. (reported by Bruno Haible for
823 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
824 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
825 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
826 permitting resolution in favour of any of the conflicting symbols.
827 (reported by Bruno Haible for CMUCL)
828 * bug fix: EQUAL compiler optimizations is less aggressive on
829 strings which can potentially compare true despite having distinct
830 specialized array element types.
831 * bug fix: unit enumerations can be defined without dividing by
832 zero. (thanks to Vincent Arkesteijn)
833 * FORMAT compile-time argument count checking has been enhanced.
834 (report from Bruno Haible for CMUCL)
835 * a partial workaround for the bug 262: the compiler does not try to
836 inline-expand a local function doing RETURN-FROM from a deleted
837 BLOCK. (thanks to Peter Denno for the bug report and to David
838 Wragg for the simple test case)
839 * fixed some bugs revealed by Paul Dietz' test suite:
840 ** RENAME-PACKAGE allows all package designators as new package
842 ** constraint propagation and lambda variable substitution are
843 more cautious in dealing with partially deleted code.
844 ** compiler sometimes lost reoptimization passes.
845 ** CERROR, given a condition as condition designator, passes its
846 remaining arguments to the continue format control without
848 ** Case-altering FORMAT directives work correctly on non-ASCII
850 ** The REAL type specifier handles bounds outside the floating
851 point ranges without signalling FLOATING-POINT-OVERFLOW.
852 ** Functions with IR1-transformations can create intercomponent
853 references to global functions.
854 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
856 ** FORMAT ~R treats a nil value for its first parameter correctly.
858 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
859 * enhancement: saving cores with foreign code loaded is now
860 supported on x86/NetBSD and sparc/Linux in addition to the previously
862 * bug fix: on some platforms repeated installations caused multiple
863 copies of HTML documentation to be installed -- should not happen
864 any more. (reported by Stefan Scholl)
865 * bug fix: parsing self-recursive alien record types multiple times
866 no longer causes infinite recursion. (reported by Thomas F. Burdick,
867 original patch by Helmut Eller for CMUCL)
868 * bug fix: stack-exhaustion detection works now on NetBSD as well.
869 (thanks to Richard Kreuter)
870 * bug fix: defining classes whose accessors are methods on existing
871 generic functions in other (locked) packages no longer signals
872 bogus package lock violations. (reported by François-René Rideau)
873 * bug fix: special variables as DEFMETHOD parameters no longer have
874 associated bogus type declarations. (reported by David Wragg and
876 * bug fix: read-write consistency on streams of element-type
877 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
878 * bug fix: redefiniton of the only method of a generic function with
879 no DEFGENERIC no longer emits a full WARNING. In addition,
880 redefinition of generic functions with no DEFGENERIC to an
881 incompatible lambda list now signals an error. (thanks to Zach
883 * bug fix: DEFGENERIC now works even when there's a function of the
884 same name in an enclosing lexical environment. (thanks to Zach
886 * fixed compiler failure, caused by instrumenting code during
887 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
888 * optimization: added loop analysis and improved register allocation
889 * fixed some bugs revealed by Paul Dietz' test suite:
890 ** POSITION on displaced vectors with non-zero displacement
891 returns the right answer.
892 ** (SIMPLE-STRING) is a valid type specifier for sequence
894 ** *PRINT-LEVEL* handling for slotless structures is pedantically
896 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
897 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
899 ** The justification version of the FORMAT ~< directive treats
900 non-zero minpad parameter correctly.
902 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
903 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
904 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
905 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
906 the supported interface.
907 * new feature: Single-stepping of code compiled with DEBUG 2 or
908 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
909 * new feature: saving cores with foreign code loaded is now
910 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
911 Timothy Moore's work for CMUCL)
912 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
913 parameters to * instead of NIL if no initform is supplied.
914 (reported by Johan Bockgaard)
915 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
916 correctly when FUNCALL appears as the car of the form. Note:
917 despite this FUNCALL forms are not currently subject to
918 compiler-macro expansion. (port of Raymond Toy's fix for the
919 same from CMUCL, reported by Johan Bockgaard)
920 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
921 (thanks for Teemu Kalvas)
922 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
923 case instead of terminating the iteration (thanks for Julian
924 Squires, Sean Champ and Raymond Toy)
925 * bug fix: incorrect expansion of defgeneric that caused a style
926 warning. (thanks for Zach Beane)
927 * bug fix: slot accessor effective method computation works properly
928 for classes with multiple non-standard applicable methods on
929 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
930 * on x86 compiler supports stack allocation of results of LIST and
931 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
932 CMUCL implementation by Gerd Moellmann)
933 * fixed some bugs revealed by Paul Dietz' test suite:
934 ** FORMAT strings with both the ~<~:;~> form of the justification
935 directive and pretty-printing directives cause an error.
937 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
938 * incompatible change: the internal functions
939 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
940 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
941 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
942 instead of the old functions.
943 * new feature: on platforms where "dladdr" is available foreign
944 function names now appear in backtraces. (based on Helmut Eller's
946 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
947 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
949 * documentation: ASDF manual is now distributed with SBCL.
950 * bug fix: SBCL can now load its contributed modules with REQUIRE
951 even if the system-provided entries have been removed from
952 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
954 * bug fix: backtraces involving undefined functions or assembly
955 routines are more informative. (thanks to Brian Downing)
956 * bug fix: mutually referent alien structures now work correctly.
957 (reported by Rick Taube)
958 * bug fix: structures defined by WITH-ALIEN can be referred to
959 within other definitions in the same WITH-ALIEN.
960 * bug fix: division operators (MOD, TRUNCATE and the like) with
961 constant zero divisors and integer dividends no longer generate
963 * bug fix: provide default methods for INPUT-STREAM-P and
964 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
965 * bug fix: improve the detection and resolution of MOP metacycles
966 (where effective method computation depends on the generic function
967 having its effective method computed). (reported by Bruno Haible)
968 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
969 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
970 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
971 * bug fix: pathnames with relative directory components can be
972 represented relative to default pathnames.
973 * optimization: in taking the GCD of bignums, reduce the two bignums
974 to approximately the same size (using Euclid's algorithm) before
975 applying the more sophisticated binary GCD. (thanks to Juho
977 * optimization: COUNT on bitvectors now operates word-at-a-time.
978 * optimization: ASH with a positive, but not necessarily constant,
979 (leftwards) shift, when in a modular context, is compiled to a
981 * fixed some bugs revealed by Paul Dietz' test suite:
982 ** FORMAT variable parameters ("~V<char>") are defaulted properly
983 if the corresponding argument is NIL.
984 ** FORMAT directives accepting a minpad parameter treat negative
986 ** literal commas as character parameters to FORMAT directives are
988 ** literal spaces directly after ~<Newline> directives within a
989 format-logical-block (~:< ~@:>) do not induce :FILL-style
990 conditional newlines.
991 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
992 as specified: it no longer includes conditional newlines).
993 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
995 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
996 :PER-LINE-PREFIX argument does not evaluate to a string.
998 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
999 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1000 the manual for details; for now, package locks can be disabled by
1001 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1002 if no major problems are found then it is likely that they will be
1003 compiled in unconditionally.
1004 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1005 unsupported operators on all platforms. To load a shared library
1006 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1007 object file, link it into a shared library outside of SBCL and
1008 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1010 * bug workaround: ROOM T and the GC were not working together
1011 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1012 reporting facility aren't preserved. That reporting has been
1013 disabled, so now until and unless someone figures out how to make it
1014 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1015 * minor incompatible change: as threatened around sbcl-0.8.0, the
1016 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1017 The symbols are also exported from SB-PCL for backwards
1018 compatibility, but more so than before SB-PCL should be treated as
1019 an implementation-internal package.
1020 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1022 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1023 It is known as of this release not to work on the Alpha, however.
1024 * fixed bug #167: errors signalled due to illegal syntax in method
1025 bodies are now more legible.
1026 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1027 designators and can hence be used with TYPEP.
1028 * fixed bug #333: CHECK-TYPE now ensures that the type error
1029 signalled, if any, has the right object to be accessed by
1030 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1031 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1032 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1034 * fixed bug #334: programmatic addition of slots using specialized
1035 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1036 and :ALLOCATION :CLASS slots.
1037 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1038 float exponents. (rereported by Peter Seibel)
1039 * fixed a bug: #\Space (and other whitespace characters) are no
1040 longer considered to be macro characters in standard syntax by
1041 GET-MACRO-CHARACTER.
1042 * fixed bug: initialization of condition class metaobjects no longer
1043 causes an instance of the condition to be created. (reported by
1045 * fixed bug: it is now possible to have slots such that
1046 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1047 description is neither :INSTANCE nor :CLASS.
1048 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1049 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1050 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1051 non-standard SLOT-DEFINITION-ALLOCATION values.
1052 * fixed bugs: various race conditions handling exiting threads.
1053 CL-PPCRE's thread torture test now passes.
1054 * fixed arguably-a-bug: GC time proportional to number of threads in
1055 system even when most of them are idle
1056 * optimization: improved performance of BIT and SBIT on bit-vectors.
1057 * .fasl file incompatibility: The fasl file version number has
1058 been incremented because of changes associated with package locks.
1060 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1061 * minor incompatible change: the system no longer provides
1062 optimization hints (conditions which are TYPEP
1063 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1064 mode; these hints are still emitted when higher SPEED optimization
1066 * new contrib module: a sampling profiler (profiling by statistical
1067 sampling, rather than by instrumenting functions) is available as
1068 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1069 * the behaviour of the standard function ED is now customizeable by
1070 third parties through a hook variable: see ED's documentation
1071 string for information on the protocol.
1072 * the compiler no longer emits efficiency notes for (FUNCALL X)
1073 when the type of X is uncertain under default optimization
1075 * fixed bug 276: mutating a binding of a specialized parameter to a
1076 method to something that is not TYPEP the specializer is now
1078 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1079 exhibit the required behaviour on the x86 platform. (thanks to
1080 Peter van Eynde, Eric Marsden and Bruno Haible)
1081 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1082 even for difficult arguments. (reported by Peter Graves)
1083 * fixed bug 141a: the backquote printer now descends quoted
1085 * fixed another bug in backquote printing: no more destructive
1086 modification of the form's list structure. (reported by Brian
1088 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1089 move between its address being taken and the call to
1090 interrupt_thread, fixing a crashing race condition.
1091 * the SB-POSIX contrib implementation has been adjusted so that it
1092 no longer exhibits ridiculously poor performance when constructing
1093 instances corresponding to C structs.
1095 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1096 * minor incompatible change: the sb-grovel contrib now treats C
1097 structures as alien (in the sense of SB-ALIEN) objects rather than
1098 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1099 has implications for memory management of client code
1100 (sb-grovel-returned objects must now be manually managed) and for
1101 type safety (alien objects now have full types).
1102 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1103 used to control emission of compiler diagnostics, rather than the
1104 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1105 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1106 quality should be considered deprecated.
1107 * install.sh now installs the user manual as well
1108 * (not quite a new documentable feature, but worth considering in
1109 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1110 beginnings of a semantically meaningful condition hierarchy is
1111 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1113 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1114 designator as the defaults argument.
1115 * fixed bug: Displaced arrays whose displaced-to array has become
1116 too small now cause ARRAY-DIMENSION to signal an error, providing
1117 for safer bounds-checking. (reported by Bruno Haible)
1118 * fixed bug: DEFCLASS slot definitions with identical :READER and
1119 :WRITER names now signal a reasonable error. (reported by Thomas
1121 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1123 * fixed bug: Invalid slot specification errors now print correctly.
1124 (thanks to Zach Beane)
1125 * fixed bug 320: Shared to local slot value transfers in class
1126 redefinitions now happen corrently from superclasses as
1127 well. (reported by Bruno Haible)
1128 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1130 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1131 improper lists correctly. (reported by Bruno Haible)
1132 * fixed bug 313: source-transform for <fun-name> was erroneously
1133 applied to a call of a value of a variable with name <fun-name>.
1134 (reported by Antonio Menezes Leitao)
1135 * fixed bug 307: The obsolete instance protocol ensures that
1136 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1137 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1138 remove dead unknown-values globs from the middle of the stack.
1139 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1140 easier to resume long computations after using *BREAK-ON-SIGNALS*
1141 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1142 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1143 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1144 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1145 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1146 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1148 * fixed a (fairly theoretical) bug in string printing: if
1149 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1150 does not have array-element-type equal to the most general string
1152 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1153 function name. (thanks to Nikodemus Siivola)
1154 * fixed bug: compile-time format string checker failed on
1155 non-closed ~{. (reported by Thomas F Burdick)
1156 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1157 don't warn on legal loop code involving more than one aggregate
1158 boolean. (thanks to Nikodemus Siivola)
1159 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1160 in keyword position to MAKE-PACKAGE was not being reported as
1162 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1163 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1165 * fixed bugs 280 and 312: the checking for multiple definitions in a
1166 file is less likely to become confused by uses of inline
1168 * fixed bug: the #S reader macro performs the keyword coercion
1169 specified for slot names. (reported by Kalle Niemitalo)
1170 * fixed bug: lambda lists may contain symbols whose names start with
1171 & but are not lambda-list-keywords; their occurrence triggers a
1173 * fixed bug 321: define-method-combination argument lambda lists do
1174 not cause invalid code to be generated when &optional or &aux
1175 variables have default values. (reported by Bruno Haible)
1176 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1177 have CLOS classes; this appears also to have fixed reported
1178 instability in stack exhaustion detection.
1179 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1180 MUFFLE-WARNING when no associated restart is present is now
1182 * optimization: rearranged the expansion of various defining macros
1183 so that each expands into only one top-level form in a
1184 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1186 * optimization: used a previously little-used slot in symbols to
1187 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1189 * fixed some bugs revealed by Paul Dietz' test suite:
1190 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1192 ** Fixed an optimization bug related to inheritance of initforms
1193 from local to shared slots.
