1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
3 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
4 be used for bundling the runtime and the core file into one
5 executable binary. This feature is not currently supported on all SBCL
6 platforms. (thanks to James Bielman and NIIMI Satoshi)
7 * minor incompatible change: the method by which SBCL finds its
8 contributed modules has changed; it no longer relies on symbolic
9 links from an $SBCL_HOME/systems directory, but searches directly
10 in the subdirectories of $SBCL_HOME.
11 * enhancement: the dynamic heap size on the Linux/PPC platform is
13 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
14 certain complicated string types. (reported by Gary King)
15 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
16 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
17 * fixed bug: DOCUMENTATION on structures no longer signals an
18 error if no structure type of the right name exists. (reported by
20 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
21 applicable non-standard primary or :AROUND method on
22 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
23 no longer cause unbound variable errors. (reported by Kevin Reid)
25 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
26 * new platform: experimental support for the Windows operating
27 system has been added. (thanks to Alastair Bridgewater)
28 * fixed several bugs in and robustified the PPC FFI (including
29 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
30 * bug fix: allow non-simple string symbol names (reported by Paul
32 * bug fix: interrupt handling on NetBSD (thanks to Richard M
34 * bug fix: saving a core corrupted callbacks on x86/x86-64
35 * bug fix: closed a loophole in metacircularity detection and
36 grounding in the PCL implementation of CLOS.
37 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
38 (reported by tomppa on #lisp)
39 * bug fix: clear the direction flag on Lisp -> C transitions, as
40 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
41 (reported by Andrej Grozin and Hendrik Maryns)
42 * optimization: major improvements to GC efficiency on GENCGC platforms
43 * optimization: faster implementation of EQUAL
44 * optimization: emit more efficient opcodes for some common
45 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
46 * optimization: several other minor code-generation improvements on x86-64
47 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
49 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
50 * minor incompatible change: (SETF CLASS-NAME) and (SETF
51 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
52 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
53 by ANSI as a generic function, and for consistency (SETF
54 GENERIC-FUNCTION-NAME) is treated likewise.
55 * fixed bug #233.b: make constraint propagation notice when a variable
56 value is changed after it is referenced but before it is used
57 * fixed bug #296: no more arbitrary behaviour from filenames with
58 odd characters as --load arguments. (reported by Adam Warner)
59 * fixed bug #390: :CHARACTER-SET pathname components now work as
60 expected. (reported by Tim Daly Jr)
61 * fixed bug #391: complicated :TYPE intersections in slot
62 definitions no longer cause an error in PCL internals.
63 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
64 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
65 returning the number of octets which would be written to the
66 file-stream. (thanks to Robert J. Macomber)
67 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
68 arguments to the internal stream functions. (thanks to David
70 * bug fix: stack exhaustion is now handled on the x86/SunOS
71 platform. (thanks to Daisuke Homma)
72 * bug fix: more accurate ROOM results on GENCGC platforms
73 * optimization: improved type inference for arithmetic-for
74 index variables in LOOP
75 * optimization: faster floating-point SQRT on x86-64
76 * fixed some bugs revealed by Paul Dietz' test suite:
77 ** DOCUMENTATION returns NIL instead of "" for method combinations
78 that don't have a docstring
80 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
81 * minor incompatible change: (SETF CLASS-NAME) and (SETF
82 GENERIC-FUNCTION-NAME) are no longer generic functions, and
83 therefore are not customizeable by user code (as seems to be at
84 least permitted and maybe required by AMOP). As a consolation,
85 however, the SBCL implementation of these functions now calls
86 REINITIALIZE-INSTANCE as specified by AMOP.
87 * bug fix: slot-definition documentation is propagated as per ANSI
88 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
89 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
90 argument, as do eslotd initialization methods. (from Pascal
91 Costanza's "Closer" project)
92 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
93 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
95 * bug fix: it is now possible to have more than one subclass of
96 STANDARD-GENERIC-FUNCTION without causing stack overflow.
97 (reported by Bruno Haible, Pascal Costanza and others)
98 * bug fix: the dependent update protocol now works for generic
99 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
101 * bug fix: condition-class instances corresponding to
102 DEFINE-CONDITION forms are now created eagerly. (reported by
103 Kalle Olavi Niemitalo on comp.lang.lisp)
104 * bug fix: floating point printing is more accurate in some
105 circumstances. (thanks to Simon Alexander)
106 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
107 merged with *DEFAULT-PATHNAME-DEFAULTS*.
108 * bug fix: callbacks on OS X now preserve stack-alignment.
109 (thanks to Cyrus Harmon)
110 * enhancement: the x86-64 disassembler is much better at
111 disassembling SSE instructions. (thanks to Lutz Euler)
112 * enhancement: CHECK-TYPE on a variable now allows type-inference
113 (thanks to Vincent Arkesteijn)
114 * optimization: improved performance of EUC-JP external format.
115 (thanks to NIIMI Satoshi)
116 * optimization: performance improvements to IO on file streams of
117 :ELEMENT-TYPE CHARACTER
118 * optimization: much faster memory allocation on x86-64
119 * optimization: faster garbage collections (latency and throughput)
121 * optimization: faster bignum subtraction and division on x86 and x86-64
122 (thanks to Lutz Euler)
123 * optimization: more accurate type-inference for loops that use a
124 floating point index variable or a negative step.
126 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
127 * bug fix: add a workaround to SBCL looping infinitely at startup on
128 Linux kernels with apparently buggy implementations of personality().
129 (thanks to Svein Ove Aas)
130 * bug fix: Unicode symbols are correctly printed in LDB backtraces
131 (thanks to David Lichteblau)
132 * bug fix: local bindings shadow global symbol macros for
133 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
134 * optimization: non-open coded uses of numeric comparison operators
135 (e.g. >) no longer cons when called with more than one parameter
136 on platforms supporting dynamic-extent allocation.
137 * enhancement: saving cores with foreign code loaded is now supported
138 on MIPS/Linux in addition to the previously supported platforms.
139 * bug fix: FIND-RESTART now tests for activity, not applicability when given
140 a restart object as identifier. (reported by Helmut Eller for CMUCL)
141 * bug fix: division by zero in sb-sprof when no samples were collected
142 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
143 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
144 (thanks to Alexander Kjeldaas)
145 * new feature: ignoring the return values of destructive functions
146 like NREVERSE signals a compile-time style-warning.
147 (thanks to Kevin Reid)
148 * new feature: the alignment of alien structure fields can be
149 explicitly specified. (thanks to Cyrus Harmon)
150 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
151 and dump core on SIGQUIT
153 ** incompatible change: threads do not inherit values of specials
154 from their parents (see manual)
155 ** bug fix: threads stacks belonging to dead threads are freed by the
156 next exiting thread, no need to gc to collect thread stacks anymore
157 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
158 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
159 ** bug fix: dynamic variable and thread start related gc lossage
160 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
162 * fixed some bugs revealed by Paul Dietz' test suite:
163 ** SUBTYPEP is slightly more accurate on heinously complicated
164 CONS types where some of the members have uncertain (in the
165 NIL, NIL sense) type relationships to each other.
167 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
168 no longer purified unless :PURIFY T is explicitly specified.
169 ** Non-purified cores are significantly smaller than before
171 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
172 * new feature: timers based on Zach Beane's excellent timer package
173 * added support for the following external formats: koi8-u,
174 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
175 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
176 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
177 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
178 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
179 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
180 * incompatible change: a threaded SBCL will no longer revert to
181 non-threaded mode on non-NPTL systems, but refuse to start entirely.
182 * bug fix: interrupts are disabled until startup is complete; no
183 more sigsegvs when receiving a signal to soon
184 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
186 * bug fix: add a workaround for the memory randomization features in
187 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
188 workaround will only be in effect on systems with the proc filesystem
190 * bug fix: printing objects of type HASH-TABLE signals a
191 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
193 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
194 have been running for over 50 days. (reported by Gilbert Baumann)
195 * bug fix: the logic for getting names of functions gets less
196 confused when confronded with alternate-metaclass
197 funcallable-instances. (reported by Cyrus Harmon)
198 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
199 consistent, even on internal alternate-metaclass objects.
200 * bug fix: flush closure information collected by physical
201 environment analysis prepass before the main pass. (bug reported
203 * bug fix: compiler pack phase does not modify a hashtable, which is
204 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
205 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
206 may be expanded. (reported by Alexander <malishev> on c.l.l)
207 * bug fix: no more style-warnings when compiling code using LAST
208 under high-SPEED low-SPACE settings. (reported by David Wallin
210 * fixed some incompatibilities between SBCL's MOP and the MOP
212 ** the METAOBJECT class is now implemented;
213 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
214 STANDARD-OBJECT, as required;
215 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
216 now compatible (as required by VALIDATE-SUPERCLASS); there
217 remains a consistency requirement over the presence or absence
218 of the FUNCTION class in the superclass list of finalized
219 classes; see the manual for more details;
220 ** the :AROUND method for COMPUTE-SLOTS on
221 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
222 requested slot ordering.
224 ** bug fix: parent thread now can be gc'ed even with a live
226 ** bug fix: binding a special with PROGV to no value is not
227 the same as that symbol not having been bound (thanks to
229 ** bug fix: binding specials is thread safe (thanks to
231 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
232 and SB-SPROF do not die with 'no handler for signal XX in
233 interrupt_handle_now(..)' anymore
234 ** bug fix: WITH-TIMEOUT works with multiple running threads
235 * fixed some bugs revealed by Paul Dietz' test suite:
236 ** ENSURE-GENERIC-FUNCTION should take a method class object for
237 the :method-class keyword argument.
239 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
240 * new port: the Solaris operating system on x86 processors is now
241 mostly supported, though some rough edges in the environment
242 remain. (thanks to Daisuke Homma)
243 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
244 space for its heap. (thanks to Thiemo Seufer)
245 * optimization: SBCL on MIPS platforms now supports dynamic-extent
246 closures. (thanks to Thiemo Seufer)
247 * minor incompatible change: eof selects abort in the debugger.
248 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
249 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
251 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
252 deprecated, and will go away in a future revision of SBCL.
253 * minor incompatible change: GC-ON and GC-OFF are no longer
254 implemented with a counter, it does not matter how many times gc
255 is switched on or off
256 * bug fix: discriminating functions for generic function classes
257 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
258 make invalid assumptions about method precedence order. (reported
260 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
261 platforms. (thanks to Thiemo Seufer)
262 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
263 * bug fix: correct pathname printing with printer escaping is on.