1194 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1195 ** CAST optimizer forgot to flush argument derived type.
1196 ** print/read consistency on floats is now orders of magnitude
1197 more likely. (thanks also to Bruno Haible for a similar report
1199 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1200 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1201 for objects of type REAL. Make it so.
1202 ** SXHASH is specified (infelicitously) to respect similarity,
1203 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1204 it so. (thanks to Markus Ziegler)
1205 ** on the Alpha, the compiler succeeds in compiling functions
1206 returning a known number of arguments greater than 63.
1207 ** fixed handling of invalid NIL arguments in keyword position
1209 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1211 ** provided a readably-printable representation for RANDOM-STATE
1213 ** ensured that pathnames, where they have a namestring, always
1214 print using #P"..." syntax.
1216 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1217 * Support for the forthcoming 2.0 version of the NetBSD kernel
1218 running on x86 hardware has been added. (thanks to Perry
1219 E. Metzger most immediately, and others for their past work)
1220 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1221 current release version; SBCL's OpenBSD support had been broken
1222 since about the time of OpenBSD's switch to ELF binary format.)
1223 * [placeholder for DX summary]
1224 ** user code with &REST lists declared dynamic-extent, under high
1225 speed or space and low safety and debug optimization policy.
1226 * The manual has been converted to Texinfo format and the debugger
1227 chapter from the cmucl manual has been added.
1228 * A facility has been added to extract documentation strings from
1229 sbcl and store them as Texinfo-formatted snippets for inclusion in
1230 the manual (via Texinfo's @include directive)
1231 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1232 the test case to Dave Roberts)
1233 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1234 pointers. (reported by Sean Ross)
1235 * bug fix: PROFILE output is printed nicely even for large numerical
1236 values. (thanks to Zach Beane)
1237 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1238 greater than 32 handle EOF correctly.
1239 * bug fix: on X86 an immediate argument of the IMUL instruction is
1240 correctly printed in disassembly. (thanks to Lutz Euler)
1241 * bug fix: class slots in redefined classes preserve their old
1242 values. (thanks to Bruno Haible and Nikodemus Siivola)
1243 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1244 succeeds. (reported by Marco Baringer)
1245 * fixed some bugs revealed by Paul Dietz' test suite:
1246 ** READ-SEQUENCE now works on ECHO-STREAMs.
1247 ** RATIONALIZE works more according to its specification. (thanks
1250 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1251 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1252 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1253 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1254 should matter to you only if you rebind the printer control
1255 variables and then find you want different bindings in the
1256 debugger than in the ordinary execution of your program.)
1257 * The runtime build system has been tweaked to support building
1258 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1259 assembler and linker. (thanks to Nikodemus Siivola)
1260 * Unbound, undefined, undeclared variables now trigger full
1261 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1262 more often programmer error than deliberate exploitation of undefined
1264 * optimization: the hash algorithm for strings has changed to one
1265 that is less vulnerable to spurious collisions. (thanks to Juho
1267 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1268 needless bounds checking. (thanks to Juho Snellman)
1269 * optimization: implemented multiplication as a modular
1270 (UNSIGNED-BYTE 32) operation on the PPC backend.
1271 * fixed some bugs revealed by Paul Dietz' test suite:
1272 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1274 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1275 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1276 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1278 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1280 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1282 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1283 their output stream on EOF from read.
1284 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1285 have been read to end-of-file.
1286 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1288 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1289 description of determination of which consecutive characters
1291 ** Printing the "Space" character with escaping on now yields "#\\ ",
1292 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1293 ** Reading floating-point numbers with *READ-BASE* set to a number
1294 less than 10 works correctly.
1295 ** Reading floating-point numbers with *READ-BASE* set to a number
1296 more than 10 works correctly.
1297 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1298 the readtable currently in effect.
1300 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1301 * minor incompatible change: parsing of namestrings on a physical
1302 (Unix) host has changed; numbers after the final #\. in a
1303 namestring are no longer interpreted as a version field. This is
1304 intented to be largely invisible to the user, except that the
1305 meaning of the namestring "*.*.*" has changed: it now refers to a
1306 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1307 should usually be replaced by
1308 (make-pathname :name :wild :type :wild :version :wild)
1309 with the added benefit that this is more likely to be portable.
1310 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1311 option to OPEN now signals an error if the file being opened
1312 exists; this may have an impact on existing code.
1313 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1314 signals received, on the PowerPC platforms (both Linux and
1315 Darwin). (thanks to Pierre Mai for pointing out the location of
1317 * several fixes on OS X: The system now builds and runs cleanly on
1318 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1319 (thanks to Brian Mastenbrook)
1320 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1321 second-resolution: integer multiples of 1/3600 between -24 and 24.
1322 (thanks to Vincent Arkesteijn)
1323 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1324 type when called with 1 argument; PEEK-CHAR checked type of
1325 PEEK-TYPE only after having read first character from a
1326 stream. (reported by Peter Graves)
1327 * bug fix: the garbage collector now has much better locality
1328 behaviour, and in particular no longer treats all memory as being
1329 exhausted when many small objects point to each other in a deeply
1331 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1332 recognized as being TYPEP their class.
1333 * bug fix: the PUSHNEW documentation string has been corrected.
1334 (thanks to Vincent Arkesteijn)
1335 * bug fix: defaulting of the value for the last of an atypically
1336 large number of multiple values being bound was not being
1337 performed correctly on the Alpha or PPC platforms
1338 * optimization: implemented multiplication as a modular
1339 (UNSIGNED-BYTE 32) operation on the x86 backend.
1340 * optimization: SEARCH on simple-base-strings can now be open-coded.
1341 (see also contrib/compiler-extras.lisp for inspiration for
1342 teaching the compiler about the Boyer-Moore algorithm).
1343 * value, returned by MAX (and MIN) called with several EQUALP, but
1344 not EQL, arguments now does not depend on compiler settings.
1345 * fixed some bugs revealed by Paul Dietz' test suite:
1346 ** in stack analysis liveness information is propagated from
1347 non-local entry points.
1348 ** pathwise CAST removing failed when the CAST node did not start
1350 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1351 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1353 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1354 host is already defined.
1355 ** RENAME-FILE works on streams instead of signalling an internal
1357 ** PEEK-CHAR uses the current readtable when determining whether
1358 or not a character is whitespace.
1359 ** MERGE-PATHNAMES handles the case when the pathname does not
1360 specify a name while the default-pathname specifies a version.
1361 ** Pathnames now stand a better chance of respecting print/read
1363 ** Attempting to use standardized file system operators with a
1364 pathname with invalid :DIRECTORY components signals a
1366 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1368 ** DIRECTORY on logical pathnames is more correct.
1369 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1370 signal a TYPE-ERROR if their argument is not a stream.
1371 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1372 designator argument does not designate a stream.
1373 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1374 examining the synonym.
1375 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1377 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1378 element-type larger than ([UN]SIGNED-BYTE 32).
1380 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1381 * When built with the :SB-FUTEX feature, threaded builds now take
1382 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1383 for faster/more reliable mutex and condition variable support.
1384 * Incompatible change (but one you probably shouldn't have been using
1385 anyway): the interface and code for arbitrating between multiple
1386 threads in the same user session has been redesigned.
1387 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1388 error when called without an explicit environment argument.
1389 (thanks to Nikodemus Siivola)
1390 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1391 confused about their position. (thanks to Adam Warner and Gerd
1393 * bug fix: Pretty printing backquoted forms with unquotations in the
1394 argument list position of various code constructs such as LAMBDA
1395 now works correctly. (reported by Paul Dietz)
1396 * bug fix: Pretty printing unquotations no longer loses all
1397 stream position information.
1398 * optimization: performance of string output streams is now less
1399 poor for multiple small sequence writes.
1400 * optimization: performance of CSUBTYPEP in the presence of complex
1401 expressions involving CONS and NOT many times has been improved.
1402 (reported by Paul Dietz)
1403 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1405 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1406 streams for the REPL, for the debugger, and so on. (thanks to
1408 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1409 * fixed some bugs revealed by Paul Dietz' test suite:
1410 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1411 is ignored if it is too small, rather than propagating through
1412 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1413 ** extremely complex negations of CONS types were not being
1414 sufficiently canonicalized, leading to inconsistencies in
1416 ** VALUES tranformer lost derived type.
1418 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1419 * fixed a bootstrapping bug: the build process no longer assumes
1420 that the various BOOLE-related constants have the same value in
1421 host and target lisps. (noted by Paul Dietz' test suite on an
1422 SBCL binary built from CLISP)
1423 * The system can now be dynamically linked on the MIPS platform,
1424 which enables dynamic loading of foreign code from Lisp. (thanks
1425 to Ralf Baechle for discussions on the MIPS ABI)
1426 * The system now records debugging information for its own source
1427 files in a filesystem-position-independent manner, relative to
1428 the "SYS" logical host.
1429 * fixed a compiler bug: MV-LET convertion did not check references
1430 to the "max args" entry point. (reported by Brian Downing)
1431 * tweaked disassembly notes to be less confident about proclaiming
1432 some instruction as an LRA. (thanks to Brian Downing)
1433 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1434 now each have their own history, command character, and other
1435 characteristics. (thanks to David Lichteblau)
1436 * fixed some bugs revealed by Paul Dietz' test suite:
1437 ** compiler failure in compiling LOGAND expressions including a
1439 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1440 shift greater than 32.
1441 ** FLUSH-DEST did not mark blocks for type check regeneration.
1442 ** HANDLER-CASE failed to accept declarations in handler clauses
1443 in some circumstances.
1445 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1446 * New code in contrib/sb-introspect (still probably not entirely
1447 stable yet) provides some support for smart Lisp development
1448 environments like SLIME.
1449 * The conditions signalled for errors occurring when loading .fasl
1450 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1451 in a way which should help ASDF recover gracefully.
1452 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1453 stuff has been cleaned up. If you code contrib/ stuff, this might
1454 affect you, and you can look at contrib/README, contrib/STANDARDS,
1455 and/or the 0.8.4.27 diff to check.
1456 * In full calls the compiler now does not generate checks for declared
1457 argument types for all arguments.
1458 * various threading fixes
1459 ** and some experimental patches which didn't make it into
1460 the main tree for this release, but which are shipped in
1461 contrib/experimental-thread.patch as a possible fix for some
1462 failures (deadlock, spinning...) in GC-intensive multithreaded
1464 * fixed PPC build problem (source code incompatibility of different
1465 library versions): added offsetof() hackery which attempts to divine
1466 where glibc maintainers put uc_mcontext today
1467 * fixed bug 282: compiler does not trust type assertions while passing
1468 arguments to a full call.
1469 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1470 &OPTIONAL VALUES type parameter.
1471 * fix bug 214: algorithm for noting rejected templates is now more
1472 similar to that of template seletion. (also reported by rydis on
1474 * fixed bug 141b: printing backquoted information readably and prettily
1475 inserts a space where necessary.
1476 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1477 (CHARACTER 10) are now reported as errors, rather than propagated
1478 as unknown types. (reported by piso on #lisp)
1479 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1480 with values NIL and :ERROR. (thanks to Milan Zamazal)
1481 * fixed bug 191c: CLOS now does proper keyword argument checking as
1482 described in CLHS 7.6.5 and 7.6.5.1.
1483 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1484 longer raise an error; further, using a list as a for-as-arithmetic
1485 counter now raises a meaningful error.
1486 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1487 now understood by sequence creation functions such as MAKE-SEQUENCE
1489 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1490 a STRING-INPUT-STREAM.
1491 * compiler enhancement: SIGNUM is now better able to derive the type
1493 * type declarations inside WITH-SLOTS are checked. (reported by
1495 * fixed some bugs revealed by Paul Dietz' test suite:
1496 ** incorrect optimization of TRUNCATE for a positive first
1497 argument and negative second.
1498 ** compiler failure in let-convertion during flushing dead code.
1499 ** compiler failure while deriving type of TRUNCATE on an
1500 interval, containing 0.
1501 ** ASH of a negative bignum by a negative bignum count now returns
1503 ** intersection of CONS types now canonicalizes properly, fixing
1504 inconsistencies in SUBTYPEP.
1506 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1507 * incompatible change: The --disable-debugger command line
1508 option now clobbers the debugger at a more fundamental
1509 level, by redefining #'INVOKE-DEBUGGER instead of by
1510 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1511 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1512 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1513 enter the debugger REPL and then suffer recursive errors
1514 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1515 while under the new system, BREAK in --disable-debugger mode
1516 terminates the system just as an unhandled error would.
1517 * fixed compiler performance when processing loops with a step >1;
1518 * bug fix: DOCUMENTATION now retrieves generic function
1519 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1520 support has been systematized, and now supports the methods
1521 specified by ANSI, along with a default method and a method for
1522 slot documentation. (reported by Nathan Froyd)
1523 * bug fix: effective methods associated with a generic function are
1524 no longer cached over a change of that generic function's method
1525 combination. (reported by Andreas Fuchs)
1526 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1527 names a non-existent directory. (thanks to Andreas Fuchs)
1528 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1529 the PPC platform, lead to stack corruption; this has been fixed.
1530 (reported by Rainer Joswig)
1531 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1532 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1533 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1534 arithmetic in general, is now much more reliable on the PPC
1536 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1537 platform now returns the right answer.
1538 * optimization: restored some effective method precomputation in
1539 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1540 precomputation is now tunable.
1541 * optimization: compiler-internal data structure use has been
1542 reviewed, and changes have been made that should improve the
1543 performance of the compiler by about 20%.
1544 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1545 simple-base-strings and simple-bit-vectors is improved.
1546 * optimization: the optimization of 32-bit logical and arithmetic
1547 functions introduced in version 0.8.3 on the x86 has been
1548 implemented on the mips, ppc and sparc platforms; an
1549 implementation of the same facility, but for 64-bit arithmetic,
1550 has been added for the alpha.