264 (thanks to Kevin Reid)
265 * bug fix: complex VOP definitions in "user-space" no longer trigger
266 package locks. (reported by Zach Beane)
267 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
268 for funcallable-instance functions, and is no different from
270 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
271 characters in the current readtable. (reported by Nicholas Neuss)
272 * bug fix: SUBTYPEP on various CONS types returns more a more
273 accurate acknowledgment of its certainty.
274 * optimizations: REMOVE-DUPLICATES now runs in linear time on
275 lists in some cases. This partially fixes bug 384.
276 * flush all standard streams before prompting in the REPL and the
278 * bug fix: signal handling and triggering gc do not conflict
279 directly anymore, in particular a high frequency sb-sprof does
280 not prevent gc from running
281 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
282 approximation for timezone and DST information between the
283 universal time epoch and the smallest negative 32-bit time_t.
284 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
285 year 1899 so long as the encoded time is non-negative.
286 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
287 the correct answer for non-constant rotate amounts when performing
288 an inline 32-bit rotation.
290 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
291 there is only one thread in the session
292 ** bug fix: memory leak for streams created in one thread and
293 written to in another
294 ** bug fix: lockup when compiled with gcc4
295 ** bug fix: race that allows the gc to be triggered when gc is
297 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
298 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
300 ** bug fix: finished threads are gc'ed properly
301 * fixed some bugs revealed by Paul Dietz' test suite:
302 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
303 has arguments with a different set of applicable methods from
304 the orignal arguments.
305 ** The type error thrown by MAP now has a correct expected-type
307 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
308 name a compiled function.
309 ** Three MISC tests where a large bignum was improperly coerced to
310 a float that couldn't represent that bignum during type
311 derivation were fixed.
312 ** SUBTYPEP can now handle types involving the negation of a
313 list-form FUNCTION type.
314 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
315 as (COMPLEX (AND RATIO (NOT FIXNUM))).
316 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
318 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
319 * New feature: Experimental support for bivalent streams: streams
320 opened with :element-type :default now allow character and binary
321 (unsigned-byte 8) I/O
322 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
323 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
325 * bug fix: no more highly sporadic "couldn't check whether ~S is
326 readable" when reading a stream and an interrupt hits in the middle
327 of a select system call
328 * compiler better recognizes complex arrays (reported by Eduardo
330 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
331 for variables on SBCL built with linkage-tables. (reported by Luis
333 * various error reporting improvements.
334 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
335 (Thanks to Thiemo Seufer)
336 * bug fix: nested reader invokations work correctly
337 * bug fix: it is possible to have simultaneous references to foreign
338 code and foreign data with the same name.
340 ** added x86-64 support
341 ** incompatible change: the threading api now works with thread
342 objects instead of thread ids
343 ** bug fix: threads are protected from signals and interruption when
344 starting up or going down
345 ** bug fix: a race where an exiting thread could lose its stack to gc
346 ** bug fix: don't halt on infinite error in threads if possible
347 ** fixed numerous gc deadlocks introduced in the pthread merge
348 ** bug fix: fixed thread safety issues in read and print
349 ** bug fix: debugger doesn't hang on session lock if interrupted at
350 an inappropriate moment
351 ** bug fix: run-program is now thread safe(r)
352 ** bug fix: inner with-recursive-lock no longer releases the mutex
353 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
354 * fixed some bugs revealed by Paul Dietz' test suite:
355 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
356 EXPECTED-TYPE slots filled.
357 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
359 ** MAKE-CONDITION accepts classes as type-designators.
360 ** COMPILE may never return NIL.
361 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
362 range before calling Unix time functions
364 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
365 * numerous signal handling fixes to increase stability
366 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
367 * minor incompatible change: we now correctly canonize default
368 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
369 as per the MOP, rather than the historical (INITARG INITFUNCTION
370 INITFORM). (reported by Bruno Haible)
371 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
372 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
373 for more information.
374 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
375 pathname is a directory pathname.
376 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
377 * dynamic space size on PPC has been increased to 768Mb. (thanks to
379 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
380 :METACLASS argument in addition to a class name. (reported by
381 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
382 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
383 (reported by Bruno Haible; patch from Gerd Moellmann)
384 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
386 * bug fix: more cleanups to the floating point exception handling on
387 x86-64 (thanks to James Knight)
388 * bug fix: the compiler does not try to constant fold calls of
389 COERCE and BIT-* functions when they return freshly constructed
390 objects. (reported by Nikodemus Siivola and Paul Dietz)
391 * optimization: improved the MIPS versions of generic (in the
392 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
393 * optimization: direct conversion of (unsigned-byte 32) to floats on
394 the PowerPC platform.
395 * optimization: structure instances with raw slots now use less
396 memory, and probably show better memory locality. (thanks to
398 * optimization: DYNAMIC-EXTENT declarations for lists and closures
399 are treated as requests for stack allocation on the x86-64,
400 Alpha-32, PPC and SPARC platforms.
401 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
402 the wrong file descriptor; implementation of SOCKET-OPEN-P.
403 (thanks to Tony Martinez)
405 ** gcing a dead thread can no longer lead to lockups
406 ** threads block signals until they are set up properly
407 ** errno is no longer shared by threads
408 ** interrupt-thread restores the eflags register on x86
409 ** fixed some lockups due to gc/thread interaction
410 * fixed some bugs revealed by Paul Dietz' test suite:
411 ** invalid dotted lists no longer raise a read error when
413 ** signal an error if a symbol that names a declaration is used
414 as the name of a type, or vice versa
415 ** allow using the (declare (typespec var*)) abbreviation for
416 (declare (type typespec var*)) with all type specifiers
417 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
418 ** always return NIL from PPRINT-POP when OBJECT is NIL
419 ** don't signal errors when pretty-printing malformed LABELS,
420 FLET or MACROLET forms
421 ** declarations in a DOLIST body are also in scope for the
423 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
425 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
428 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
429 * fixed cross-compiler leakages that prevented building a 32-bit
430 target with a 64-bit host compiler.
431 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
432 opened with :IF-EXISTS :OVERWRITE.
433 * fixed bug 281: error for an invalid qualifier in a short-form method
434 combination method is not signalled until the faulty method is called.
435 * bug fix: iteration variable type inferrer failed to deal with open
436 intervals. (reported by Alan Shields)
437 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
438 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
439 or more runtime options were provided to the sbcl binary.
440 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
441 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
442 * optimization: there's now a fast-path for fixnum arguments in the
443 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
444 * optimization: the code generated on x86-64 is more compact thanks
445 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
446 * bug fix: floating-point exception handling now partly works on
447 x86-64. (Thanks to James Knight)
448 * improvement to the MIPS backend from Thiemo Seufer: C-style
449 64-bit long long arguments and return values to alien functions
451 * fixed some bugs revealed by Paul Dietz' test suite:
452 ** the type-error signalled from WARN has a filled-in DATUM slot.
453 ** the type-error required when a stream is not associated with
454 a file has the stream as its datum.
455 ** type-errors on single-floats on x86-64 no longer have
456 :INVALID-OBJECT as the datum
457 ** the type-errors signalled for invalid function names now have
458 a correct expected type
459 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
460 for structures defined with a :TYPE. Documentation strings for
461 typed structures are no longer immediately discarded
462 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
463 broadcast streams and synonym streams. FILE-LENGTH now also works
464 on broadcast streams.
466 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
467 * incompatible change: the --noprogrammer option, deprecated since
468 version 0.7.5, has been removed. Please use the equivalent
469 --disable-debugger option instead.
470 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
472 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
473 inoperational for a while now) has been completely removed.
474 * null lexical environments are now printed as #<NULL-LEXENV>,
475 significantly reducing the amount of clutter in typical backtraces.
476 * documentation on weak pointers, finalization, and after GC hooks
477 has been added to the manual.
478 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
479 as other specialized array types, is much faster. SUBSEQ and
480 COPY-SEQ on such arrays have also been sped up.
481 * optimization: EQL is now more efficient when at least other argument
482 is known to be of type (OR FIXNUM (NOT NUMBER)).
483 * fixed bug: the runtime is now less vulnerable to changes in the
484 size of the SBCL object on OS X, and virtual memory is reserved for
485 all spaces that need to be at a fixed address.
486 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
487 * fixed bug: finalizers and after GC hooks that cause consing are now
489 * fixed bug: compiler error messages and summaries are now printed to
490 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
491 * fixed inference of the upper bound of an iteration variable.
492 (reported by Rajat Datta).
493 * fixed bug 211e: calling local functions with duplicated constant
494 keyword argument no longer causes a bogus style warning about an
496 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
497 ftype declaration to be disregarded. (reported by Dave Roberts)
498 * fixed bug 373: caused by erronous compilation of references to alien
499 variables in the runtime on ppc/darwin.
500 * fixed bug 376: CONJUGATE type deriver.
501 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
502 (reported by Baughn on #lisp)
503 * a cross-compiler bug on non-x86 platforms has been identified and
504 fixed. (thanks to Bruno Haible)
505 * improvements to the MIPS runtime code for increased stability.
506 (thanks to Thiemo Seufer)
507 * increased the maximimum compact environment size to allow
508 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
509 * partial workaround for bug 135: don't name the function we're
510 calling for hairy cases of EVAL, so as not to accumulate one environment
511 entry per call to EVAL. (reported by Kevin Reid)
512 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
513 * optimization: type testing for non-vector arrays should be faster.
514 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
516 * bug fix: setting 31st element of a bit vector to zero did not work
518 * bug fix: redefining a class definition which failed due to a
519 previous accessor / function clash now works (but see BUGS entry
520 #380 for more problems in this area). (thanks to Zach Beane)
521 * the long form of DEFINE-METHOD-COMBINATION disables method group
522 checking when given a single method group with pattern *. (thanks
524 * on x86 compiler supports stack allocation of results of simple
525 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
526 * support for building and running on Mac OS X 10.4 "Tiger" has been added
527 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
528 * fixed some bugs related to Unicode integration:
529 ** the restarts for recovering from input and output encoding
530 errors only appear when there is in fact such an error to
532 * fixed some bugs revealed by Paul Dietz' test suite:
533 ** MISC.549 and similar: late transformation of unsafe type
534 assertions into derived types caused unexpected code
536 ** SCALE-FLOAT type deriver is less wrong.
537 ** type derivers for EXP, LOG and similar functions compute result
538 types for complex arguments better.
539 ** (MISC.563) CONJUGATE type deriver works for very restricted
541 ** out-of-line type testers for character strings are available.
542 ** EQUAL compiler transform understands specialness of objects
544 ** accessing double-floats stored on the stack now works on x86-64.
545 ** debugger internals could sometimes create invalid lispobjs,
546 resulting in GC crashes.
547 ** MISC.548: type check weakening can convert required type into
549 ** initialization forms for bindings are not in scope of free special
552 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
553 * incompatible change: thread support for non-NPTL systems has
554 been removed - locking is buggy and unreliable. A threaded
555 SBCL build will now warn at startup and refuse to create
556 new threads, unless futex support is detected in the kernel
557 * incompatible change: the top level REPL now has only an ABORT
558 restart associated with it, not TOPLEVEL and ABORT as it used to.
559 TOP and TOPLEVEL are now available as debugger commands for
560 returning to the top level.
561 * incompatible change: forms evaluated in the REPL now use the
562 global optimization policy.
563 * incompatible change: user- and system-initialization files are
564 no longer processed with LOAD, but by READ and EVAL; hence the
565 global optimization policy, startup package, readtable, etc,
567 * The .fasl file format number has been incremented because of
568 various incompatible changes.
569 * internal entry point details and argument counts no longer appear
570 in backtraces unless explicitly requested by setting
571 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
572 * built-in and standard functions no longer have names like "top
573 level local call to FOO".
574 * fixed bug 32: functions defined in non-null lexical environments
575 now have more legible printed representation
576 * fixed bug 33: functions defined in non-null lexical environemnts
577 are now more amenable to inspection by INSPECT.
578 * workaround for bug 354: XEPs no longer appear in backtraces unless
579 explicitly requested.
580 * fixed bug: receiving the signal which results from attempting to
581 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
582 SIGBUS on other *BSDs) on architectures where the C stack is also
583 the Lisp stack (x86 and x86-64) from foreign code no longer leads
584 to debugger confusion or wild instability. (reported by Cheuksan
586 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
587 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
588 (reported by Lutz Euler)
589 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
590 compiled code. (reported by Paul Dietz)
591 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
592 specialized parameter to a method to something that is not TYPEP
593 the specializer is now possible.
594 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
595 face of package deletion.
596 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
597 pathname. (reported by David Tolpin on c.l.l)
598 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
599 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
600 STRING-STREAM and FILE-STREAM, respectively.
601 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
602 than previously; in addition, its test suite is now run on build.
603 (thanks to Luke Gorrie)
604 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
605 * optimization: added a immediate representation for single-floats
607 * fixed some bugs revealed by Paul Dietz' test suite:
608 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
609 ** attempting to create a package with a colliding nickname causes
610 correctable errors to be signalled.
611 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
612 ** several x86-64 backend bugs related to sign-extension of immediate
615 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
616 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
617 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
619 * fixed loading of multiply forward-referenced layouts.
620 (thanks to Cheuksan Wang)
621 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
622 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
623 (see COMPILE-FILE documentation for details of :PRINT option)
624 * fixed bugs 19 and 317: fixed-format floating point printing is
625 more accurate. This also fixes a bug reported by Adam Warner
626 related to the ~@F format directive.
627 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
629 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
630 dispatch table if the from-char is a dispatch macro character.
631 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
632 by the wordsize no longer ignore the last word. (reported by Lutz
634 * fixed bug in type checking of dynamic-extent variables. (reported
636 * optimization: sequence traversal functions use their freedom to
637 coerce function designators to functions.
638 * optimization: code with many calls to CLOS methods specialized on
639 CLOS classes has had redundant type checks removed.
640 * optimization: added declarations to speed up operations that access
641 the internal character database (for example STRING-UPCASE)
642 * optimization: comparison operations between floats and sufficiently small
643 fixnums no longer create extra rationals
644 * fixed some bugs related to Unicode integration:
645 ** portions of multibyte characters at the end of buffers for
646 character-based file input are correctly transferred to the
647 start of the buffer at the next read.
648 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
649 passing it through to OPEN.
650 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
651 argument given to internal calls to OPEN.
652 ** the built-in comment readers (introduced by character sequences
653 ";" and "#|") are more forgiving to encoding errors; they will
654 STYLE-WARN and then attempt to resync the stream at a character
655 boundary. (thanks to Teemu Kalvas)
656 * fixed some bugs in the x86-64 port:
657 ** Negative short int return values from c-calls are sign-extended
659 ** The stack is aligned to 16-bytes for c-calls, as required by
660 the ABI. (thanks to Cheuksan Wang)
661 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
662 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
663 for enums. (thanks to Vincent Arkesteijn)
664 ** Multiple small optimizations and bugfixes for floating point
666 * fixed some bugs revealed by Paul Dietz' test suite:
667 ** Space, Tab, Linefeed, Return and Page have the invalid
668 secondary constituent character trait.
669 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
671 ** WITH-INPUT-FROM-STRING should only update the index place on
673 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
675 ** Bit-array operations (BIT-AND and similar) worked incorrectly
676 with one-dimensional arrays with fill pointers.
677 ** TYPE-OF failed on a complex with an integer realpart and a
679 ** compiler failure during type inference for the code of form
680 (IF (EQL X (THE ...)) ...) (MISC.535).
682 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
683 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
684 platform. The port supports 61-bit fixnums, large memory spaces
685 and reloading shared object files.
686 * enhancement: saving cores with foreign code loaded is now
687 supported on ppc/Darwin in addition to the previously supported
689 * enhancement: the statistical profiler now walks deeper into the
690 call stack for more meaningful call-graphs and accrued time
691 reports (x86/x86-64 only). It also now reports time spent in
693 * enhancement: it is now possible to trace most individual methods
694 of a generic function in addition to tracing the generic function
696 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
698 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
699 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
700 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
701 produces an error. (thanks to Vincent Arkesteijn)
702 * bug fix: NAMESTRING on pathnames with :WILD components in their
703 directories works correctly. (thanks to Artem V. Andreev)
704 * fixed bug 125: compiler preserves identity of closures. (reported
706 * bug fixed: functions with &REST arguments sometimes failed with
707 "Undefined function" when compiled with (DEBUG 3). (reported by
709 * bug fix: overflow during compiling of setting element of a bit
710 vector with constant index and value. (reported by Timmy Douglas)
711 * build fix: fixed the dependence on *LOAD-PATHNAME* and
712 *COMPILE-FILE-PATHNAME* being absolute pathnames.
713 * on x86 compiler partially supports stack allocation of dynamic-extent
715 * GO and RETURN-FROM do not check the extent of their exit points
716 when compiled with SAFETY 0.
717 * fixed some bugs related to Unicode integration:
718 ** encoding and decoding errors are now much more robustly
719 handled; it should now be possible to recover even from invalid
720 input or output to the terminal. (thanks to Teemu Kalvas)
721 ** provided a first cut at implementing STRING-TO-OCTETS and
722 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
723 ** altered the SB-MD5 contributed module slightly, changing the
724 interface just enough for it to be supportable for builds where
725 lisp characters are not eight bits.
726 * fixed some bugs revealed by Paul Dietz' test suite:
727 ** the FORMATTER-generated functions for ~V[ conditionals require
728 the correct number of arguments.
729 ** READ-FROM-STRING returns the mandated second value when applied
730 to displaced strings.
731 ** the #\Rubout and #\Backspace characters are treated as invalid
732 constituent characters by the tokenizer.
734 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
735 * new feature: reloading changed shared object files with
736 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
737 * new feature: references to foreign variables and functions
738 can now be compiled and loaded before the corresponding shared
739 object file is loaded, as long as the foreign definitions are
740 available at runtime.
741 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
742 supported. (thanks to Dan Debertin)
743 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
744 just keywords. (thanks to Vincent Arkesteijn)
745 * fixed bug #331: structure-class instances corresponding to
746 DEFSTRUCT forms are now created eagerly.
747 * fixed bug #345: backtraces from calls to undefined functions work
748 on x86 as well. Related bug #61 is now also partially fixed on x86
749 (backtraces from throws to unknown catch tags.)
750 * bug fix: lambda-list parsing is now stricter vrt. order and number
751 of lambda-list keywords.
752 * bug fix: as specified by AMOP, an error is signalled if a
753 class-option appears multiple times in a DEFCLASS form. (reported
755 * bug fix: RANDOM can be compiled when the compiler derives the type
756 of its numeric argument as a disjoint set of small integers.
757 (reported by Paul Dietz)
758 * bug fix: starting a core saved with shared objects loaded when
759 those objects are not available no longer causes threaded SBCL to
760 hang. (reported by Sean Ross)
761 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
763 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
764 argument, which must always be NIL. (reported by Kalle Niemitalo)
765 * bug fix: printing 1.0d+23 no longer results in an error.
766 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
767 * bug fix: structure slot setters preserve evaluation order. (thanks
769 * fixed some bugs related to Unicode integration:
770 ** RUN-PROGRAM can allow its child to take input from a Lisp
771 stream. (reported by Stefan Scholl)
772 ** ASDF-INSTALL successfully downloads .tar.gz files in all
773 locales. (reported by Ken Causey)
774 * fixed some bugs revealed by Paul Dietz' test suite:
775 ** INCF, DECF and REMF evaluate their place form as specified in
777 ** FORMATTER expands ~{ ~} iteration directives with V or #
778 parameters correctly.
779 ** FORMATTER deals with the ~@[ ~] conditional directive where the
780 consequent uses no arguments correctly.
781 ** the system has a partial understanding of the (COMPLEX RATIO)
783 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
786 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
787 * new feature: a build-time option (controlled by the :SB-UNICODE
788 keyword feature, enabled by default) for building the system with
789 support for the entire 21-bit character space defined by the
791 * new feature: the system now has rudimentary external-format
792 support; the primary user-visible change at this time is that
793 characters with the high bit set (such as lower-case-e-acute) will
794 print correctly to a terminal in a UTF-8 environment.
795 * minor incompatible change: BASE-CHAR no longer names a class;
796 however, CHARACTER continues to do so, as required by ANSI.
797 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
798 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
799 has been moved to the SB-EXT package (temporarily re-exported from
801 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
802 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
804 * fixed bug #351: better error-handlind and reporting for malformed
806 * fixed bug #350: bignum-printing is now more memory-efficient,
807 allowing printing of very large bignums, eg. (expt 2 10000000).
808 (reported by Bruno Haible)
809 * fixed bug #302: better primitive-type selection for intersection
811 * fixed bug #308: non-graphic characters now all have names, as
812 required. (reported by Bruno Haible)
813 * bug fix: redefining a class with different superclasses now correctly
814 removes it from the direct-subclasses of its previous superclasses.