1551 * microoptimization: the compiler is better able to make use of the
1552 x86 LEA instruction for multiplication by constants.
1553 * bug fix: in some situations compiler did not report usage of
1554 generic arithmetic in (SPEED 3) policy.
1555 * bug 145b fix: compiler used wrong type specifier while converting
1556 MEMBER-types to numeric.
1557 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1559 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1560 index argument now works properly on the Alpha platform.
1561 * bug fix: floating point exception treatment on the Alpha platform
1563 * bug fix: FILE-POSITION works much better on string input and
1564 output streams. (thanks to Nikodemus Siivola)
1565 * bug fix: many threading/garbage collection symptoms sorted.
1566 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1567 might be pseudo-atomic.
1568 * internal change: Stopping for GC is now done with signals not
1569 ptrace. GC is now done in whichever thread wanted it, instead of
1571 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1573 * bug fix: result form in DO is not contained in the implicit
1575 * incompatible change: ICR structure is changed; the value part of
1576 CONTINUATION is now called LVAR; corresponding functions are
1577 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1578 * added type deriver for ISQRT (thanks to Robert E. Brown).
1579 * bug fix: better support for loading from the command line when an
1580 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1582 * fixed some bugs revealed by Paul Dietz' test suite:
1583 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1584 ** ROUND and FROUND now give the right answer when given very
1585 small float arguments.
1586 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1588 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1589 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1590 ** LCM should return a non-negative integer.
1591 ** PARSE-INTEGER returned the index of a terminator instead of the
1592 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1593 ** PARSE-INTEGER returned an incorrect index being applied to a
1595 ** LCM with two arguments of 0 returns 0 rather than signalling
1597 ** unsigned addition of a 32-bit constant with the high bit set no
1598 longer causes an internal compiler error on the x86.
1599 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1600 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1601 with negative last argument.
1602 ** byte specifiers with zero size and position no longer cause
1603 an error during type derivation.
1604 ** bignum multiplication on the Alpha platform now returns the
1606 * porting: The system now builds on SuSE AMD64, although it still
1607 generates a 32-bit binary.
1608 * .fasl file incompatibility: The fasl file version number has
1609 been incremented (because of the changes to internal compiler
1610 data structures referred to above).
1612 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1613 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1614 more accurately, on the Darwin kernel running on PowerPC hardware.
1615 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1616 * Compiler code deletion notes now signal a condition of type
1617 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1618 an associated MUFFLE-WARNING restart.
1619 * The compiler now performs limited argument count validation of
1620 constant format strings in FORMAT, and where appropriate in ERROR,
1621 CERROR and WARN. (thanks to Gerd Moellmann)
1622 * New ASDF-INSTALL contrib can be used for automatic download and
1623 installation of third-party Lisp code from CCLAN or other sites
1625 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1626 which forces another thread to execute a function supplied by the
1628 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1629 now accept and act upon their :ELEMENT-TYPE keyword argument.
1630 (reported by Martin Atzmueller, Edi Weitz)
1631 * bug fix: FILE-POSITION now accepts position designators up to
1632 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1633 is the greater. (thanks to Patrik Nordebo)
1634 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1635 Kalle Olavi Niemitalo)
1636 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1637 printable. (reported by Eric Marsden)
1638 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1639 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1640 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1641 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1642 (thanks to Stig Sandoe)
1643 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1644 UNREAD-CHAR now correctly includes the unread character in the
1645 target sequence. (thanks to Gerd Moellmann)
1646 * bug fix in threaded builds: the system can now be suspended and
1647 resumed by shell job control with minimal disruption.
1648 * bug fixes in times and timezones >2038AD
1649 * better handling of "where is GNU make?" problem in build scripts
1650 (thanks to Nikodemus Siivola)
1651 * new optimization: inside a named function any reference to a
1652 function with the same name is considered to be a self-reference;
1653 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1654 optimization quality.
1655 * new optimization on x86: logical functions and + now have
1656 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1657 used when the result is truncated to 32 bits.
1658 * VALUES declaration is partially enabled.
1659 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1660 * fixed some bugs revealed by Paul Dietz' test suite:
1661 ** The system now obeys the constraint imposed by
1662 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1663 types form a lattice under type intersection.
1664 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1665 ** ASSOC now ignores NIL elements in an alist.
1666 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1667 and (1+ MOST-POSITIVE-FIXNUM) answers.
1668 ** The addition of a method with invalid qualifiers to a generic
1669 function does not cause an error to be signalled immediately;
1670 a warning is signalled, and the error is generated only on
1671 calling the generic function.
1672 * changed .fasl file version number, in part to add type codes for
1673 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1674 obscure ANSI requirements
1676 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1677 * fixed bug 148: failure to inline-expand a local function left
1678 garbage, confusing the compiler.
1679 * fixed bugs 3cd: structure slot readers perform type check if the
1680 slot can have an invalid value (i.e. it is either not initialized
1681 or can be written with a less specific slot writer).
1682 * bug fix: the compiler now traps array references to elements off
1683 the end of an array; previously, the bounds checking in some
1684 circumstances could go off-by-one.
1685 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1687 * type declarations for array element types now obey the description
1688 on the CLHS page "Declaration TYPE", as per discussions on
1689 sbcl-help around 2003-05-08. This means that a declaration
1690 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1691 declaration, all references to BAR will be asserted or assumed
1692 (with THE, so dependent on compiler policy) to involve objects of
1693 type FOO. Note that no such declaration is implied in
1694 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1695 * declared types of functions from the "Conditions"
1696 chapter. (reported by Paul Dietz)
1697 * bug fix: CERROR accepts a function as its first argument.
1698 * bug fix: NTH an NTHCDR accept a bignum as index
1699 arguments. (reported by Adam Warner)
1700 * optimization: character compare routines now optimize comparing
1701 against a constant character. (reported by Gilbert Baumann)
1702 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1703 argument now works properly on the MIPS platform.
1704 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1705 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1706 declared to have a functional type, does not cause an error
1708 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1709 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1710 * fixed bug: if last continuation of a deleted block has a
1711 destination, this destination should be deleted too. (reported by
1713 * fixed a bug in the bootstrap process: the host compiler's values
1714 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1715 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1716 test case from Patrik Nordebo)
1717 * improved the ability of the disassembler on the PPC platform to
1718 provide helpful disassembly notes.
1719 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1720 the class in more cases than previously.
1721 * bug fix: FILE-POSITION now understands :START and :END for
1722 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1723 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1724 8) data. (thanks to David Lichteblau)
1725 * bug fix: it is possible to add a method to a generic function
1726 without lambda list.
1727 * bug fix: reader failed to signal END-OF-FILE inside an
1728 object representation. (reported by Nikodemus Siivola)
1729 * fixed some bugs revealed by Paul Dietz' test suite:
1730 ** LAST and [N]BUTLAST should accept a bignum.
1731 ** condition slot accessors are methods.
1732 ** (VECTOR NIL) is a subtype of STRING.
1734 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1735 * minor incompatible change: some nonsensical specialized lambda
1736 lists (used in DEFMETHOD) which were previously ignored now signal
1738 * minor incompatible change: the system is now aware of the types of
1739 variables in the COMMON-LISP package, and will signal errors for
1740 most violations of these type constraints (where previously they
1741 were silently accepted).
1742 * minor incompatible change: COMPILE-FILE now uses the freedom
1743 afforded (ANSI 3.2.2.3) to use derived function types for
1744 functions defined in the same file. This also permits the system
1745 to warn on static type mismatches and function
1746 redefinition. (Currently it does not work with high DEBUG level.)
1747 * minor incompatible change: VALUES declaration is disabled.
1748 * When issuing notes, the compiler now signals a condition of type
1749 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1750 restart for use in user handlers. It is expected that the
1751 COMPILER-NOTE condition will eventually become a condition
1752 supertype to a hierarchy of note types, which will then be
1753 handleable in a similar fashion. However, at the moment, no such
1754 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1755 but it's an implementation detail, not a classification for the
1757 * Changes in type checking closed the following bugs:
1758 ** type checking of unused values (192b, 194d, 203);
1759 ** template selection based on unsafe type assertions (192c, 236);
1760 ** type checking in branches (194bc).
1761 * A short form of VALUES type specifier has ANSI meaning (it has
1762 increased the number of situations when SBCL cannot perform type
1764 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1765 keywords or constants is permissible.
1766 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1767 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1768 classes. (thanks to Antonio Martinez)
1769 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1770 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1771 argument to a pretty-print stream. Instead, it leaves any such
1772 operations to DESCRIBE-OBJECT methods.
1773 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1775 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1776 lambda lists. (thanks to David Lichteblau)
1777 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1778 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1779 (thanks to Gerd Moellmann)
1780 * fixed bug 246: increased compilation speed of long
1781 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1783 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1785 * DEFCONSTANT now throws a condition of type
1786 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1787 constant to a non-EQL value; CONTINUE and ABORT restarts
1788 respectively change and preserve the value.
1789 * fixed bug 63: The code walker, part of the implementation of CLOS,
1790 is now better at handling symbol macros.
1791 * bug fix: There is no longer an internal implementation type named
1792 CL:LENGTH. (reported by Raymond Toy)
1793 * bug fix: In macro-like defining macros/special operators the
1794 implicit block does not enclose the lambda list.
1795 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1796 enclosed in parentheses) are not suitable as type specifiers, and
1797 their use properly signals an error now.
1798 * bug fix: An argument count mismatch for a type specifier in code
1799 being compiled no longer causes an unhandled error at compile
1800 time, but signals a compile-time warning.
1801 * fixed simple vector readable printing
1802 * bug fix: DESCRIBE takes more care over whether the class
1803 precedence list slot of a class is bound before accessing it.
1804 (reported by Markus Krummenacker)
1805 * bug fix: FORMATTER can successfully compile pretty-printer format
1806 strings which use variants of the ~* directive inside.
1807 * bug fix: SEARCH now applies its TEST predicate to the elements of
1808 the arguments in the correct order. (thanks to Wolfhard Buss)
1809 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1810 in type inference. (thanks to Robert E. Brown)
1811 * bug fix: Reading in symbols with an explicit package name of ""
1812 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1813 * fixed some bugs revealed by Paul Dietz' test suite:
1814 ** NIL is now allowed as a structure slot name.
1815 ** Arbitrary numbers, not just REALs, are allowed in certain
1816 circumstances in LOOP for-as-arithmetic clauses.
1817 ** Multiple class redefinitions before slot access no longer
1818 causes a type error.
1819 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1820 association between the name and a class.
1821 ** Generic functions with non-standard method-combination and over
1822 five methods all of which return constants no longer return NIL
1823 after the first few invocations. (thanks to Gerd Moellmann)
1824 ** CALL-NEXT-METHOD with no arguments now passes the original
1825 values of the arguments, even in the presence of assignment.
1826 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1828 ** DISASSEMBLE works with closures and funcallable instances.
1829 ** ADD-METHOD now returns the generic function, not the new method.
1830 ** FIND-METHOD signals an error if the lengths of the specializers
1831 is incompatible with the generic function, even if the ERRORP
1833 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1834 which its argument is a member.
1835 ** DEFCLASS only redefines the class named by its class-name
1836 argument if that name is the proper name of the class;
1837 otherwise, it creates a new class.
1838 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1839 of the UNBOUND-SLOT condition to the name of the slot.
1840 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1841 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1842 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1843 SLOT-MAKUNBOUND in the specified fashion.
1845 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1846 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1847 cross-compilation host. As a consequence, we can now bootstrap our
1848 way up to SBCL starting with a bare gcc toolchain and human-readable
1849 source code (first the source to CLISP, then the source to SBCL).
1850 * A contributed module containing a partial implementation of the
1851 simple-streams interface has been included. (thanks to Rudi
1853 * A contributed module implementing the RFC1321 Message Digest
1854 Algorithm, known as MD5, has been included.
1855 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1856 no longer has any effect, as the code controlled by this feature
1857 has been deleted. (As far as we know, no-one has ever built using
1858 this feature, and its semantics were confused in any case).
1859 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1860 work on conditions (as required by the ANSI specification),
1861 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1862 expected behaviour on conditions. Users should note, however,
1863 that such behaviour is not required by the ANSI specification,
1864 and so use of this behaviour may render their code unportable.
1865 * fixed some bugs revealed by Paul Dietz' test suite:
1866 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1868 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1869 ** SXHASH on condition objects no longer returns NIL.
1870 ** :ALLOCATION :CLASS slots are better treated; their values are
1871 updated on class redefinition, and initforms inherited from
1872 superclasses are applied.
1873 ** REMOVE-METHOD returns its generic function argument even when
1874 no method was removed.
1875 ** SHARED-INITIALIZE now initializes the values of the requested
1876 slots, including those with :ALLOCATION :CLASS.
1877 ** ALLOCATE-INSTANCE now works on structure classes defined via
1878 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1880 ** SLOT-EXISTS-P now works on conditions, as well as structures
1882 ** MAKE-LOAD-FORM now has the required methods on
1883 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1884 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1885 keyword, and now implements the SLOT-NAMES argument.
1886 ** methods with &OPTIONAL arguments no longer allow too many
1887 arguments to be passed in the call without error.
1888 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1889 option is consistent with the required arguments of the generic
1890 function lambda list.
1891 * bug fix: REQUIRE accepts a string designator. (Thanks to
1893 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1894 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1895 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1896 * bug fix: The compiler now checks for duplicated variables in macro
1898 * bug fix: SETQ on globals returns the correct value.
1899 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1900 not a COMPILER-ERROR (followed by some other strange error on
1901 choosing the CONTINUE restart).
1902 * bug fix: make.sh and friends are now more consistent in the way that
1903 they look for GNU "make".
1905 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1906 * experimental native threads support (on x86 Linux >=2.4 only).
1907 This is not compiled in by default: you need to add :SB-THREAD to
1908 the target features. See the "Beyond ANSI" chapter of the manual
1910 * fix for longstanding nonANSIism: The old distinction between
1911 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1912 The return value from CL:FIND-CLASS is now a CLOS class, and
1913 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1914 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1915 * An interface to the MetaObject Protocol, as described in Kiczales,
1916 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1917 MIT Press, 1991, is available from the SB-MOP package.
1918 * incompatible change: the SB-PCL package should now be considered
1919 a private implementation detail, and no longer a semi-private MOP
1921 * minor incompatible change: due to rearrangement for threads, the
1922 control stack and binding stack are now allocated at arbitrary
1923 addresses instead of being hardcoded per-port. Users affected by
1924 this probably have to be doing advanced things with shared
1925 libraries, and will know who they are.
1926 * minor incompatible change: Previously, all --eval forms used were
1927 processed with READ before any of them were processed with EVAL.
1928 Now each --eval form is processed with both READ and EVAL before
1929 the next --eval form is processed. (Thus package operations like
1930 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1931 work as the user might reasonably expect.)
1932 * minor incompatible change: *STANDARD-INPUT* is now only an
1933 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1935 * minor incompatible change: Y-OR-N-P is now character-oriented,
1936 not line oriented. Also, YES-OR-NO-P now works without errors.
1937 (thanks to Antonio Martinez)
1938 * sb-aclrepl module improvements: an integrated inspector, added
1939 repl features, and a bug fix to :trace command.
1940 * Known functions, which cannot be open coded by the backend, are
1941 considered to be able to check types of their arguments. (fixing
1942 a bug report by Nathan J. Froyd)
1943 * fixed a bug in computing method discriminating functions: It is
1944 now possible to define methods specialized on classes which have
1945 forward-referenced superclasses. (thanks to Gerd Moellmann)
1946 * fixed evaluation order in optional entries (reported by Gilbert
1948 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1949 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1950 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1951 required-by-AMOP NAME argument, as well as CLASS and
1952 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1953 * fixed bug 20: DEFMETHOD can define methods using names that are
1954 not the proper names of classes to designate class specializers.
1955 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1956 Unix file descriptors, instead of blowing up. (thanks to Antonio
1958 * Garbage collection refactoring: user-visible change is that a
1959 call to the GC function during WITHOUT-GCING will not do garbage
1960 collection until the end of the WITHOUT-GCING. If you were doing
1961 this you were probably losing anyway.
1962 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1963 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1964 TYPEP the latter but not the former.
1965 * The compiler issues a full WARNING for calls to undefined functions
1966 with names from the CL package.
1967 * MAP-INTO for a vector destination is open coded. (reported by
1968 Brian Downing on c.l.l)
1969 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1970 documentation string.
1971 * fixed some bugs revealed by Paul Dietz' test suite:
1972 ** COPY-ALIST now signals an error if its argument is a dotted
1974 ** Condition slots are now accessed more correctly in the presence
1975 of multiple initargs for a given slot.
1976 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1977 exclude restarts of the same name associated with a different
1979 ** DEFCLASS of forward-referenced classes with another
1980 forward-referenced class in the superclasses list no longer
1982 ** Condition slots are now initialized once each, not multiple
1983 times. (thanks to Gerd Moellmann)
1984 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1986 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1987 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1988 arguments contain duplicated elements.
1989 ** RESTART-CASE understands local macros.
1990 ** RESTART-CASE associates exactly its own restarts with a condition.
1991 ** ENDP in safe mode checks its argument to be of type LIST.
1992 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1993 in question is unbound.
1994 ** Optimized MAKE-INSTANCE functions no longer cause internal
1995 assertion failures in the presence of duplicate initargs.
1996 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1997 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1998 propagated even in the ctor optimized implementation of
2000 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2002 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2003 * a better implementation of SXHASH on (simple) bit vectors,
2004 measured both in execution speed and in distribution of results
2005 over the positive fixnums, has been installed. Likewise, a better
2006 implementation of EQUAL for simple bit vectors is now available.
2007 * fixed CEILING optimization for a divisor of form 2^k.
2008 * fixed bug 240 (emitting extra style warnings "using the lexical
2009 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2010 by Antonio Martinez)
2011 * fixed SXHASH, giving different results for NIL depending on type
2012 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2013 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2014 variable. (found by Rolf Wester)
2015 * SBCL does not ignore type declarations for special
2016 variables. (reported by rif on c.l.l 2003-03-05)
2017 * some bug fixes in contrib/sb-aclrepl/
2018 * fixed some bugs revealed by Paul Dietz' test suite:
2019 ** a bug in the CONS type specifier, whereby the CAR and CDR
2020 types got intertwined, has been fixed;
2021 ** the type system is now able to reason about the interaction
2022 between INTEGER and RATIO types more completely;
2023 ** APPEND, [N]REVERSE and NRECONC check that those their
2024 arguments, which must be proper lists, are really so;
2025 ** An array specialized to be unable to hold elements has been
2026 implemented, as required -- yes, really -- by ANSI;
2027 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2028 on malformed property lists;
2030 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2031 * incompatible packaging change: in line with Unix convention,
2032 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2033 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2034 * REQUIRE and PROVIDE are now optionally capable of doing something
2035 useful. See the documentation string for REQUIRE.
2036 * infrastructure for a managed SBCL contrib system: contributed
2037 modules in this release include:
2038 ** the ASDF system definition facility;
2039 ** an interface to the BSD Sockets API;
2040 ** an ACL-like convenience interface to the repl;
2041 (thanks to Kevin Rosenberg)
2042 ** an implementation of ROTATE-BYTE, with efficient implementation
2044 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2045 gives a closer approximation to the right answer than previously.
2046 (thanks to Raymond Toy)
2047 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2048 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2049 optional environment arguments, as required by ANSI.
2050 * fixed bugs in other functions taking environment objects, allowing
2051 calls with an explicit NIL environment argument to be compiled
2053 * fixed bug 228: primary return values from
2054 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2055 COMPILE or FUNCTION.
2056 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2057 :TYPE will no longer signal errors on innocuous objects.
2058 * fixed bug 231b: SETQ is better at respecting type declarations in
2059 the lexical environment.
2060 * fixed a bug in DEFCLASS: classes named by symbols with no or
2061 unprintable packages can now be defined.
2062 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2063 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2064 * fixed compiler failure related to checking types of functions.
2065 (reported by Robert E. Brown)
2066 * the compiler is now much more consistent in its error-checking
2067 treatment of bounding index arguments to sequence functions: in
2068 (SAFETY 3) code, errors will be signalled in almost all cases if
2069 invalid sequence bounding indices are passed to functions defined
2070 by ANSI to operate on sequences.
2071 * fixed a bug in the build procedure: documentation of SBCL-specific
2072 packages is now preserved and available in the final Lisp image.
2073 * lifted FDEFINITION lookup out of loops in the implementation of
2074 many list operations. (thanks to Robert E. Brown)
2075 * fixed a bug in the reader: the #n# reader macro now works for
2076 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2077 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2078 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2079 E. Brown and Tony Martinez respectively)
2080 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2081 SLOT-BOUNDP: in particular, it is now able to optimize them much
2082 better, and is now not vulnerable to having packages renamed.
2083 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2084 always called when a slot is not present in an instance. (thanks
2086 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2087 longer signal an error on unreading a character following EOF on
2088 the previous constituent stream. (thanks to Tony Martinez)
2089 * fixed some bugs revealed by Paul Dietz' test suite:
2090 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2091 not just nonnegative fixnums;
2092 ** the logical bit-array operators such as BIT-AND now accept an
2093 explicit NIL for their "opt-arg" argument (to indicate a
2094 freshly-consed result bit-array);
2095 ** ELT now signals an error on an invalid sequence index in safe
2097 ** the type system is now cleverer about negations of numeric
2098 types, and consequently understands the BIGNUM and RATIO types
2100 ** the type system is now cleverer about the interaction between
2101 INTEGER and RATIO types: while bugs still remain, many more
2102 cases are accurately computed;
2103 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2104 if it is in the last clause;
2105 ** CONSTANTLY now correctly returns a side-effect-free function in
2107 ** DECLARE is no longer treated as a special-operator; in
2108 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2109 * incremented fasl file version number due to the change in the
2110 DEFSTRUCT-SLOT-DESCRIPTION structure.
2112 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2113 * minor incompatible change: code processed by the "interpreter" or
2114 EVAL now has a compilation optimization policy of (DEBUG 2)
2115 (changed from (DEBUG 1)) to improve debuggability of interactive
2116 development, and to allow the use of the debug RETURN command in
2118 * an experimental implementation of the RETURN command for the
2119 debugger has been included. (thanks to Frederik Kuivinen)
2120 * fixed bug 62: constraints were not propagated into a loop.
2121 * fixed bug in embedded calls of SORT (reported and investigated by
2123 * fixed some bugs revealed by Paul Dietz' test suite:
2124 ** printing and reading of arrays with some dimensions having
2125 length 0 (thanks to Gerd Moellmann);
2126 ** BOA constructor with &AUX argument without a default value does
2127 not cause a type error;
2128 ** CONSTANTP now returns true for all self-evaluating objects.
2130 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2131 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2132 accessors that are related by inheritance, as specified in the
2133 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2134 Valtteri Vuorikoski)
2135 * The compiler is now able to inline functions that were defined in
2136 a complex lexical environment (e.g. inside a MACROLET).
2137 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2139 * Support for the upcoming FreeBSD-5.0 release has been included.
2140 (thanks to Dag-Erling Smorgrav)
2141 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2142 effect when it is not in a toplevel context.
2143 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2144 stand a better chance of being correct. (thanks to Gerd
2146 * fixed bug in COERCE, which now signals an error on coercing a
2147 rational to a bounded real type which excludes the expected
2149 * The compiler is now able to derive types more accurately from the
2150 COERCE and COMPILE functions.
2151 * fixed bug 223: functional binding is considered to be constant
2152 only for symbols in the CL package.
2153 * fixed bug 231: SETQ did not check the type of a variable being set
2154 (reported by Robert E. Brown)
2155 * A new optimization for MAKE-INSTANCE has been included, fixing
2156 various bugs (including relating to :ALLOCATION :CLASS slots and
2157 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2159 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2160 ** As required by ANSI, LOOP now disallows anonymous collection
2161 clauses such as COLLECT I in conjunction with aggregate boolean
2162 clauses such as THEREIS (= I 1);
2163 ** LOOP now signals an error when any variable is reused in the
2164 same loop (including the potentially useful construct analogous
2165 to WITH A = 1 WITH A = (1+ A);
2166 ** IT is only a special loop symbol within the first clause of a
2167 conditional loop clause;
2168 ** LOOP with a typed iteration variable over a hashtable now
2169 signals a type error iff it should.
2170 * fixed some other bugs revealed by Paul Dietz' test suite:
2171 ** FILE-STREAM now names the class previously known as FD-STREAM;
2172 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2173 argument) no longer signals an error;
2174 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2175 of the slot symbol, rather than using the current package
2176 ((:CONC-NAME "") continues to intern the slot's name in the
2178 * incremented fasl file version number, because of the incompatible
2179 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2180 of the new implementation of DEFINE-COMPILER-MACRO.
2182 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2183 * Support for building SBCL for MIPS platforms running in
2184 little-endian mode has now been checked in, and basic
2185 functionality on said platforms verified.
2186 * minor incompatible change: PCL now records the pathname of a file
2187 in which methods and the like are defined, rather than its
2189 * minor incompatible change: TRUENAME now considers the truename of
2190 a file naming a directory to be the pathname with :DIRECTORY
2191 component indicating that directory.
2192 * minor incompatible change: a NAMED clause in the extended form of
2193 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2194 reason for the previous behaviour is unclear.
2195 * more systematization and improvement of CLOS and MOP conformance
2196 in PCL (thanks to Gerd Moellman and Pierre Mai):
2197 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2199 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2200 primary methods with no specializers;
2201 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2203 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2204 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2205 FORWARD-REFERENCED-CLASSes; error reporting on
2206 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2208 ** SXHASH on CLOS instances now uses a slot internal to the
2209 instance to return different numbers on distinct instances,
2210 while preserving the same return value through invocations of
2212 ** DEFMETHOD signals errors when methods with longer incongruent
2213 lambda lists are added to generic functions;
2214 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2215 CLASS, as specified in AMOP;
2216 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2217 on the order returned by the primary method for classes of
2218 class STANDARD-CLASS;
2219 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2220 * fixed some bugs shown by Paul Dietz' test suite:
2221 ** DOLIST puts its body in TAGBODY;
2222 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2224 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2225 value producing form;
2226 ** if more variables are given to PROGV than values, extra
2227 variables are bound and made to have no value;
2228 ** NSUBSTITUTE on list arguments gets the right answer with
2230 ** ELT signals an error of type TYPE-ERROR when the index argument
2231 is not a valid sequence index;
2232 ** LOOP signals (at macroexpansion time) an error of type
2233 PROGRAM-ERROR when duplicate variable names are found;
2234 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2235 ** FUNCALL of special-operators now cause an error of type
2237 ** PSETQ now works as required in the presence of side-effecting
2238 symbol-macro places;
2239 ** NCONC accepts any object as its last argument;
2240 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2242 ** loop-for-as-package does not require a package to be explicitely
2244 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2246 * fixed bug 166: compiler preserves "there is a way to go"
2247 invariant when deleting code.