815 (reported by David Morse)
816 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
817 new value now works. (reported by Bruno Haible)
818 * bug fix: correct canonicalization of multiple non-standard slot
819 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
820 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
821 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
822 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
823 now exists, an signals an error.
824 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
825 during macroexpansion for non-iteration variables as well. (reported
826 by Bruno Haible for CMUCL)
827 * bug fix: Cyclic structures and unprintable objects in compiler
828 messages no longer cause errors. (reported by Bruno Haible)
829 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
830 and READ-FROM-STRING all now return a primary value of NIL if
831 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
832 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
833 (reported by Bruno Haible for CMUCL)
834 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
835 do not designate a condition. (reported by Bruno Haible for
837 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
838 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
839 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
840 permitting resolution in favour of any of the conflicting symbols.
841 (reported by Bruno Haible for CMUCL)
842 * bug fix: EQUAL compiler optimizations is less aggressive on
843 strings which can potentially compare true despite having distinct
844 specialized array element types.
845 * bug fix: unit enumerations can be defined without dividing by
846 zero. (thanks to Vincent Arkesteijn)
847 * FORMAT compile-time argument count checking has been enhanced.
848 (report from Bruno Haible for CMUCL)
849 * a partial workaround for the bug 262: the compiler does not try to
850 inline-expand a local function doing RETURN-FROM from a deleted
851 BLOCK. (thanks to Peter Denno for the bug report and to David
852 Wragg for the simple test case)
853 * fixed some bugs revealed by Paul Dietz' test suite:
854 ** RENAME-PACKAGE allows all package designators as new package
856 ** constraint propagation and lambda variable substitution are
857 more cautious in dealing with partially deleted code.
858 ** compiler sometimes lost reoptimization passes.
859 ** CERROR, given a condition as condition designator, passes its
860 remaining arguments to the continue format control without
862 ** Case-altering FORMAT directives work correctly on non-ASCII
864 ** The REAL type specifier handles bounds outside the floating
865 point ranges without signalling FLOATING-POINT-OVERFLOW.
866 ** Functions with IR1-transformations can create intercomponent
867 references to global functions.
868 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
870 ** FORMAT ~R treats a nil value for its first parameter correctly.
872 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
873 * enhancement: saving cores with foreign code loaded is now
874 supported on x86/NetBSD and sparc/Linux in addition to the previously
876 * bug fix: on some platforms repeated installations caused multiple
877 copies of HTML documentation to be installed -- should not happen
878 any more. (reported by Stefan Scholl)
879 * bug fix: parsing self-recursive alien record types multiple times
880 no longer causes infinite recursion. (reported by Thomas F. Burdick,
881 original patch by Helmut Eller for CMUCL)
882 * bug fix: stack-exhaustion detection works now on NetBSD as well.
883 (thanks to Richard Kreuter)
884 * bug fix: defining classes whose accessors are methods on existing
885 generic functions in other (locked) packages no longer signals
886 bogus package lock violations. (reported by François-René Rideau)
887 * bug fix: special variables as DEFMETHOD parameters no longer have
888 associated bogus type declarations. (reported by David Wragg and
890 * bug fix: read-write consistency on streams of element-type
891 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
892 * bug fix: redefiniton of the only method of a generic function with
893 no DEFGENERIC no longer emits a full WARNING. In addition,
894 redefinition of generic functions with no DEFGENERIC to an
895 incompatible lambda list now signals an error. (thanks to Zach
897 * bug fix: DEFGENERIC now works even when there's a function of the
898 same name in an enclosing lexical environment. (thanks to Zach
900 * fixed compiler failure, caused by instrumenting code during
901 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
902 * optimization: added loop analysis and improved register allocation
903 * fixed some bugs revealed by Paul Dietz' test suite:
904 ** POSITION on displaced vectors with non-zero displacement
905 returns the right answer.
906 ** (SIMPLE-STRING) is a valid type specifier for sequence
908 ** *PRINT-LEVEL* handling for slotless structures is pedantically
910 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
911 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
913 ** The justification version of the FORMAT ~< directive treats
914 non-zero minpad parameter correctly.
916 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
917 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
918 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
919 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
920 the supported interface.
921 * new feature: Single-stepping of code compiled with DEBUG 2 or
922 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
923 * new feature: saving cores with foreign code loaded is now
924 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
925 Timothy Moore's work for CMUCL)
926 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
927 parameters to * instead of NIL if no initform is supplied.
928 (reported by Johan Bockgaard)
929 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
930 correctly when FUNCALL appears as the car of the form. Note:
931 despite this FUNCALL forms are not currently subject to
932 compiler-macro expansion. (port of Raymond Toy's fix for the
933 same from CMUCL, reported by Johan Bockgaard)
934 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
935 (thanks for Teemu Kalvas)
936 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
937 case instead of terminating the iteration (thanks for Julian
938 Squires, Sean Champ and Raymond Toy)
939 * bug fix: incorrect expansion of defgeneric that caused a style
940 warning. (thanks for Zach Beane)
941 * bug fix: slot accessor effective method computation works properly
942 for classes with multiple non-standard applicable methods on
943 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
944 * on x86 compiler supports stack allocation of results of LIST and
945 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
946 CMUCL implementation by Gerd Moellmann)
947 * fixed some bugs revealed by Paul Dietz' test suite:
948 ** FORMAT strings with both the ~<~:;~> form of the justification
949 directive and pretty-printing directives cause an error.
951 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
952 * incompatible change: the internal functions
953 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
954 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
955 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
956 instead of the old functions.
957 * new feature: on platforms where "dladdr" is available foreign
958 function names now appear in backtraces. (based on Helmut Eller's
960 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
961 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
963 * documentation: ASDF manual is now distributed with SBCL.
964 * bug fix: SBCL can now load its contributed modules with REQUIRE
965 even if the system-provided entries have been removed from
966 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
968 * bug fix: backtraces involving undefined functions or assembly
969 routines are more informative. (thanks to Brian Downing)
970 * bug fix: mutually referent alien structures now work correctly.
971 (reported by Rick Taube)
972 * bug fix: structures defined by WITH-ALIEN can be referred to
973 within other definitions in the same WITH-ALIEN.
974 * bug fix: division operators (MOD, TRUNCATE and the like) with
975 constant zero divisors and integer dividends no longer generate
977 * bug fix: provide default methods for INPUT-STREAM-P and
978 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
979 * bug fix: improve the detection and resolution of MOP metacycles
980 (where effective method computation depends on the generic function
981 having its effective method computed). (reported by Bruno Haible)
982 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
983 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
984 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
985 * bug fix: pathnames with relative directory components can be
986 represented relative to default pathnames.
987 * optimization: in taking the GCD of bignums, reduce the two bignums
988 to approximately the same size (using Euclid's algorithm) before
989 applying the more sophisticated binary GCD. (thanks to Juho
991 * optimization: COUNT on bitvectors now operates word-at-a-time.
992 * optimization: ASH with a positive, but not necessarily constant,
993 (leftwards) shift, when in a modular context, is compiled to a
995 * fixed some bugs revealed by Paul Dietz' test suite:
996 ** FORMAT variable parameters ("~V<char>") are defaulted properly
997 if the corresponding argument is NIL.
998 ** FORMAT directives accepting a minpad parameter treat negative
1000 ** literal commas as character parameters to FORMAT directives are
1002 ** literal spaces directly after ~<Newline> directives within a
1003 format-logical-block (~:< ~@:>) do not induce :FILL-style
1004 conditional newlines.
1005 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1006 as specified: it no longer includes conditional newlines).
1007 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1009 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1010 :PER-LINE-PREFIX argument does not evaluate to a string.
1012 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1013 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1014 the manual for details; for now, package locks can be disabled by
1015 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1016 if no major problems are found then it is likely that they will be
1017 compiled in unconditionally.
1018 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1019 unsupported operators on all platforms. To load a shared library
1020 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1021 object file, link it into a shared library outside of SBCL and
1022 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1024 * bug workaround: ROOM T and the GC were not working together
1025 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1026 reporting facility aren't preserved. That reporting has been
1027 disabled, so now until and unless someone figures out how to make it
1028 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1029 * minor incompatible change: as threatened around sbcl-0.8.0, the
1030 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1031 The symbols are also exported from SB-PCL for backwards
1032 compatibility, but more so than before SB-PCL should be treated as
1033 an implementation-internal package.
1034 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1036 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1037 It is known as of this release not to work on the Alpha, however.
1038 * fixed bug #167: errors signalled due to illegal syntax in method
1039 bodies are now more legible.
1040 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1041 designators and can hence be used with TYPEP.
1042 * fixed bug #333: CHECK-TYPE now ensures that the type error
1043 signalled, if any, has the right object to be accessed by
1044 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1045 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1046 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1048 * fixed bug #334: programmatic addition of slots using specialized
1049 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1050 and :ALLOCATION :CLASS slots.
1051 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1052 float exponents. (rereported by Peter Seibel)
1053 * fixed a bug: #\Space (and other whitespace characters) are no
1054 longer considered to be macro characters in standard syntax by
1055 GET-MACRO-CHARACTER.
1056 * fixed bug: initialization of condition class metaobjects no longer
1057 causes an instance of the condition to be created. (reported by
1059 * fixed bug: it is now possible to have slots such that
1060 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1061 description is neither :INSTANCE nor :CLASS.
1062 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1063 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1064 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1065 non-standard SLOT-DEFINITION-ALLOCATION values.
1066 * fixed bugs: various race conditions handling exiting threads.
1067 CL-PPCRE's thread torture test now passes.
1068 * fixed arguably-a-bug: GC time proportional to number of threads in
1069 system even when most of them are idle
1070 * optimization: improved performance of BIT and SBIT on bit-vectors.
1071 * .fasl file incompatibility: The fasl file version number has
1072 been incremented because of changes associated with package locks.
1074 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1075 * minor incompatible change: the system no longer provides
1076 optimization hints (conditions which are TYPEP
1077 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1078 mode; these hints are still emitted when higher SPEED optimization
1080 * new contrib module: a sampling profiler (profiling by statistical
1081 sampling, rather than by instrumenting functions) is available as
1082 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1083 * the behaviour of the standard function ED is now customizeable by
1084 third parties through a hook variable: see ED's documentation
1085 string for information on the protocol.
1086 * the compiler no longer emits efficiency notes for (FUNCALL X)
1087 when the type of X is uncertain under default optimization
1089 * fixed bug 276: mutating a binding of a specialized parameter to a
1090 method to something that is not TYPEP the specializer is now
1092 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1093 exhibit the required behaviour on the x86 platform. (thanks to
1094 Peter van Eynde, Eric Marsden and Bruno Haible)
1095 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1096 even for difficult arguments. (reported by Peter Graves)
1097 * fixed bug 141a: the backquote printer now descends quoted
1099 * fixed another bug in backquote printing: no more destructive
1100 modification of the form's list structure. (reported by Brian
1102 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1103 move between its address being taken and the call to
1104 interrupt_thread, fixing a crashing race condition.
1105 * the SB-POSIX contrib implementation has been adjusted so that it
1106 no longer exhibits ridiculously poor performance when constructing
1107 instances corresponding to C structs.
1109 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1110 * minor incompatible change: the sb-grovel contrib now treats C
1111 structures as alien (in the sense of SB-ALIEN) objects rather than
1112 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1113 has implications for memory management of client code
1114 (sb-grovel-returned objects must now be manually managed) and for
1115 type safety (alien objects now have full types).
1116 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1117 used to control emission of compiler diagnostics, rather than the
1118 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1119 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1120 quality should be considered deprecated.
1121 * install.sh now installs the user manual as well
1122 * (not quite a new documentable feature, but worth considering in
1123 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1124 beginnings of a semantically meaningful condition hierarchy is
1125 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1127 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1128 designator as the defaults argument.
1129 * fixed bug: Displaced arrays whose displaced-to array has become
1130 too small now cause ARRAY-DIMENSION to signal an error, providing
1131 for safer bounds-checking. (reported by Bruno Haible)
1132 * fixed bug: DEFCLASS slot definitions with identical :READER and
1133 :WRITER names now signal a reasonable error. (reported by Thomas
1135 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1137 * fixed bug: Invalid slot specification errors now print correctly.
1138 (thanks to Zach Beane)
1139 * fixed bug 320: Shared to local slot value transfers in class
1140 redefinitions now happen corrently from superclasses as
1141 well. (reported by Bruno Haible)
1142 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1144 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1145 improper lists correctly. (reported by Bruno Haible)
1146 * fixed bug 313: source-transform for <fun-name> was erroneously
1147 applied to a call of a value of a variable with name <fun-name>.
1148 (reported by Antonio Menezes Leitao)
1149 * fixed bug 307: The obsolete instance protocol ensures that
1150 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1151 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1152 remove dead unknown-values globs from the middle of the stack.
1153 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1154 easier to resume long computations after using *BREAK-ON-SIGNALS*
1155 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1156 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1157 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1158 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1159 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1160 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1162 * fixed a (fairly theoretical) bug in string printing: if
1163 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1164 does not have array-element-type equal to the most general string
1166 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1167 function name. (thanks to Nikodemus Siivola)
1168 * fixed bug: compile-time format string checker failed on
1169 non-closed ~{. (reported by Thomas F Burdick)
1170 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1171 don't warn on legal loop code involving more than one aggregate
1172 boolean. (thanks to Nikodemus Siivola)
1173 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1174 in keyword position to MAKE-PACKAGE was not being reported as
1176 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1177 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1179 * fixed bugs 280 and 312: the checking for multiple definitions in a
1180 file is less likely to become confused by uses of inline
1182 * fixed bug: the #S reader macro performs the keyword coercion
1183 specified for slot names. (reported by Kalle Niemitalo)
1184 * fixed bug: lambda lists may contain symbols whose names start with
1185 & but are not lambda-list-keywords; their occurrence triggers a
1187 * fixed bug 321: define-method-combination argument lambda lists do
1188 not cause invalid code to be generated when &optional or &aux
1189 variables have default values. (reported by Bruno Haible)
1190 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1191 have CLOS classes; this appears also to have fixed reported
1192 instability in stack exhaustion detection.
1193 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1194 MUFFLE-WARNING when no associated restart is present is now
1196 * optimization: rearranged the expansion of various defining macros
1197 so that each expands into only one top-level form in a
1198 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1200 * optimization: used a previously little-used slot in symbols to
1201 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1203 * fixed some bugs revealed by Paul Dietz' test suite:
1204 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1206 ** Fixed an optimization bug related to inheritance of initforms
1207 from local to shared slots.
1208 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1209 ** CAST optimizer forgot to flush argument derived type.
1210 ** print/read consistency on floats is now orders of magnitude
1211 more likely. (thanks also to Bruno Haible for a similar report
1213 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1214 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1215 for objects of type REAL. Make it so.
1216 ** SXHASH is specified (infelicitously) to respect similarity,
1217 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1218 it so. (thanks to Markus Ziegler)
1219 ** on the Alpha, the compiler succeeds in compiling functions
1220 returning a known number of arguments greater than 63.
1221 ** fixed handling of invalid NIL arguments in keyword position
1223 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1225 ** provided a readably-printable representation for RANDOM-STATE
1227 ** ensured that pathnames, where they have a namestring, always
1228 print using #P"..." syntax.
1230 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1231 * Support for the forthcoming 2.0 version of the NetBSD kernel
1232 running on x86 hardware has been added. (thanks to Perry
1233 E. Metzger most immediately, and others for their past work)
1234 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1235 current release version; SBCL's OpenBSD support had been broken
1236 since about the time of OpenBSD's switch to ELF binary format.)
1237 * [placeholder for DX summary]
1238 ** user code with &REST lists declared dynamic-extent, under high
1239 speed or space and low safety and debug optimization policy.
1240 * The manual has been converted to Texinfo format and the debugger
1241 chapter from the cmucl manual has been added.
1242 * A facility has been added to extract documentation strings from
1243 sbcl and store them as Texinfo-formatted snippets for inclusion in
1244 the manual (via Texinfo's @include directive)
1245 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1246 the test case to Dave Roberts)
1247 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1248 pointers. (reported by Sean Ross)
1249 * bug fix: PROFILE output is printed nicely even for large numerical
1250 values. (thanks to Zach Beane)
1251 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1252 greater than 32 handle EOF correctly.
1253 * bug fix: on X86 an immediate argument of the IMUL instruction is
1254 correctly printed in disassembly. (thanks to Lutz Euler)
1255 * bug fix: class slots in redefined classes preserve their old
1256 values. (thanks to Bruno Haible and Nikodemus Siivola)
1257 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1258 succeeds. (reported by Marco Baringer)
1259 * fixed some bugs revealed by Paul Dietz' test suite:
1260 ** READ-SEQUENCE now works on ECHO-STREAMs.
1261 ** RATIONALIZE works more according to its specification. (thanks
1264 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1265 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1266 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1267 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1268 should matter to you only if you rebind the printer control
1269 variables and then find you want different bindings in the
1270 debugger than in the ordinary execution of your program.)
1271 * The runtime build system has been tweaked to support building
1272 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1273 assembler and linker. (thanks to Nikodemus Siivola)
1274 * Unbound, undefined, undeclared variables now trigger full
1275 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1276 more often programmer error than deliberate exploitation of undefined
1278 * optimization: the hash algorithm for strings has changed to one
1279 that is less vulnerable to spurious collisions. (thanks to Juho
1281 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1282 needless bounds checking. (thanks to Juho Snellman)
1283 * optimization: implemented multiplication as a modular
1284 (UNSIGNED-BYTE 32) operation on the PPC backend.
1285 * fixed some bugs revealed by Paul Dietz' test suite:
1286 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1288 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1289 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1290 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1292 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1294 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1296 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1297 their output stream on EOF from read.
1298 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1299 have been read to end-of-file.
1300 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1302 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1303 description of determination of which consecutive characters
1305 ** Printing the "Space" character with escaping on now yields "#\\ ",
1306 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1307 ** Reading floating-point numbers with *READ-BASE* set to a number
1308 less than 10 works correctly.
1309 ** Reading floating-point numbers with *READ-BASE* set to a number
1310 more than 10 works correctly.
1311 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1312 the readtable currently in effect.
1314 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1315 * minor incompatible change: parsing of namestrings on a physical
1316 (Unix) host has changed; numbers after the final #\. in a
1317 namestring are no longer interpreted as a version field. This is
1318 intented to be largely invisible to the user, except that the
1319 meaning of the namestring "*.*.*" has changed: it now refers to a
1320 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1321 should usually be replaced by
1322 (make-pathname :name :wild :type :wild :version :wild)
1323 with the added benefit that this is more likely to be portable.
1324 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1325 option to OPEN now signals an error if the file being opened
1326 exists; this may have an impact on existing code.
1327 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1328 signals received, on the PowerPC platforms (both Linux and
1329 Darwin). (thanks to Pierre Mai for pointing out the location of
1331 * several fixes on OS X: The system now builds and runs cleanly on
1332 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1333 (thanks to Brian Mastenbrook)
1334 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1335 second-resolution: integer multiples of 1/3600 between -24 and 24.
1336 (thanks to Vincent Arkesteijn)
1337 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1338 type when called with 1 argument; PEEK-CHAR checked type of
1339 PEEK-TYPE only after having read first character from a
1340 stream. (reported by Peter Graves)
1341 * bug fix: the garbage collector now has much better locality
1342 behaviour, and in particular no longer treats all memory as being
1343 exhausted when many small objects point to each other in a deeply
1345 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1346 recognized as being TYPEP their class.
1347 * bug fix: the PUSHNEW documentation string has been corrected.
1348 (thanks to Vincent Arkesteijn)
1349 * bug fix: defaulting of the value for the last of an atypically
1350 large number of multiple values being bound was not being
1351 performed correctly on the Alpha or PPC platforms
1352 * optimization: implemented multiplication as a modular
1353 (UNSIGNED-BYTE 32) operation on the x86 backend.
1354 * optimization: SEARCH on simple-base-strings can now be open-coded.
1355 (see also contrib/compiler-extras.lisp for inspiration for
1356 teaching the compiler about the Boyer-Moore algorithm).
1357 * value, returned by MAX (and MIN) called with several EQUALP, but
1358 not EQL, arguments now does not depend on compiler settings.
1359 * fixed some bugs revealed by Paul Dietz' test suite:
1360 ** in stack analysis liveness information is propagated from
1361 non-local entry points.
1362 ** pathwise CAST removing failed when the CAST node did not start
1364 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1365 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1367 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1368 host is already defined.
1369 ** RENAME-FILE works on streams instead of signalling an internal
1371 ** PEEK-CHAR uses the current readtable when determining whether
1372 or not a character is whitespace.
1373 ** MERGE-PATHNAMES handles the case when the pathname does not
1374 specify a name while the default-pathname specifies a version.