2248 * fixed bug 172: macro lambda lists with required arguments after
2249 &REST arguments now cause an error to be signalled. (thanks to
2251 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2252 bug. (thanks to Matthew Danish)
2253 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2255 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2256 arguments are lexically rebound. (thanks to Gerd Moellmann and
2258 * fixed bug 194: error messages are now more informative when there
2259 is no primary method applicable in a call to a generic
2260 function. (thanks to Gerd Moellmann)
2261 * fixed bug in command line argument checking (thanks to Julian
2263 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2264 (thanks to Matthew Danish)
2265 * incremented fasl file version number, because of the
2266 SXHASH-related changes in the layout of CLOS data structures
2268 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2269 * minor incompatible change: The runtime (the Unix executable named
2270 "sbcl") is now much pickier about the .core files it will load.
2271 Essentially it now requires .core files to descend from the same
2272 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2273 as the runtime does. (The intent is to prevent the crashes which
2274 can occur, and which can even be reported as mysterious failures,
2275 when people patch the sources or change the build parameters
2276 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2277 sbcl and .core files.)
2278 * fixed bug: VALUES-LIST is no longer optimized away.
2279 * fixed bug 142: The FFI conversion of C string values to Lisp
2280 string values no longer conses excessively. (thanks to Nathan
2281 Froyd porting Raymond Toy's fix to CMU CL)
2282 * began to systematize and improve MOP conformance in PCL (thanks to
2283 Nathan Froyd, Gerd Moellman and Pierre Mai):
2284 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2286 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2287 ** FINALIZE-INHERITANCE is now called on class finalization;
2288 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2289 argument precedence order.
2290 * fixed bug 202: The compiler no longer fails on functions whose
2291 derived types contradict their declared type.
2292 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2293 so it can be non-toplevel.
2294 * The fasl file version number has changed (because of the new
2295 implementation of DEFMACRO).
2296 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2297 safe code, that any length requirement by their type-specifier
2298 argument is valid. The exceptions to this are described in bug
2300 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2301 functions now check the types of their inputs as required by ANSI.
2302 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2303 introduced symbol is DECLAREd to be SPECIAL.
2304 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2305 * fixed bug: :COUNT argument to sequence functions may be negative
2306 * fixed bug: body of DO-SYMBOLS may contain declarations
2307 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2308 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2309 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2310 symbol macro only once
2311 * fixed printing of call frame when argument list is unavailable
2312 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2313 * compiler no longer signals WARNING on unknown keyword
2316 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2317 * A beta-quality port to the mips architecture running Linux,
2318 based on the old CMUCL backend, has been made. It has been tested
2319 on a big-endian kernel, and works sufficiently well to be able to
2320 rebuild itself; it has not been tested in little-endian mode.
2321 * fixed an inconsistency between gencgc.c and purify.c which made
2322 dumping/loading .core files unreliable
2323 * fixed bug 120a: The compiler now deals correctly with IFs where
2324 the consequent is the same as the alternative, instead of
2325 misderiving the return type. (thanks to Alexey Dejneka)
2326 * fixed bug 113: Logical pathnames are now dumpable (the logical
2327 host is resolved at load-time, throwing an error if it is not
2329 * fixed bug 174: FORMAT's error message is slightly clearer when a
2330 non-printing character is used in a format directive.
2331 * fixed several bugs in compiler checking of type declarations, i.e.
2332 violations of the Python "declarations are assertions" principle
2333 (thanks to Alexey Dejneka)
2334 * fixed several bugs in PCL's error checking (thanks to Gerd
2336 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2338 * fixed bug in compilation of functions as first class values
2339 (thanks to Antonio Martinez-Shotton)
2340 * The compiler's handling TYPE-ERRORs which it can prove will
2341 inevitably happen at runtime has been cleaned up and corrected
2342 in several ways. (thanks to Alexey Dejneka)
2343 * improved argument type checking for various basic arithmetic
2344 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2345 lavished on them in the past that they can be compiled in many
2346 ways in different special cases
2347 * fixed bug 181: compiler checks validity of user supplied type
2349 * cleaned up code flushing in optimization: Function calls which
2350 should signal errors for safety purposes (e.g. which ANSI says
2351 should signal errors when their arguments are of incorrect type)
2352 are no longer optimized away.
2353 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2354 * incremented fasl file version number, because changes in the
2355 implementation of sequence functions like COERCE caused
2356 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2357 in old inline expansions) to become undefined. (Actually these
2358 changes were later undone, so we might very well be binary
2359 compatible with 0.7.7 after all, but leaving the version number
2360 incremented seemed like the simplest and most conservative
2363 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2364 * An alpha-quality port to the parisc architecture running Linux,
2365 based on the old CMUCL backend, has been made. This, even more so
2366 than the other backends, should be considered still a work in
2367 progress; known problems include that the Linux kernel in 64-bit
2368 mode does not propagate the correct sigcontext structure to
2369 userspace, and consequently SBCL on a parisc64 kernel will not
2371 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2372 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2373 suppressed.) Also now that the compiler is looking at declarations
2374 in the environment, it checks optimization declarations as well,
2375 and suppresses inlining when (> DEBUG SPEED).
2376 * More fixes have been made to treatment of floating point exception
2377 treatment and other Unix signals. In particular, floating point
2378 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2379 * The detection and handling of control stack exhaustion (infinite
2380 or very deeply nested recursion) has changed. Stack exhaustion
2381 detection is now done by write-protecting pages at the OS level
2382 and applies at all optimization settings; when found, a
2383 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2384 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2385 be caught using IGNORE-ERRORS.
2386 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2387 correctly when transforming recursive LABELS functions to LETs.
2388 (thanks to Alexey Dejneka)
2389 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2390 that are names of constants or global variables.
2391 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2392 alien routines with docstrings.
2393 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2394 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2396 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2397 (thanks to Eric Marsden)
2398 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2399 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2400 * Bug fix: The SPARC backend can now compile functions involving
2401 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2402 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2403 to a functions expecting a PATHNAME-DESIGNATOR.
2404 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2405 lambda lists. (thanks to Alexey Dejneka)
2406 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2407 functions so that EXPORTing the name of the class doesn't cause
2408 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2409 to Antonio Martinez for reporting this)
2410 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2411 constant-folded without failing an assertion. (thanks to Einar
2412 Floystad Dorum for reporting this)
2413 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2414 types are now dealt with more correctly by the compiler.
2415 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2416 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2417 behaviour analogous to RENAME-FILE. This puts its behaviour more
2418 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2420 * The fasl file version number has changed again. (because of the
2421 bug fix involving the names of PCL MAKE-INSTANCE functions)
2423 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2424 * bug fix: Floating point exceptions are treated much more
2425 consistently on the x86/Linux and PPC/Linux platforms.
2426 * Array initialization with :INITIAL-ELEMENT is now much faster for
2427 cases when the compiler cannot open code the array creation, but
2428 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2429 array accesses have also seen a speed increase.
2430 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2431 specified. (This was at the root of some bad interactions between
2432 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2434 * bug fix: Internal error arguments for undefined functions are now
2435 computed correctly on the PPC/Linux platform.
2436 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2437 Raymond Toy's patch for CMU CL)
2438 * Support for the Solaris 9 operating environment has been included
2439 (thanks to Daniel Merritt)
2440 * A very ugly but hopefully complete draft of the missing FFI chapter
2441 of the manual has been created by reformatting the corresponding
2442 CMU CL manual chapter into (currently very ugly and incoherent)
2443 DocBook and bringing it up to date for SBCL behavior. Thus, the
2444 manual is now essentially complete, at least by my extreme
2445 once-and-only-once standards, whereby it's acceptable to refer to
2446 the doc strings of SB-EXT functions as the primary documentation.
2447 * The fasl file version number has changed again, due to cleanup of
2448 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2449 is no longer a static symbol.)
2451 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2452 * SBCL now builds with OpenMCL (version 0.12) as the
2453 cross-compilation host; also, more progress has been made toward
2454 bootstrapping under CLISP.
2455 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2457 * bug 158 fixed: The compiler can now deal with integer loop
2458 increments different from 1; fixing this turned out also to fix
2460 * bug 169 fixed: no more bogus warnings about using lexical bindings
2461 despite the presence of perfectly good SPECIAL declarations (thanks
2462 to David Lichteblau)
2463 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2464 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2465 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2467 * bug 180 fixed: Method combination specifications no longer ignore
2468 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2469 * bug fix: Structure type predicate functions now check their argument
2470 count as they should.
2471 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2472 correctly. (thanks to Pierre Mai)
2473 * minor incompatible change: The --noprogrammer option is deprecated
2474 in favor of the new --disable-debugger option, which is very similar.
2475 (The major difference is that it takes effect at a slightly different
2476 time at startup, causing handling of errors in --sysinit and
2477 --userinit files will be affected differently.) The
2478 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2479 been added to allow this functionality to be controlled from ordinary
2480 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2481 maintainers, who might want to run non-interactive scripts to
2482 build SBCL cores which will later be used interactively.)
2483 * minor incompatible change: The LOAD function no longer, when given
2484 a wild pathname to load, loads all files matching that pathname.
2485 Instead, an error of type FILE-ERROR is signalled.
2487 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2488 * bug 147 fixed: The compiler preserves its block link/count
2489 invariants more correctly now so that it doesn't crash. (thanks
2491 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2493 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2494 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2495 no-value-for-this-character as NIL (as specified by ANSI).
2496 * HOST-NAMESTRING on physical pathnames now returns a string that is
2497 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2498 (thanks to Christophe Rhodes)
2499 * The Alpha port handles icache flushing more correctly. (thanks to
2501 * More progress has been made toward bootstrapping under CLISP. (thanks
2502 to Christophe Rhodes)
2503 * The fasl file format has changed again, because dynamic loading
2504 on OpenBSD (which has non-ELF object files) motivated some cleanups
2505 in the way that foreign symbols are transformed and passed around.
2506 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2507 is no longer treated as whitespace by the reader, but instead as
2508 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2509 |A
\7fB|, instead of A as it used to.
2511 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2512 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2513 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2514 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2515 this as of 0.7.1.45, but was left out of the previous news section
2516 (thanks to Dan Barlow)
2517 * SBCL now runs on the Solaris operating system on SPARC architectures
2518 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2519 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2520 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2521 * SPARC backend cleanups, allowing builds of cores optimized for V8
2522 and V9 SPARCS, and also emission of code targeted to a particular
2523 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2525 * SBCL is closer to bootstrapping under CLISP, thanks to various
2526 fixes by Christophe Rhodes.
2527 * The fasl file format has changed again, to allow the compiler's
2528 INFO database to support symbol macros.
2529 * The user manual (in doc/) is formatted into HTML more nicely.
2530 (thanks to coreythomas)
2531 * The system is smarter about SUBTYPEP relationships, especially
2532 those involving NOT types (including types such as ATOM which are
2533 represented internally using NOT types). Thus SUBTYPEP is less
2534 likely to return (VALUES NIL NIL) in general, and in particular
2535 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2537 * The fasl file format has changed again, because the internal
2538 representation of types now includes a new slot to support the new
2539 SUBTYPEP-of-NOT-types logic.
2540 * (not a change in the main branch of SBCL, but a related prototype
2541 which can hopefully be merged into the main branch of SBCL in the
2542 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2543 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2544 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2545 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2546 dispatch macro character combinations. (thanks to Alexey Dejneka)
2547 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2548 that superficially look like logical namestrings correctly.
2549 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2551 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2552 default supplied-p) for &optional and &key arguments. (thanks to
2555 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2556 * incompatible change: The compiler is now less aggressive about
2557 tail call optimization, doing it only when (> SPACE DEBUG) or
2558 (> SPEED DEBUG). (This is an incompatible change because there are
2559 programs which relied on the old CMU-CL-style behavior to optimize
2560 away their unbounded recursion which will now die of stack overflow.)
2561 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2562 for non-GENCGC systems has been increased to 20M (since that
2563 seems much closer to the likely performance optimum for modern
2564 systems than the old 4M value was)
2565 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2566 and *DEBUG-PRINT-LEVEL*
2567 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2568 of CMU CL's support for SPARC, and various endianness and other
2569 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2570 * new syntactic sugar for the Unix command line: --load foo.bar is now
2571 an alternate notation for --eval '(load "foo.bar")'.
2573 ** The system now detects stack overflow and handles it gracefully,
2574 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2575 optimization settings. (This is a good thing in general, and
2576 its introduction in this version should be particularly timely
2577 for anyone whose code fails because of suppression of tail
2579 ** The system now hunts for the C variable "environ" in a more
2580 devious way, to avoid segfaults when the C library version differs
2581 between compile time and run time. (thanks to Christophe Rhodes)
2582 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2583 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2584 ** The compiler no longer issues bogus style warnings for undefined
2585 classes in the same source file as the DEFCLASSes which defined
2586 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2588 ** fixes in CONDITION class precedence list for undefined function
2589 errors (thanks to Alexei Dejneka)
2590 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2591 correctly. (thanks to Dan Barlow)
2592 ** portability fixes aiming at bootstrapping under CLISP (thanks
2593 to Dave McDonald and Christophe Rhodes)
2594 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2595 ** fixes in type translation and and type inference (thanks to
2597 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2598 ** various fixes in the new ports (thanks to Dan Barlow)
2599 * several changes related to debugging:
2600 ** suppression of tail recursion, as noted above
2601 ** stack overflow detection, as noted above
2602 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2603 is now the default. (For some time encapsulation has been more
2604 reliable than the breakpoint-based :ENCAPSULATE NIL
2605 implementation, at least on X86 systems; and I just noticed that
2606 encapsulation also seems closer to the spirit of the ANSI
2609 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2611 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2612 up properly again. (There was a packaging bug in 0.7.0 which
2613 left their definitions in SB-SYS::LOAD-FOREIGN and
2614 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2615 vital for most things which interface to C-level interfaces,
2616 like extensions working with sockets or databases or
2617 Perl-compatible regexes or whatever, and the need to fix
2618 this bug is the main reason that 0.7.1 was released so
2620 ** DEFGENERIC is now choosier about the methods it redefines, so that
2621 reLOADing a previously-LOADed file containing DEFGENERICs does
2622 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2623 cycle now works as it should. (thanks to Alexey Dejneka)
2624 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2625 fixed by Christophe Rhodes. (He actually submitted this patch
2626 months ago, and I delayed until after 0.7.0.)
2627 ** Bug 111 (internal compiler confusion about runtime checks on
2628 FUNCTION types) was fixed by Alexey Dejneka.
2629 * Some internal cleanups (getting rid of variables which aren't
2630 needed now that the byte interpreter is gone) caused the fasl
2631 file format number to change again.
2633 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2634 * major incompatible change: The default fasl file extension, i.e. the
2635 default extension for files produced by COMPILE-FILE, has changed
2636 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2638 ** There are many changes in the implementation of the compiler.
2639 SBCL is now essentially a compiler-only implementation of ANSI
2640 Common Lisp. EVAL still "interprets" a few special cases, but
2641 almost all the interesting cases are handled by creating
2642 a LAMBDA expression, calling COMPILE on it, then calling
2643 FUNCALL on the result.
2644 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2645 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2646 Since the code is newer, there might still be some new bugs
2647 (though not as many as before Martin Atzmueller's fixes:-). But
2648 the new code is substantially simpler and clearer, and hopefully
2649 any remaining bugs will be simpler, less fundamental, and more
2650 fixable then the bugs in the old code.
2651 ** The revised compiler is still a little unsteady on its feet.
2653 *** The debugging information it produces (particularly the names
2654 of FUNCTION objects) is sometimes much less useful than what
2655 the old compiler produced.
2656 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2657 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2658 within a MACROLET) has been temporarily weakened.
2659 ** There are new compiler optimizations for various functions:
2660 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2661 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2662 *** the math functions TRUNCATE, FLOOR, and CEILING
2663 *** the function-of-all-trades COERCE
2664 Mostly these should be transparent, but there's one
2665 potentially-annoying problem (bug 117): when the compiler
2666 inline-expands a function and does type analysis on the result,
2667 it can create control paths which have type mismatches, and
2668 when it can't prove that those control paths aren't taken,
2669 it will issue WARNINGs about the type mismatches. This is
2670 a particular problem in practice for the new sequence functions.
2671 It's not clear how this should be fixed, and for now, a
2672 workaround is given in the entry for 117 in the BUGS file.
2673 ** (Because of the interaction between the two previous items --
2674 occasional inlining problems and new inline expansions -- some
2675 of the new sequence function optimizations won't really kick in
2676 completely until debugging information, and then inlining, are
2677 straightened out in some future version.)
2678 * minor incompatible changes:
2679 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2680 DIRECTORY no longer implicitly promotes NIL slots of its
2681 pathname argument to :WILD. In particular, when you ask for the
2682 contents of a directory (which you used to be able to do without
2683 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2684 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2685 ** changes in behavior that ANSI explicitly defines to be
2686 implementation dependent:
2687 *** The new compiler-only implementation still conforms with ANSI,
2688 but acts a little different than before. Besides the obvious
2689 changes in performance tradeoffs (that the cost per form passed
2690 to EVAL has gone up, and the cost per form executed by EVAL
2691 has gone down), the behavior of the system changes a little
2692 because there are no longer any interpreted function objects.
2693 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2694 e.g. doing COMPILE on the output of interactive DEFUN is
2696 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2698 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2699 has changed from (:CL) to NIL.
2700 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2701 e.g. #\Newline and #\Tab, don't have names specified in the
2702 ANSI Common Lisp standard, is now based on their ASCII symbolic
2703 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2704 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2705 are no longer used for output.
2706 ** changes in internal implementation constants:
2707 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2708 4 million. (If your application spends a lot of time GCing and
2709 you have a lot of RAM, you might want to experiment with
2710 increasing it even more.)
2711 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2712 However, almost all old code should still continue to work without
2713 immediate update, as SB-C-CALL is now a (deprecated) nickname
2715 ** Old operator names in the style DEF-FOO are now deprecated in
2716 favor of new corresponding names DEFINE-FOO, for consistency with
2717 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2718 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2719 internal symbols, but a few supported extensions like
2720 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2721 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2722 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2723 etc. as you get deeper into recursive calls to the debugger
2724 command loop, instead of the old "5]", "5]]", "5]]]"
2725 sequence. (I was motivated to do this when squabbles between
2726 ILISP and SBCL left me very deeply nested in the debugger. In the
2727 short term, this change will probably provoke more ILISP/SBCL
2728 squabbles, but hopefully it will be an improvement in the long run.)
2729 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2730 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2731 function, since declaiming return types involving aliens is
2732 (1) annoyingly messy to do by hand and (2) vital to efficient
2733 compilation of code which calls such functions.
2734 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2735 longer reexported by the SB-EXT package. They're solely useful
2736 for alien code, so it seems more logical that you should get
2737 them from the SB-ALIEN package, not in SB-EXT.
2738 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2739 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2740 features. Instead, the code that they used to control is always
2741 built into the system.
2742 * many other bug fixes
2743 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2744 advantage of the new EVAL-WHEN stuff and to clean them up in
2745 general, and they are now more ANSI-compliant in a number of
2746 ways. Martin Atzmueller is responsible for a lot of this.
2747 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2749 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2750 *** correct ERROR type for various file operations
2751 *** some fixes for Lisp streams
2752 *** DEFMETHOD syntax checking
2753 *** changing old weird representation of debug information as
2754 strings (which, among their other deficiencies, don't transform
2755 correctly when you rename packages, and don't change their
2756 print representation when you change things like *PACKAGE*
2757 and *PRINT-LENGTH*) to symbols and lists of symbols
2758 He also made several improvements and fixed several bugs in DESCRIBE.
2759 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2761 *** misbehavior of WRITE-STRING/WRITE-LINE
2762 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2763 and several other LOOP problems as well
2764 *** DIRECTORY when similar filenames are present
2765 *** DEFGENERIC with :METHOD options
2766 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2767 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2768 *** argument ordering in FIND with :TEST option
2769 *** mishandled package designator argument in APROPOS-LIST
2770 *** various problems in the backquote readmacro
2771 *** a bug in APROPOS
2772 *** probably some others that I'm not describing very well here,
2773 since the CVS log documents them by reference to sbcl-devel
2774 messages, and the SourceForge archives aren't working well.:-(
2775 ** Dan Barlow improved the Alpha port (and is making progress on the
2776 PPC port, for those of you who think different).
2777 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2778 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2779 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2780 continues to work on the SPARC port (for those of you in a position
2781 to look down upon our little PC-compatible boxes from a great height).
2782 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2783 and uses that copy, rather than the current dynamic value, when
2784 it's trying to decide whether to truncate output. Thus e.g.
2785 (let ((*print-lines* 50))
2786 (pprint-logical-block (stream nil)
2788 (let ((*print-lines* 8))
2789 (print (aref possiblybigthings i) stream)))))
2790 should now truncate the logical block only at 50 lines, instead of
2791 often truncating it at 8 lines, as it did before.
2792 * The doc/cmucl/ directory, containing old CMU CL documentation
2793 from the time of the fork, is no longer part of the base system.
2794 SourceForge has shut down its anonymous FTP service, and with it
2795 my original plan for distributing the old CMU CL documentation
2796 there. For now, if you need these files you can download an old
2797 SBCL source release and extract them from it.
2798 * The fasl file version number changed again, for dozens of reasons,
2799 some of which are apparent above.
2801 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2802 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2803 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2804 patch, so that DISASSEMBLE output is much nicer.
2805 * The code in the SB-PROFILE package now seems reasonably stable.
2806 I still haven't decided what the final interface should look like
2807 (I'd like PROFILE to interact cleanly with TRACE, since both
2808 facilities use function encapsulation) but if you have a need
2809 for profiling now, you can probably use it successfully with
2810 the current CMU-CL-style interface.
2811 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2812 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2813 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2814 link now returns the dangling link itself, and for similar
2815 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2816 link itself. (In these cases the old code signalled an error and
2817 looped endlessly, respectively.) Thus, DIRECTORY now works even
2818 in the presence of dangling and cyclic symbolic links.
2819 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2820 is now a supported extension again, since the consensus on
2821 sbcl-devel was that it can be useful for ordinary development
2822 work, not just for debugging SBCL itself.
2823 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2824 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2825 that currently-defined functions might be redefined later with
2826 different return types.
2827 * Hash tables can be printed readably, as inspired by CMU CL code
2828 of Eric Marsden and SBCL code of Martin Atzmueller.
2829 * better error handling in CLOS method combination, thanks to
2830 Martin Atzmueller porting Pierre Mai's CMU CL patches
2831 * more overflow fixes for >16Mbyte I/O buffers
2832 * A bug in READ has been fixed, so that now a single Ctrl-D
2833 character suffices to cause end-of-file on character streams.
2834 In particular, now you only need one Ctrl-D at the command
2835 line (not two) to exit SBCL.
2836 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2838 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2839 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2840 does the right thing.
2841 * The implementation of some type tests, especially for CONDITION
2842 types, is now tidier and maybe faster, due to CMU CL code
2843 originally by Douglas Crosher, ported by Martin Atzmueller.
2844 * Some math functions have been fixed, and there are new
2845 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2846 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2847 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2848 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2849 likely to be built into sbcl-0.7.0.)
2850 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2851 CL, ported to SBCL by Martin Atzmueller)
2852 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2853 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2854 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2855 * There's a new slam.sh hack to shorten the edit/compile/debug
2856 cycle for low-level changes to SBCL itself, and a new
2857 :SB-AFTER-XC-CORE target feature to control the generation of
2858 the after-xc.core file needed by slam.sh.
2859 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2860 COMPILE-FILE is no longer supported, so that now every function
2861 gets an entry point, so that block compilation looks a little
2862 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2863 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2864 returns the number of bytes consed since the system started,
2865 rather than the number consed since the first time the function
2866 was called. (The new definition parallels ANSI functions like
2867 CL:GET-INTERNAL-RUN-TIME.)
2868 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2869 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2870 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2871 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2872 * Fasl file version numbers are now independent of the target CPU,
2873 since historically most system changes which required version
2874 number changes have affected all CPUs equally. Similarly,
2875 the byte fasl file version is now equal to the ordinary
2878 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2879 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2880 is no longer recognized. I apologize for this, because it was
2881 listed in SB-EXT as a supported extension, but I found that
2882 its existing behavior was poorly specified, as well as incorrectly
2883 specified, and it looked like too much of a mess to straighten it
2884 out. I have enough on my hands trying to get ANSI stuff to work..
2885 * many patches ported from CMU CL by Martin Atzmueller, with
2886 half a dozen bug fixes in pretty-printing and the debugger, and
2887 half a dozen others elsewhere
2888 * fixed bug 13: Floating point infinities are now supported again.
2889 They might still be a little bit flaky, but thanks to bug reports
2890 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2891 as flaky as they were.
2892 * The --noprogrammer command line option is now supported. (Its
2893 behavior is slightly different in detail from what the old man
2894 page claimed it would do, but it's still appropriate under the
2895 same circumstances that the man page talks about.)
2896 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2897 are now supported, and enabled by default. Thus, the compiler can
2898 handle many floating point and complex operations much less
2899 inefficiently. (Thus e.g. you can implement a complex FFT
2901 * The compiler now detects type mismatches between DECLAIM FTYPE
2902 and DEFUN better, and implements CHECK-TYPE more correctly, and
2903 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2904 thanks to patches from Martin Atzmueller.
2905 * various fixes to make the cross-compiler more portable to
2906 ANSI-conforming-but-different cross-compilation hosts (notably
2907 Lispworks for Windows, following bug reports from Arthur Lemmens)
2908 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2909 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2910 fixed thanks to Pierre Mai's CMU CL patches.
2911 * fixes to keep the system from overflowing internal counters when
2912 it tries to use i/o buffers larger than 16M bytes
2913 * fixed bug 45a: Various internal functions required to support
2914 complex special functions have been merged from CMU CL sources.
2915 (When I was first setting up SBCL, I misunderstood a compile-time
2916 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2917 * improved support for type intersection and union, fixing bug 12
2918 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2919 more obscure bugs as well
2920 * some steps toward byte-compiling non-performance-critical
2921 parts of the system, courtesy of patches from Martin Atzmueller
2922 * Christophe Rhodes has made some debian packages of sbcl at
2923 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2924 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2925 stable, but are nonetheless usable. When he's ready, I'd be happy
2926 to add them to the SourceForge "File Releases" section. (And if
2927 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2928 * new fasl file format version number (because of changes in
2929 internal representation of (OR ..) types to accommodate the new
2930 support for (AND ..) types, among other things)
2932 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2933 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2935 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2937 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2938 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2939 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2940 * DESCRIBE now works on structure objects again.
2941 * Most function call argument type mismatches are now handled as
2942 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2943 know whether the function will be redefined before the call is
2944 executed. (The compiler could flag local calls with full WARNINGs,
2945 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2946 it doesn't keep track of enough information to know whether calls
2947 are local in this sense.)
2948 * Compiler output is now more verbose, with messages truncated
2949 later than before. (There should be some supported way for users
2950 to override the default verbosity, but I haven't decided how to
2951 provide it yet, so this behavior is still controlled by the internal
2952 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2953 src/compiler/ir1util.lisp.)
2954 * Fasl file format version numbers have increased again, because
2955 support for the Gray streams extension changes the layout of the
2956 system's STREAM objects.
2957 * The Gray subclassable streams extension now works, thanks to a
2958 patch from Martin Atzmueller.
2959 * The full LOAD-FOREIGN extension (not just the primitive
2960 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2961 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2962 but like most other programs, it defaults to copying the Unix
2963 environment from the original process instead of starting the
2964 new process in an empty environment.