1375 ** Pathnames now stand a better chance of respecting print/read
1377 ** Attempting to use standardized file system operators with a
1378 pathname with invalid :DIRECTORY components signals a
1380 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1382 ** DIRECTORY on logical pathnames is more correct.
1383 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1384 signal a TYPE-ERROR if their argument is not a stream.
1385 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1386 designator argument does not designate a stream.
1387 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1388 examining the synonym.
1389 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1391 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1392 element-type larger than ([UN]SIGNED-BYTE 32).
1394 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1395 * When built with the :SB-FUTEX feature, threaded builds now take
1396 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1397 for faster/more reliable mutex and condition variable support.
1398 * Incompatible change (but one you probably shouldn't have been using
1399 anyway): the interface and code for arbitrating between multiple
1400 threads in the same user session has been redesigned.
1401 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1402 error when called without an explicit environment argument.
1403 (thanks to Nikodemus Siivola)
1404 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1405 confused about their position. (thanks to Adam Warner and Gerd
1407 * bug fix: Pretty printing backquoted forms with unquotations in the
1408 argument list position of various code constructs such as LAMBDA
1409 now works correctly. (reported by Paul Dietz)
1410 * bug fix: Pretty printing unquotations no longer loses all
1411 stream position information.
1412 * optimization: performance of string output streams is now less
1413 poor for multiple small sequence writes.
1414 * optimization: performance of CSUBTYPEP in the presence of complex
1415 expressions involving CONS and NOT many times has been improved.
1416 (reported by Paul Dietz)
1417 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1419 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1420 streams for the REPL, for the debugger, and so on. (thanks to
1422 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1423 * fixed some bugs revealed by Paul Dietz' test suite:
1424 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1425 is ignored if it is too small, rather than propagating through
1426 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1427 ** extremely complex negations of CONS types were not being
1428 sufficiently canonicalized, leading to inconsistencies in
1430 ** VALUES tranformer lost derived type.
1432 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1433 * fixed a bootstrapping bug: the build process no longer assumes
1434 that the various BOOLE-related constants have the same value in
1435 host and target lisps. (noted by Paul Dietz' test suite on an
1436 SBCL binary built from CLISP)
1437 * The system can now be dynamically linked on the MIPS platform,
1438 which enables dynamic loading of foreign code from Lisp. (thanks
1439 to Ralf Baechle for discussions on the MIPS ABI)
1440 * The system now records debugging information for its own source
1441 files in a filesystem-position-independent manner, relative to
1442 the "SYS" logical host.
1443 * fixed a compiler bug: MV-LET convertion did not check references
1444 to the "max args" entry point. (reported by Brian Downing)
1445 * tweaked disassembly notes to be less confident about proclaiming
1446 some instruction as an LRA. (thanks to Brian Downing)
1447 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1448 now each have their own history, command character, and other
1449 characteristics. (thanks to David Lichteblau)
1450 * fixed some bugs revealed by Paul Dietz' test suite:
1451 ** compiler failure in compiling LOGAND expressions including a
1453 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1454 shift greater than 32.
1455 ** FLUSH-DEST did not mark blocks for type check regeneration.
1456 ** HANDLER-CASE failed to accept declarations in handler clauses
1457 in some circumstances.
1459 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1460 * New code in contrib/sb-introspect (still probably not entirely
1461 stable yet) provides some support for smart Lisp development
1462 environments like SLIME.
1463 * The conditions signalled for errors occurring when loading .fasl
1464 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1465 in a way which should help ASDF recover gracefully.
1466 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1467 stuff has been cleaned up. If you code contrib/ stuff, this might
1468 affect you, and you can look at contrib/README, contrib/STANDARDS,
1469 and/or the 0.8.4.27 diff to check.
1470 * In full calls the compiler now does not generate checks for declared
1471 argument types for all arguments.
1472 * various threading fixes
1473 ** and some experimental patches which didn't make it into
1474 the main tree for this release, but which are shipped in
1475 contrib/experimental-thread.patch as a possible fix for some
1476 failures (deadlock, spinning...) in GC-intensive multithreaded
1478 * fixed PPC build problem (source code incompatibility of different
1479 library versions): added offsetof() hackery which attempts to divine
1480 where glibc maintainers put uc_mcontext today
1481 * fixed bug 282: compiler does not trust type assertions while passing
1482 arguments to a full call.
1483 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1484 &OPTIONAL VALUES type parameter.
1485 * fix bug 214: algorithm for noting rejected templates is now more
1486 similar to that of template seletion. (also reported by rydis on
1488 * fixed bug 141b: printing backquoted information readably and prettily
1489 inserts a space where necessary.
1490 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1491 (CHARACTER 10) are now reported as errors, rather than propagated
1492 as unknown types. (reported by piso on #lisp)
1493 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1494 with values NIL and :ERROR. (thanks to Milan Zamazal)
1495 * fixed bug 191c: CLOS now does proper keyword argument checking as
1496 described in CLHS 7.6.5 and 7.6.5.1.
1497 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1498 longer raise an error; further, using a list as a for-as-arithmetic
1499 counter now raises a meaningful error.
1500 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1501 now understood by sequence creation functions such as MAKE-SEQUENCE
1503 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1504 a STRING-INPUT-STREAM.
1505 * compiler enhancement: SIGNUM is now better able to derive the type
1507 * type declarations inside WITH-SLOTS are checked. (reported by
1509 * fixed some bugs revealed by Paul Dietz' test suite:
1510 ** incorrect optimization of TRUNCATE for a positive first
1511 argument and negative second.
1512 ** compiler failure in let-convertion during flushing dead code.
1513 ** compiler failure while deriving type of TRUNCATE on an
1514 interval, containing 0.
1515 ** ASH of a negative bignum by a negative bignum count now returns
1517 ** intersection of CONS types now canonicalizes properly, fixing
1518 inconsistencies in SUBTYPEP.
1520 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1521 * incompatible change: The --disable-debugger command line
1522 option now clobbers the debugger at a more fundamental
1523 level, by redefining #'INVOKE-DEBUGGER instead of by
1524 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1525 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1526 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1527 enter the debugger REPL and then suffer recursive errors
1528 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1529 while under the new system, BREAK in --disable-debugger mode
1530 terminates the system just as an unhandled error would.
1531 * fixed compiler performance when processing loops with a step >1;
1532 * bug fix: DOCUMENTATION now retrieves generic function
1533 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1534 support has been systematized, and now supports the methods
1535 specified by ANSI, along with a default method and a method for
1536 slot documentation. (reported by Nathan Froyd)
1537 * bug fix: effective methods associated with a generic function are
1538 no longer cached over a change of that generic function's method
1539 combination. (reported by Andreas Fuchs)
1540 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1541 names a non-existent directory. (thanks to Andreas Fuchs)
1542 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1543 the PPC platform, lead to stack corruption; this has been fixed.
1544 (reported by Rainer Joswig)
1545 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1546 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1547 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1548 arithmetic in general, is now much more reliable on the PPC
1550 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1551 platform now returns the right answer.
1552 * optimization: restored some effective method precomputation in
1553 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1554 precomputation is now tunable.
1555 * optimization: compiler-internal data structure use has been
1556 reviewed, and changes have been made that should improve the
1557 performance of the compiler by about 20%.
1558 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1559 simple-base-strings and simple-bit-vectors is improved.
1560 * optimization: the optimization of 32-bit logical and arithmetic
1561 functions introduced in version 0.8.3 on the x86 has been
1562 implemented on the mips, ppc and sparc platforms; an
1563 implementation of the same facility, but for 64-bit arithmetic,
1564 has been added for the alpha.
1565 * microoptimization: the compiler is better able to make use of the
1566 x86 LEA instruction for multiplication by constants.
1567 * bug fix: in some situations compiler did not report usage of
1568 generic arithmetic in (SPEED 3) policy.
1569 * bug 145b fix: compiler used wrong type specifier while converting
1570 MEMBER-types to numeric.
1571 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1573 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1574 index argument now works properly on the Alpha platform.
1575 * bug fix: floating point exception treatment on the Alpha platform
1577 * bug fix: FILE-POSITION works much better on string input and
1578 output streams. (thanks to Nikodemus Siivola)
1579 * bug fix: many threading/garbage collection symptoms sorted.
1580 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1581 might be pseudo-atomic.
1582 * internal change: Stopping for GC is now done with signals not
1583 ptrace. GC is now done in whichever thread wanted it, instead of
1585 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1587 * bug fix: result form in DO is not contained in the implicit
1589 * incompatible change: ICR structure is changed; the value part of
1590 CONTINUATION is now called LVAR; corresponding functions are
1591 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1592 * added type deriver for ISQRT (thanks to Robert E. Brown).
1593 * bug fix: better support for loading from the command line when an
1594 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1596 * fixed some bugs revealed by Paul Dietz' test suite:
1597 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1598 ** ROUND and FROUND now give the right answer when given very
1599 small float arguments.
1600 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1602 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1603 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1604 ** LCM should return a non-negative integer.
1605 ** PARSE-INTEGER returned the index of a terminator instead of the
1606 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1607 ** PARSE-INTEGER returned an incorrect index being applied to a
1609 ** LCM with two arguments of 0 returns 0 rather than signalling
1611 ** unsigned addition of a 32-bit constant with the high bit set no
1612 longer causes an internal compiler error on the x86.
1613 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1614 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1615 with negative last argument.
1616 ** byte specifiers with zero size and position no longer cause
1617 an error during type derivation.
1618 ** bignum multiplication on the Alpha platform now returns the
1620 * porting: The system now builds on SuSE AMD64, although it still
1621 generates a 32-bit binary.
1622 * .fasl file incompatibility: The fasl file version number has
1623 been incremented (because of the changes to internal compiler
1624 data structures referred to above).
1626 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1627 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1628 more accurately, on the Darwin kernel running on PowerPC hardware.
1629 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1630 * Compiler code deletion notes now signal a condition of type
1631 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1632 an associated MUFFLE-WARNING restart.
1633 * The compiler now performs limited argument count validation of
1634 constant format strings in FORMAT, and where appropriate in ERROR,
1635 CERROR and WARN. (thanks to Gerd Moellmann)
1636 * New ASDF-INSTALL contrib can be used for automatic download and
1637 installation of third-party Lisp code from CCLAN or other sites
1639 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1640 which forces another thread to execute a function supplied by the
1642 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1643 now accept and act upon their :ELEMENT-TYPE keyword argument.