2965 * Extensions which manipulate the Unix environment now support
2966 an :ENVIRONMENT keyword option which doesn't smash case or
2967 do other bad things. The CMU-CL-style :ENV option is retained
2968 for porting convenience.
2969 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2970 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2972 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2974 * A patch from Martin Atzmueller seems to have solved the SIGINT
2975 problem, and as far as we know, signal-handling now works cleanly.
2976 (If you find any new bugs, please report them!)
2977 * The system no longer defaults Lisp source file names to types
2978 ".l", ".cl", or ".lsp", but only to ".lisp".
2979 * The compiler no longer uses special default file extensions for
2980 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2981 seems to expect a single default extension for all compiled code,
2982 and there's no compelling reason to try to stretch the standard
2983 to allow two different extensions.) Instead, byte-compiled files
2984 default to the same extension as native-compiled files.
2985 * Fasl file format version numbers have increased again, because
2986 a rearrangement of internal implementation packages made some
2987 dumped symbols in old fasl files unreadable in new cores.
2988 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2989 many fewer weird special cases.
2990 * Bug #17 (differing COMPILE-FILE behavior between logical and
2991 physical pathnames) has been fixed, and some related misbehavior too,
2992 thanks to a patch from Martin Atzmueller.
2993 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2994 by Tim Moore, ported to SBCL by Martin Atzmueller.
2995 * Martin Atzmueller fixed several filesystem-related problems,
2996 including bug #36, in part by porting CMU CL patches, which were
2997 written in part by Paul Werkowski.
2998 * More compiler warnings in src/runtime/ are gone, thanks to
2999 more patches from Martin Atzmueller.
3000 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3003 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3005 * DESCRIBE now works on CONDITION objects.
3006 * The debugger now handles errors which arise when trying to print
3007 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3009 * The build system now uses an additional file, customize-target-features.lisp,
3010 to allow local modifications to the target *FEATURES* list. (The point of
3011 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3012 debugging features enabled, without having to worry about propagating it
3013 into everyone's system when I do a "cvs update".) When no
3014 customize-target-features.lisp file exists, the target *FEATURES* list
3015 should be constructed the same way as before.
3016 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3017 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3018 warn about behavior which is undefined under the ANSI spec. Note: This
3019 is specified by ANSI, but it's not very popular with programmers.
3020 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3021 macro in the SBCL sources for an example of a workaround which you
3022 might use to make portable ANSI-standard code which does what you want.
3023 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3024 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3025 when it's used as a non-toplevel form. (This is required in order
3026 to implement the DEFCONSTANT-EQX macro.)
3027 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3028 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3029 funny things, due to bugs in the implementation of EVAL-WHEN
3030 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3031 non-toplevel DEFCONSTANTs are uncommon.)
3032 * The core file version number and fasl file version number have been
3033 incremented, because the old noncompliant DEFCONSTANT behavior involved
3034 calling functions which no longer exist, and because I also took the
3035 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3036 * fixed bug 1 (error handling before read-eval-print loop starts), and
3037 redid debugger restarts and related debugger commands somewhat while
3039 ** The QUIT debugger command is gone, since it did something
3040 rather different than the SB-EXT:QUIT command, and since it never
3041 worked properly outside the main toplevel read/eval/print loop.
3042 Invoking the new TOPLEVEL restart provides the same functionality.
3043 ** The GO debugger command is also gone, since you can just invoke
3044 the CONTINUE restart directly instead.
3045 ** The TOP debugger command is also gone, since it's redundant with the
3046 FRAME 0 command, and since it interfered with abbreviations for the
3048 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3050 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3051 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3052 some porting work by Martin Atzmueller)
3053 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3054 with a boatload of patches which he ported to SBCL. Now that those
3056 ** The system tries to make sure that its low-priority messages
3057 are prefixed by semicolons, to help people who like to use
3058 syntax highlighting in their ILISP buffer. (This patch
3059 was originally due to Raymond Toy.)
3060 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3061 patches originally written by Raymond Toy.
3062 ** The compiler understands coercion between single-value and
3063 multiple-VALUES type expressions better, getting rid of some very
3064 weird behavior, thanks to patches originally by Robert MacLachlan
3065 and Douglas Crosher.
3066 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3067 lambda lists, thanks to a patch originally by Pierre Mai.
3068 ** The system no longer bogusly warns about "abbreviated type
3070 ** The compiler gets less confused by inlining and RETURN-FROM,
3071 thanks to some patches originally by Tim Moore.
3072 ** The system no longer hangs when dumping circular lists to fasl
3073 files, thanks to a patch originally from Douglas Crosher.
3074 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3075 undefined function error.
3076 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3077 in structure slot initforms) for now, documented workaround instead:-|
3078 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3079 * fixed bug 5: added stubs for various Gray stream functions called
3080 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3081 installed, at least appropriate type errors are generated
3082 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3083 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3084 consistently in DEFMETHOD forms.
3085 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3086 it doesn't seem to affect SBCL after all
3087 * The C runtime system now builds with better optimization and many
3088 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3090 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3092 * The system is now under CVS at SourceForge (instead of the
3093 CVS repository on my home machine).
3094 * The new signal handling code has been tweaked to treat register
3095 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3096 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3097 downstream, e.g. in debug-int.lisp, has implicit dependencies
3098 on the unsignedness of integer representation of machine words,
3099 and that caused the system to bomb out with infinite regress
3100 when trying to recover from type errors involving signed values,
3101 e.g. (BUTLAST '(1 2 3) -1).)
3102 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3103 (This was one of the bugs Peter Van Eynde reported back in July.)
3104 * The system now uses code inspired by Colin Walters' O(N)
3105 implementation of MAP (from the cmucl-imp@cons.org mailing
3106 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3107 inline the MAP operation, and there is more than one
3108 sequence argument to the MAP call (so that it can't just
3109 do ETYPECASE once and for all based on the type of the
3110 single sequence argument). (The old non-inline implementation
3111 of the general M-argument sequence-of-length-N case required
3112 O(M*N*N) time when any of the sequence arguments were LISTs.)
3113 * The QUIT :UNIX-CODE keyword argument has been renamed to
3114 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3115 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3117 * Raymond Wiker's patches to port dynamic loading from Linux to
3118 FreeBSD have been added.
3119 * The BUGS file is now more nearly up to date, thanks in large part
3120 to Martin Atzmueller's review of it.
3121 * The debugger now flushes standard output streams before it begins
3122 its output ("debugger invoked" and so forth).
3123 * The core version number and fasl file version number have both
3124 been incremented, because of incompatible changes in the layout
3126 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3127 used to not be called for a saved Lisp image.)
3128 * Martin Atzmueller's version of a patch to fix a compiler crash,
3129 as posted on sbcl-devel 13 September 2000, has been installed.
3130 * Instead of installing Martin Atzmueller's patch for the
3131 compiler transform for SUBSEQ, I deleted the compiler transform,
3132 and transforms for some similar consing operations.
3133 * A bug in signal handling which kept TRACE from working on OpenBSD
3135 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3136 away by constant folding
3137 * The system now defines its address space constants in one place
3138 (in the Lisp sources), and propagates them automatically elsewhere
3139 (through GENESIS and the sbcl.h file). Therefore, patching the
3140 address map is less unnecessarily tedious and error-prone. The
3141 Lisp names of address space constants have also been systematized.
3142 * CVS tags like dollar-Header-dollar have been removed from
3143 the sources, because they have never saved me trouble and
3144 they've been source of trouble working with patches and other
3145 diff-related operations.
3146 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3147 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3149 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3151 * The system has been ported to OpenBSD.
3152 * The system now compiles with a simple "sh make.sh" on the systems
3153 that it's supported on. I.e., now you no longer need to tweak
3154 text in the target-features.lisp-expr and symlinks in src/runtime/
3155 by hand, the make.sh takes care of it for you.
3156 * The system is no longer so grossly inefficient when compiling code
3157 involving vectors implemented as general (not simple) vectors (VECTOR T),
3158 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3159 which dares to use the various sequence functions on non-simple
3160 vectors, takes less of a performance hit.
3161 * There is now a primitive type predicate VECTOR-T-P
3162 to test for the (VECTOR T) type, so that e.g.
3163 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3164 can now be compiled with some semblance of efficiency. (The old code
3165 turned the type declaration into a full call to %TYPEP at runtime!)
3166 * AREF on (VECTOR T) is still not fast, since it's still compiled
3167 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3168 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3169 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3170 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3171 without making full calls to SUBTYPEP at runtime.
3172 (Some analogous efficiency issues for non-simple vectors specialized to
3173 element types other than T, or for non-simple multidimensional arrays,
3174 have not been addressed. They could almost certainly be handled the
3175 same way if anyone is motivated to do so.)
3176 * The changes in array handling break binary compatibility, so
3177 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3178 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3179 instead of (VALUES T T).
3180 * By following the instructions that Dan Barlow posted to sbcl-devel
3181 on 2 July 2000, I was able to enable primitive dynamic object
3182 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3183 functionality is not implemented (since it calls ld to resolve
3184 library references automatically, requiring RUN-PROGRAM for its
3185 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3186 try to resolve library references) is now supported.
3187 * The system now flushes the standard output streams when it terminates,
3188 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3189 them at several other probably-convenient times, e.g. in each pass of
3190 the toplevel read-eval-print loop, and after evaluating a form given
3191 as an "--eval" command-line option. (These changes were motivated by a
3192 discussion of stream flushing issues on cmucl-imp in August 2000.)
3193 * The source transform for TYPEP of array types no longer assumes
3194 that an array whose element type is a not-yet-defined type
3195 is implemented as an array of T, but instead punts, so that the
3196 type will be interpreted at runtime.
3197 * There is now some support for cross-compiling in make.sh: each of
3198 the phases of make.sh has its own script. (This should be transparent
3199 to people doing ordinary, non-cross-compile builds.)
3200 * Since my laptop doesn't have hundreds of megabytes of memory like
3201 my desktop machine, I became more motivated to do some items on
3202 my to-do list in order to reduce the size of the system a little:
3203 ** Arrange for various needed-only-at-cold-init things to be
3204 uninterned after cold init. To support this, those things have
3205 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3206 symbols with such names are now uninterned after cold init).
3207 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3208 fasl files for cold load.
3209 ** Remove the old compiler structure pooling code (which used to
3210 be conditional on the target feature :SB-ALLOC) completely.
3211 ** Redo the representation of some data in cold init to be more compact.
3212 (I also looked into supporting byte compiled code at bootstrap time,
3213 which would probably reduce the size of the system a lot, but that
3214 looked too complicated, so I punted for now.)
3215 * The maximum signal nesting depth in the src/runtime/ support code has
3216 been reduced from 4096 to 256. (I don't know any reason for the very
3217 large old value. If the new smaller value turns out to break something,
3218 I'll probably just bump it back up.)
3219 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3221 * Many, many bugs reported by Peter Van Eynde have been added to
3222 the BUGS list; some have even been fixed.
3223 * While enabling dynamic object file loading, I tried to make the
3224 code easier to understand, renaming various functions and variables
3225 with less ambiguous names, and changing some function calling
3226 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3227 * While trying to figure out how to do the OpenBSD port, I tried to
3228 clean up some of the code in src/runtime/. In particular, I dropped
3229 support for non-POSIX signal handling, added various comments,
3230 tweaked the code to reduce the number of compilation warnings, and
3231 renamed some files to increase consistency.
3232 * To support the new automatic configuration functionality in make.sh,
3233 the source file target-features.lisp-expr has been replaced with the
3234 source file base-target-features.lisp-expr and the machine-generated
3235 file local-target-features.lisp-expr.
3236 * fixed a stupid quoting error in make.sh so that using CMU CL
3237 "lisp -batch" as cross-compilation host works again
3239 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3241 * DESCRIBE no longer tries to call itself recursively to describe
3242 bound/fbound values, so that it no longer fails on symbols which are
3243 bound to themselves (like keywords, T, and NIL).
3244 * DESCRIBE now works on generic functions.
3245 * The printer now prints less-screwed-up representations of closures
3246 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3247 * A private symbol is used instead of the :EMPTY keyword previously
3248 used to mark empty slots in hash tables. Thus
3249 (DEFVAR *HT* (MAKE-HASH-TABLE))
3250 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3251 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3252 now does what ANSI says that it should. (You can still get
3253 similar noncompliant behavior if bang on the hash table
3254 implementation with all the symbols you get back from
3255 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3256 This breaks binary compatibility, since tests for equality to
3257 :EMPTY are wired into things like the macroexpansion of
3258 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3260 * There's now a minimal placeholder implementation for CL:STEP,
3261 as required by ANSI.
3262 * An obscure bug in the interaction of the normal compiler, the byte
3263 compiler, inlining, and structure predicates has been patched
3264 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3265 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3267 * Missing ordinary arguments in a macro call are now detected even
3268 when the macro lambda list contains &KEY or &REST.
3269 * The debugger no longer complains about encountering the top of the
3270 stack when you type "FRAME 0" to explicitly instruct it to go to
3271 the top of the stack. And it now prints the frame you request even
3272 if it's the current frame (instead of saying "You are here.").
3273 * As specified by ANSI, the system now always prints keywords
3274 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3276 * The default initial SIZE of HASH-TABLEs is now smaller.
3277 * Type information from CLOS class dispatch is now propagated
3278 into DEFMETHOD bodies, so that e.g.
3279 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3281 is now basically equivalent to
3282 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3283 (DECLARE (TYPE SINGLE-FLOAT X))
3285 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3286 operation, without having to do run-time type dispatch.
3287 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3288 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3289 * Since I don't seem to be making big file reorganizations very often
3290 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3291 is overflowing my ability to conveniently back them up), I've finally
3292 checked the system into CVS. (The CVS repository is on my home system,
3293 not at SourceForge -- putting it on SourceForge might come later.)
3294 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3295 high-level GC-NOTIFY-FOO functions send their output. (There's
3296 still very little control of where low-level verbose GC functions
3297 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3298 less than it used to -- the GC-NOTIFY-FOO functions are now under
3299 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3300 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3301 in only one place in the source code, and propagates it automatically
3302 everywhere that it's needed. Thus e.g. when I bump the version from
3303 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3304 * The C source files now include boilerplate legalese and documentation
3305 at the head of each file (just as the Lisp source files already did).
3306 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3307 to his page will be replaced with a link to his new CLiki service.
3309 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3311 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3312 * The build process now looks for GNU make under the default name "gmake",
3313 instead of "make" as it used to. If GNU make is not available as "gmake"
3314 on your system, you can change this default behavior by setting the
3315 GNUMAKE environment variable.
3316 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3317 can build without error under CMU CL.
3319 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3321 * There is now a partial SBCL user manual (with some new text and some
3322 text cribbed from the CMU CL manual).
3323 * The beginnings of a profiler have been added (starting with the
3324 CMU CL profiler and simplifying and cleaning up). Eventually the
3325 main interface should be through the TRACE macro, but for now,
3326 it's still accessed through vaguely CMU-CL-style functions and macros
3327 exported from the package SB-PROFILE.
3328 * Some problems left over from porting CMU CL to the new
3329 cross-compilation bootstrap process have been cleaned up:
3330 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3331 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3332 part when getting rid of the compiler *BACKEND* stuff.)
3333 ** The value of *NULL-TYPE* was screwed up, because it was
3334 being initialized before the type system knew the final
3335 definition of the 'NULL type. This screwed up several key
3336 optimizations in the compiler, causing inefficiency in all sorts
3337 of places. (I found it because I wanted to understand why
3338 GET-INTERNAL-RUN-TIME was consing.)
3339 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3340 PROCLAIM FTYPE information. Unfortunately this broke binary
3341 compatibility again, since now the forms output by DEFGENERIC
3342 to refer to functions which didn't exist in 0.6.3.
3343 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3344 can use the new (as of 0.6.3) transform for SOME into MAP into
3346 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3347 values to (INTEGER 0 1000000), so that the time code will no longer
3348 occasionally get blown up by Linux returning 1000000 microseconds
3349 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3350 its output conform to the ANSI spec. (Alas, this makes its output
3351 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3352 * A full call to MAP NIL with a single sequence argument no longer conses.
3353 * fixes to problems pointed out by Martin Atzmueller:
3354 * The manual page no longer talks about multiprocessing as though
3355 it were currently supported.
3356 * The ILISP support patches have been removed from the distribution,
3357 because as of version 5.10.1, ILISP now supports SBCL without us
3358 having to maintain patches.
3359 * added a modified version of Raymond Toy's recent CMU CL patch for
3360 EQUALP comparison of HASH-TABLE
3362 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3364 * The system still can't cross-compile itself with
3365 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3366 entails), but at least it can compile more of itself that way
3367 than it used to be able to, and various buglets which were uncovered
3368 by trying to cross-compile itself that way have now been fixed.
3369 * This release breaks binary compatibility again. This time
3370 at least I've incremented the FASL file format version to 2, so that the
3371 problem can be detected reliably instead of just causing weird errors.
3372 * various new style warnings:
3373 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3374 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3375 ** using the lexical binding of a variable named in the *FOO* style
3376 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3377 as specified by ANSI.
3378 * *RANDOM-STATE* is no longer automatically initialized from
3379 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3380 default behavior of the system is to repeat its behavior every time
3381 it's run. If you'd like to change this behavior, you can always
3382 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3383 old convention there was no comparably easy way to get the system to
3384 repeat its behavior every time it was run.
3385 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3386 FTYPE declarations has been removed, in favor of their ANSI
3387 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3388 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3389 the types of their sequence arguments can be determined at compile time.
3390 This is done through a new open code expansion for MAP which eliminates
3391 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3392 argument types can be determined at compile time.
3393 * The optimizer now transforms COERCE into an identity operation when it
3394 can prove that the coerced object is already of the correct type. (This
3395 can be a win for machine generated code, including the output of other
3396 optimization transforms, such as the MAP transform above.)
3397 * Credit information has been moved from source file headers into CREDITS.
3398 * Source file headers have been made more standard.
3399 * The CASE macro now compiles without complaining even when it has
3402 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3404 * (Note that the way that the PCL macroexpansions were rewritten
3405 to accommodate the change in DEFGENERIC below breaks binary
3406 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3407 not run under sbcl-0.6.2. Once we get out of alpha releases,
3408 i.e. hit release 1.0.0, we'll probably try to maintain binary
3409 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3410 and sbcl-1.4.4. Until then, however, it might be fairly common
3411 for maintenance releases to break binary compatibility.)
3412 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3414 * The reporting of unhandled signals has been changed to print some
3415 explanatory text as well as the report form. (Previously only
3416 the report form was printed.)
3417 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3418 it defines, so that the compiler no longer issues undefined function
3419 warnings for compiled-but-not-yet-loaded generic functions.
3420 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3421 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3422 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3423 section of the ANSI spec are supported.
3424 * The "" nickname for the "KEYWORD" package has been removed.
3425 The reader still handles symbol tokens which begin with a package marker
3426 as keywords, but it doesn't expose its mechanism for doing so in the
3427 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3428 * The system now issues STYLE-WARNINGs for contradictory TYPE
3429 proclamations. (Warnings for contradictory FTYPE proclamations would
3430 be nice too, but those can't be done usefully unless the type system
3431 is made smarter about FUNCTION types.)
3432 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3433 systematized, so that "*target-*.lisp is supposed to exist only on the
3434 target and imply that there's a related file which exists on the
3435 host, and *host-*.lisp is supposed to exist only on the host and imply
3436 that there's a related file which exists on the target. This involves a
3437 lot of renaming. Hopefully the acute confusion caused by the renaming
3438 will be justified by the reduction in chronic confusion..
3439 ** runtime-type.lisp -> early-target-type.lisp
3440 ** target-type.lisp -> late-target-type.lisp
3441 ** early-host-format.lisp -> early-format.lisp
3442 ** late-host-format.lisp -> late-format.lisp
3443 ** host-error.lisp -> misc-error.lisp
3444 ** early-error.lisp -> early-target-error.lisp
3445 ** late-error.lisp -> late-target-error.lisp
3446 ** host-defboot.lisp -> early-defboot.lisp
3447 ** code/misc.lisp -> code/target-misc.lisp
3448 ** code/host-misc.lisp -> code/misc.lisp
3449 ** code/numbers.lisp -> code/target-numbers.lisp
3450 ** code/early-numbers.lisp -> numbers.lisp
3451 ** early-host-type.lisp -> early-type.lisp
3452 ** late-host-type.lisp -> late-type.lisp
3453 ** host-typep.lisp -> typep.lisp
3454 ** load.lisp -> target-load.lisp
3455 ** host-load.lisp -> load.lisp
3456 ** host-disassem.lisp -> disassem.lisp
3457 ** host-insts.lisp -> insts.lisp
3458 ** byte-comp.lisp -> target-byte-comp.lisp
3459 ** host-byte-comp.lisp -> byte-comp.lisp
3460 ** host-signal.lisp -> signal.lisp
3461 ** host-defstruct.lisp -> defstruct.lisp
3462 ** late-target-type.lisp -> deftypes-for-target.lisp
3463 Furthermore, several other previously target-only files foo.lisp (e.g.
3464 hash-table.lisp and random.lisp) have been split into a target-and-host
3465 foo.lisp file and a target-only target-foo.lisp file, with their key type
3466 definitions in the target-and-host part, so that the cross-compiler will
3467 know more about target types.
3468 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3469 gone away. In their place are various *BACKEND-FOO* variables
3470 corresponding to the slots of the old structure.
3471 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3472 into the target SBCL has been fixed.
3473 * The chill.lisp system for loading cold code into a running SBCL
3475 * Support for the CMU CL "scavenger hook" extension has been removed.
3476 (It was undocumented and unused in the CMU CL sources that SBCL was
3477 derived from, and stale in sbcl-0.6.1.)
3478 * Various errors in the cross-compiler type system were detected
3479 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3480 (enabling various consistency checks). Many of them were fixed,
3481 but some hard problems remain, so the compiler is back to
3482 running without *TYPE-SYSTEM-INITIALIZED* for now.
3483 * As part of the cross-compiler type system cleanup, I implemented
3484 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3485 * I have started adding UNCROSS calls throughout the type system
3486 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3487 on cross-compiler input files will be able to go away, and various
3489 * CONSTANTP now returns true for quoted forms (as explicitly required
3492 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3494 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3495 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3496 (MAKE-LIST -1). In the longer term, it ought to become true
3497 that declarations are assertions even at SAFETY 1. For now, it's not
3498 quite true even at SAFETY 3, but it's at least more nearly true..
3499 (Note that this change seems to increases the size of the system by
3500 O(5%) and to decrease the speed of the compiler by 20% or more.)
3501 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3502 for the problem of printing dozens of lines of distracting information
3503 about low-level system machinery as part of the top stack frame
3504 on entry to the debugger when an undefined function was called.
3505 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3506 is not reset to COMMON-LISP-USER.
3507 * Compilation of stuff related to dyncount.lisp has been made conditional
3508 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3509 smaller. The various dyncount-related symbols have been moved into
3510 a new "SB-DYNCOUNT" package.
3511 * tty-inspect.lisp has been renamed to inspect.lisp.
3512 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3513 feature has gone away. (When we eventually port to other flavors of
3514 libc and/or Unix, we'll try to make the differences between flavors
3515 invisible at the user level.)
3516 * Various other *FEATURES* tags, and/or their associated conditionals,
3517 have been removed if obsolescent, or given better documentation, or
3518 sometimes given more-mnemonic names.
3520 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3522 * tidied up "make.sh" script
3523 * tidied up system directory structure
3524 * better "clean.sh" behavior
3525 * added doc/FOR-CMUCL-DEVELOPERS
3526 * many many small tweaks to output format, e.g. removing possibly-confusing
3527 trailing #\. character in DESCRIBE-INSTANCE
3528 * (EQUALP #\A 'A) no longer signals an error.
3529 * new hashing code, including EQUALP hashing
3530 * tidied up Lisp initialization and toplevel
3531 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3532 * command line argument processing
3533 * added POSIX-GETENV function to deal with Unix-ish environment variables
3534 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3535 terminating SBCL on EOF
3536 * non-verbose GC by default
3537 * There is no more "sbcl" shell script; the sbcl file is now the C
3538 runtime executable (just like CMU CL).
3539 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3541 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3542 symbol and package information as Lisp native symbol and package objects
3543 instead of strings naming symbols and strings naming packages. This way,
3544 whenever packages are renamed (as in warm init), debug information is
3545 transformed along with everything else.
3546 * tweaked the optimization policy declarations which control the building
3547 of SBCL itself. Now, among other things, the system no longer saves
3548 source location debugging information. (This helps two problems at once
3549 by reducing SBCL size and by keeping SBCL from trying to look for its
3550 sources -- which may not exist -- when reporting errors.)
3551 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3552 debugging and testing purposes
3553 * cleaned up printing, making the printer call PRINT-OBJECT for
3554 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3555 methods, giving nearly-ANSI behavior
3556 * converted almost all special variables to use *FOO* naming convention
3557 * deleted PARSE-TIME functionality, since it can be done portably
3558 * moved some files out of cold init into warm init
3559 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3561 * regularized formatting of source files
3562 * added an install.sh script
3563 * fixed ridiculous memory usage of cross-compiler by making
3564 compiler/alloc.lisp not try to do pooling unless it can hook
3565 itself into the GC of the cross-compilation host. Now the system
3566 builds nicely on my old laptop.
3567 * added :SB-ALLOC in target-features.lisp-expr
3568 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3569 was not implemented)
3570 * re-did condition handling and note reporting in the compiler. Notes
3571 are no longer handled by signalling conditions. Style warnings
3572 and warnings are handled more correctly and reported in such a way
3573 that it's easy to find one or the other in your output (so that you
3574 can e.g. figure out which of many problems caused COMPILE-FILE to
3576 * changed the severity of several compiler warnings from full WARNING
3577 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3578 compiler note reporting so that it doesn't use the condition system
3579 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3580 COMPILE-FILE command)
3581 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3582 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3583 no longer supported, since their implementation was deeply intertwingled
3584 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3585 * removed bogus "support" for compiler macros named (SETF FOO), and
3586 removed the compiler macro for SETF INFO (but only after making a fool
3587 of myself on the cmucl-imp mailing list by posting a bogus patch for
3588 DEFINE-COMPILER-MACRO..)
3589 * Compiled files containing forms which have side effects on the Lisp
3590 reader (such as DEFPACKAGE forms) are now handled more correctly.
3591 (Compiler queuing of top level lambdas has been suppressed by setting
3592 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3593 * deleted various currently-unused source files, e.g. gengc.lisp. They
3594 may be added back at some point e.g. when porting to other architectures,
3595 but until they are it's distracting to distribute them and to try to
3597 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3598 there were so many of them they're just distractions, and UNCROSS is
3599 known to be able to handle the current sources
3600 * moved PROFILE functionality into TRACE, so that it will be clear
3601 how the wrapping and unwrapping of functions when you profile them
3602 interacts with the wrapping and unwrapping of functions when you
3603 trace them. (Actually, the functionality isn't there yet, but at least
3604 the interface specification is there. Hopefully, the functionality will
3605 arrive with some future maintenance release.)
3606 * removed host-oops.lisp
3607 * changed signature of QUIT function to allow UNIX-CODE argument
3608 * fixed READ-SEQUENCE bug
3609 * tweaked verbose GC output so that it looks more like the progress
3610 output that ANSI specifies for functions like LOAD
3611 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3612 * added <http://sbcl.sourceforge.com> to the banner information printed by