1644 (reported by Martin Atzmueller, Edi Weitz)
1645 * bug fix: FILE-POSITION now accepts position designators up to
1646 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1647 is the greater. (thanks to Patrik Nordebo)
1648 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1649 Kalle Olavi Niemitalo)
1650 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1651 printable. (reported by Eric Marsden)
1652 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1653 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1654 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1655 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1656 (thanks to Stig Sandoe)
1657 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1658 UNREAD-CHAR now correctly includes the unread character in the
1659 target sequence. (thanks to Gerd Moellmann)
1660 * bug fix in threaded builds: the system can now be suspended and
1661 resumed by shell job control with minimal disruption.
1662 * bug fixes in times and timezones >2038AD
1663 * better handling of "where is GNU make?" problem in build scripts
1664 (thanks to Nikodemus Siivola)
1665 * new optimization: inside a named function any reference to a
1666 function with the same name is considered to be a self-reference;
1667 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1668 optimization quality.
1669 * new optimization on x86: logical functions and + now have
1670 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1671 used when the result is truncated to 32 bits.
1672 * VALUES declaration is partially enabled.
1673 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1674 * fixed some bugs revealed by Paul Dietz' test suite:
1675 ** The system now obeys the constraint imposed by
1676 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1677 types form a lattice under type intersection.
1678 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1679 ** ASSOC now ignores NIL elements in an alist.
1680 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1681 and (1+ MOST-POSITIVE-FIXNUM) answers.
1682 ** The addition of a method with invalid qualifiers to a generic
1683 function does not cause an error to be signalled immediately;
1684 a warning is signalled, and the error is generated only on
1685 calling the generic function.
1686 * changed .fasl file version number, in part to add type codes for
1687 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1688 obscure ANSI requirements
1690 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1691 * fixed bug 148: failure to inline-expand a local function left
1692 garbage, confusing the compiler.
1693 * fixed bugs 3cd: structure slot readers perform type check if the
1694 slot can have an invalid value (i.e. it is either not initialized
1695 or can be written with a less specific slot writer).
1696 * bug fix: the compiler now traps array references to elements off
1697 the end of an array; previously, the bounds checking in some
1698 circumstances could go off-by-one.
1699 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1701 * type declarations for array element types now obey the description
1702 on the CLHS page "Declaration TYPE", as per discussions on
1703 sbcl-help around 2003-05-08. This means that a declaration
1704 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1705 declaration, all references to BAR will be asserted or assumed
1706 (with THE, so dependent on compiler policy) to involve objects of
1707 type FOO. Note that no such declaration is implied in
1708 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1709 * declared types of functions from the "Conditions"
1710 chapter. (reported by Paul Dietz)
1711 * bug fix: CERROR accepts a function as its first argument.
1712 * bug fix: NTH an NTHCDR accept a bignum as index
1713 arguments. (reported by Adam Warner)
1714 * optimization: character compare routines now optimize comparing
1715 against a constant character. (reported by Gilbert Baumann)
1716 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1717 argument now works properly on the MIPS platform.
1718 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1719 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1720 declared to have a functional type, does not cause an error
1722 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1723 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1724 * fixed bug: if last continuation of a deleted block has a
1725 destination, this destination should be deleted too. (reported by
1727 * fixed a bug in the bootstrap process: the host compiler's values
1728 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1729 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1730 test case from Patrik Nordebo)
1731 * improved the ability of the disassembler on the PPC platform to
1732 provide helpful disassembly notes.
1733 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1734 the class in more cases than previously.
1735 * bug fix: FILE-POSITION now understands :START and :END for
1736 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1737 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1738 8) data. (thanks to David Lichteblau)
1739 * bug fix: it is possible to add a method to a generic function
1740 without lambda list.
1741 * bug fix: reader failed to signal END-OF-FILE inside an
1742 object representation. (reported by Nikodemus Siivola)
1743 * fixed some bugs revealed by Paul Dietz' test suite:
1744 ** LAST and [N]BUTLAST should accept a bignum.
1745 ** condition slot accessors are methods.
1746 ** (VECTOR NIL) is a subtype of STRING.
1748 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1749 * minor incompatible change: some nonsensical specialized lambda
1750 lists (used in DEFMETHOD) which were previously ignored now signal
1752 * minor incompatible change: the system is now aware of the types of
1753 variables in the COMMON-LISP package, and will signal errors for
1754 most violations of these type constraints (where previously they
1755 were silently accepted).
1756 * minor incompatible change: COMPILE-FILE now uses the freedom
1757 afforded (ANSI 3.2.2.3) to use derived function types for
1758 functions defined in the same file. This also permits the system
1759 to warn on static type mismatches and function
1760 redefinition. (Currently it does not work with high DEBUG level.)
1761 * minor incompatible change: VALUES declaration is disabled.
1762 * When issuing notes, the compiler now signals a condition of type
1763 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1764 restart for use in user handlers. It is expected that the
1765 COMPILER-NOTE condition will eventually become a condition
1766 supertype to a hierarchy of note types, which will then be
1767 handleable in a similar fashion. However, at the moment, no such
1768 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1769 but it's an implementation detail, not a classification for the
1771 * Changes in type checking closed the following bugs:
1772 ** type checking of unused values (192b, 194d, 203);
1773 ** template selection based on unsafe type assertions (192c, 236);
1774 ** type checking in branches (194bc).
1775 * A short form of VALUES type specifier has ANSI meaning (it has
1776 increased the number of situations when SBCL cannot perform type
1778 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1779 keywords or constants is permissible.
1780 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1781 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1782 classes. (thanks to Antonio Martinez)
1783 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1784 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1785 argument to a pretty-print stream. Instead, it leaves any such
1786 operations to DESCRIBE-OBJECT methods.
1787 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1789 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1790 lambda lists. (thanks to David Lichteblau)
1791 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1792 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1793 (thanks to Gerd Moellmann)
1794 * fixed bug 246: increased compilation speed of long
1795 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1797 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1799 * DEFCONSTANT now throws a condition of type
1800 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1801 constant to a non-EQL value; CONTINUE and ABORT restarts
1802 respectively change and preserve the value.
1803 * fixed bug 63: The code walker, part of the implementation of CLOS,
1804 is now better at handling symbol macros.
1805 * bug fix: There is no longer an internal implementation type named
1806 CL:LENGTH. (reported by Raymond Toy)
1807 * bug fix: In macro-like defining macros/special operators the
1808 implicit block does not enclose the lambda list.
1809 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1810 enclosed in parentheses) are not suitable as type specifiers, and
1811 their use properly signals an error now.
1812 * bug fix: An argument count mismatch for a type specifier in code
1813 being compiled no longer causes an unhandled error at compile
1814 time, but signals a compile-time warning.
1815 * fixed simple vector readable printing
1816 * bug fix: DESCRIBE takes more care over whether the class
1817 precedence list slot of a class is bound before accessing it.
1818 (reported by Markus Krummenacker)
1819 * bug fix: FORMATTER can successfully compile pretty-printer format
1820 strings which use variants of the ~* directive inside.
1821 * bug fix: SEARCH now applies its TEST predicate to the elements of
1822 the arguments in the correct order. (thanks to Wolfhard Buss)
1823 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1824 in type inference. (thanks to Robert E. Brown)
1825 * bug fix: Reading in symbols with an explicit package name of ""
1826 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1827 * fixed some bugs revealed by Paul Dietz' test suite:
1828 ** NIL is now allowed as a structure slot name.
1829 ** Arbitrary numbers, not just REALs, are allowed in certain
1830 circumstances in LOOP for-as-arithmetic clauses.
1831 ** Multiple class redefinitions before slot access no longer
1832 causes a type error.
1833 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1834 association between the name and a class.
1835 ** Generic functions with non-standard method-combination and over
1836 five methods all of which return constants no longer return NIL
1837 after the first few invocations. (thanks to Gerd Moellmann)
1838 ** CALL-NEXT-METHOD with no arguments now passes the original
1839 values of the arguments, even in the presence of assignment.
1840 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1842 ** DISASSEMBLE works with closures and funcallable instances.
1843 ** ADD-METHOD now returns the generic function, not the new method.
1844 ** FIND-METHOD signals an error if the lengths of the specializers
1845 is incompatible with the generic function, even if the ERRORP
1847 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1848 which its argument is a member.
1849 ** DEFCLASS only redefines the class named by its class-name
1850 argument if that name is the proper name of the class;
1851 otherwise, it creates a new class.
1852 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1853 of the UNBOUND-SLOT condition to the name of the slot.
1854 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1855 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1856 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1857 SLOT-MAKUNBOUND in the specified fashion.
1859 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1860 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1861 cross-compilation host. As a consequence, we can now bootstrap our
1862 way up to SBCL starting with a bare gcc toolchain and human-readable
1863 source code (first the source to CLISP, then the source to SBCL).
1864 * A contributed module containing a partial implementation of the
1865 simple-streams interface has been included. (thanks to Rudi
1867 * A contributed module implementing the RFC1321 Message Digest
1868 Algorithm, known as MD5, has been included.
1869 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1870 no longer has any effect, as the code controlled by this feature
1871 has been deleted. (As far as we know, no-one has ever built using
1872 this feature, and its semantics were confused in any case).
1873 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1874 work on conditions (as required by the ANSI specification),
1875 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1876 expected behaviour on conditions. Users should note, however,
1877 that such behaviour is not required by the ANSI specification,
1878 and so use of this behaviour may render their code unportable.
1879 * fixed some bugs revealed by Paul Dietz' test suite:
1880 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1882 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1883 ** SXHASH on condition objects no longer returns NIL.
1884 ** :ALLOCATION :CLASS slots are better treated; their values are
1885 updated on class redefinition, and initforms inherited from
1886 superclasses are applied.
1887 ** REMOVE-METHOD returns its generic function argument even when
1888 no method was removed.
1889 ** SHARED-INITIALIZE now initializes the values of the requested
1890 slots, including those with :ALLOCATION :CLASS.
1891 ** ALLOCATE-INSTANCE now works on structure classes defined via
1892 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1894 ** SLOT-EXISTS-P now works on conditions, as well as structures
1896 ** MAKE-LOAD-FORM now has the required methods on
1897 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1898 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1899 keyword, and now implements the SLOT-NAMES argument.
1900 ** methods with &OPTIONAL arguments no longer allow too many
1901 arguments to be passed in the call without error.
1902 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1903 option is consistent with the required arguments of the generic
1904 function lambda list.
1905 * bug fix: REQUIRE accepts a string designator. (Thanks to
1907 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1908 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1909 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1910 * bug fix: The compiler now checks for duplicated variables in macro
1912 * bug fix: SETQ on globals returns the correct value.
1913 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1914 not a COMPILER-ERROR (followed by some other strange error on
1915 choosing the CONTINUE restart).
1916 * bug fix: make.sh and friends are now more consistent in the way that
1917 they look for GNU "make".
1919 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1920 * experimental native threads support (on x86 Linux >=2.4 only).
1921 This is not compiled in by default: you need to add :SB-THREAD to
1922 the target features. See the "Beyond ANSI" chapter of the manual
1924 * fix for longstanding nonANSIism: The old distinction between
1925 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1926 The return value from CL:FIND-CLASS is now a CLOS class, and
1927 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1928 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1929 * An interface to the MetaObject Protocol, as described in Kiczales,
1930 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1931 MIT Press, 1991, is available from the SB-MOP package.
1932 * incompatible change: the SB-PCL package should now be considered
1933 a private implementation detail, and no longer a semi-private MOP
1935 * minor incompatible change: due to rearrangement for threads, the
1936 control stack and binding stack are now allocated at arbitrary
1937 addresses instead of being hardcoded per-port. Users affected by
1938 this probably have to be doing advanced things with shared
1939 libraries, and will know who they are.
1940 * minor incompatible change: Previously, all --eval forms used were
1941 processed with READ before any of them were processed with EVAL.
1942 Now each --eval form is processed with both READ and EVAL before
1943 the next --eval form is processed. (Thus package operations like
1944 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1945 work as the user might reasonably expect.)
1946 * minor incompatible change: *STANDARD-INPUT* is now only an
1947 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1949 * minor incompatible change: Y-OR-N-P is now character-oriented,
1950 not line oriented. Also, YES-OR-NO-P now works without errors.
1951 (thanks to Antonio Martinez)
1952 * sb-aclrepl module improvements: an integrated inspector, added
1953 repl features, and a bug fix to :trace command.
1954 * Known functions, which cannot be open coded by the backend, are
1955 considered to be able to check types of their arguments. (fixing
1956 a bug report by Nathan J. Froyd)
1957 * fixed a bug in computing method discriminating functions: It is
1958 now possible to define methods specialized on classes which have
1959 forward-referenced superclasses. (thanks to Gerd Moellmann)
1960 * fixed evaluation order in optional entries (reported by Gilbert
1962 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1963 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1964 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1965 required-by-AMOP NAME argument, as well as CLASS and
1966 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1967 * fixed bug 20: DEFMETHOD can define methods using names that are
1968 not the proper names of classes to designate class specializers.
1969 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1970 Unix file descriptors, instead of blowing up. (thanks to Antonio
1972 * Garbage collection refactoring: user-visible change is that a
1973 call to the GC function during WITHOUT-GCING will not do garbage
1974 collection until the end of the WITHOUT-GCING. If you were doing
1975 this you were probably losing anyway.
1976 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1977 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1978 TYPEP the latter but not the former.
1979 * The compiler issues a full WARNING for calls to undefined functions
1980 with names from the CL package.
1981 * MAP-INTO for a vector destination is open coded. (reported by
1982 Brian Downing on c.l.l)
1983 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1984 documentation string.
1985 * fixed some bugs revealed by Paul Dietz' test suite:
1986 ** COPY-ALIST now signals an error if its argument is a dotted
1988 ** Condition slots are now accessed more correctly in the presence
1989 of multiple initargs for a given slot.
1990 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1991 exclude restarts of the same name associated with a different
1993 ** DEFCLASS of forward-referenced classes with another
1994 forward-referenced class in the superclasses list no longer
1996 ** Condition slots are now initialized once each, not multiple
1997 times. (thanks to Gerd Moellmann)
1998 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2000 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2001 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2002 arguments contain duplicated elements.
2003 ** RESTART-CASE understands local macros.
2004 ** RESTART-CASE associates exactly its own restarts with a condition.
2005 ** ENDP in safe mode checks its argument to be of type LIST.
2006 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2007 in question is unbound.
2008 ** Optimized MAKE-INSTANCE functions no longer cause internal
2009 assertion failures in the presence of duplicate initargs.
2010 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2011 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2012 propagated even in the ctor optimized implementation of
2014 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2016 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2017 * a better implementation of SXHASH on (simple) bit vectors,
2018 measured both in execution speed and in distribution of results
2019 over the positive fixnums, has been installed. Likewise, a better
2020 implementation of EQUAL for simple bit vectors is now available.
2021 * fixed CEILING optimization for a divisor of form 2^k.
2022 * fixed bug 240 (emitting extra style warnings "using the lexical
2023 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2024 by Antonio Martinez)
2025 * fixed SXHASH, giving different results for NIL depending on type
2026 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2027 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2028 variable. (found by Rolf Wester)
2029 * SBCL does not ignore type declarations for special
2030 variables. (reported by rif on c.l.l 2003-03-05)
2031 * some bug fixes in contrib/sb-aclrepl/
2032 * fixed some bugs revealed by Paul Dietz' test suite:
2033 ** a bug in the CONS type specifier, whereby the CAR and CDR
2034 types got intertwined, has been fixed;
2035 ** the type system is now able to reason about the interaction
2036 between INTEGER and RATIO types more completely;
2037 ** APPEND, [N]REVERSE and NRECONC check that those their
2038 arguments, which must be proper lists, are really so;
2039 ** An array specialized to be unable to hold elements has been
2040 implemented, as required -- yes, really -- by ANSI;
2041 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2042 on malformed property lists;
2044 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2045 * incompatible packaging change: in line with Unix convention,
2046 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2047 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2048 * REQUIRE and PROVIDE are now optionally capable of doing something
2049 useful. See the documentation string for REQUIRE.
2050 * infrastructure for a managed SBCL contrib system: contributed
2051 modules in this release include:
2052 ** the ASDF system definition facility;
2053 ** an interface to the BSD Sockets API;
2054 ** an ACL-like convenience interface to the repl;
2055 (thanks to Kevin Rosenberg)
2056 ** an implementation of ROTATE-BYTE, with efficient implementation
2058 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2059 gives a closer approximation to the right answer than previously.
2060 (thanks to Raymond Toy)
2061 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2062 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2063 optional environment arguments, as required by ANSI.
2064 * fixed bugs in other functions taking environment objects, allowing
2065 calls with an explicit NIL environment argument to be compiled
2067 * fixed bug 228: primary return values from
2068 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2069 COMPILE or FUNCTION.
2070 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2071 :TYPE will no longer signal errors on innocuous objects.
2072 * fixed bug 231b: SETQ is better at respecting type declarations in
2073 the lexical environment.
2074 * fixed a bug in DEFCLASS: classes named by symbols with no or
2075 unprintable packages can now be defined.
2076 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2077 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2078 * fixed compiler failure related to checking types of functions.
2079 (reported by Robert E. Brown)
2080 * the compiler is now much more consistent in its error-checking
2081 treatment of bounding index arguments to sequence functions: in
2082 (SAFETY 3) code, errors will be signalled in almost all cases if
2083 invalid sequence bounding indices are passed to functions defined
2084 by ANSI to operate on sequences.
2085 * fixed a bug in the build procedure: documentation of SBCL-specific
2086 packages is now preserved and available in the final Lisp image.
2087 * lifted FDEFINITION lookup out of loops in the implementation of
2088 many list operations. (thanks to Robert E. Brown)
2089 * fixed a bug in the reader: the #n# reader macro now works for
2090 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2091 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2092 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2093 E. Brown and Tony Martinez respectively)
2094 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2095 SLOT-BOUNDP: in particular, it is now able to optimize them much
2096 better, and is now not vulnerable to having packages renamed.
2097 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2098 always called when a slot is not present in an instance. (thanks
2100 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2101 longer signal an error on unreading a character following EOF on
2102 the previous constituent stream. (thanks to Tony Martinez)
2103 * fixed some bugs revealed by Paul Dietz' test suite:
2104 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2105 not just nonnegative fixnums;
2106 ** the logical bit-array operators such as BIT-AND now accept an
2107 explicit NIL for their "opt-arg" argument (to indicate a
2108 freshly-consed result bit-array);
2109 ** ELT now signals an error on an invalid sequence index in safe
2111 ** the type system is now cleverer about negations of numeric
2112 types, and consequently understands the BIGNUM and RATIO types
2114 ** the type system is now cleverer about the interaction between
2115 INTEGER and RATIO types: while bugs still remain, many more
2116 cases are accurately computed;
2117 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2118 if it is in the last clause;
2119 ** CONSTANTLY now correctly returns a side-effect-free function in
2121 ** DECLARE is no longer treated as a special-operator; in
2122 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2123 * incremented fasl file version number due to the change in the
2124 DEFSTRUCT-SLOT-DESCRIPTION structure.
2126 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2127 * minor incompatible change: code processed by the "interpreter" or
2128 EVAL now has a compilation optimization policy of (DEBUG 2)
2129 (changed from (DEBUG 1)) to improve debuggability of interactive
2130 development, and to allow the use of the debug RETURN command in
2132 * an experimental implementation of the RETURN command for the
2133 debugger has been included. (thanks to Frederik Kuivinen)
2134 * fixed bug 62: constraints were not propagated into a loop.
2135 * fixed bug in embedded calls of SORT (reported and investigated by
2137 * fixed some bugs revealed by Paul Dietz' test suite:
2138 ** printing and reading of arrays with some dimensions having
2139 length 0 (thanks to Gerd Moellmann);
2140 ** BOA constructor with &AUX argument without a default value does
2141 not cause a type error;
2142 ** CONSTANTP now returns true for all self-evaluating objects.
2144 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2145 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2146 accessors that are related by inheritance, as specified in the
2147 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2148 Valtteri Vuorikoski)
2149 * The compiler is now able to inline functions that were defined in
2150 a complex lexical environment (e.g. inside a MACROLET).
2151 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2153 * Support for the upcoming FreeBSD-5.0 release has been included.
2154 (thanks to Dag-Erling Smorgrav)
2155 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2156 effect when it is not in a toplevel context.
2157 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2158 stand a better chance of being correct. (thanks to Gerd
2160 * fixed bug in COERCE, which now signals an error on coercing a
2161 rational to a bounded real type which excludes the expected
2163 * The compiler is now able to derive types more accurately from the
2164 COERCE and COMPILE functions.
2165 * fixed bug 223: functional binding is considered to be constant
2166 only for symbols in the CL package.
2167 * fixed bug 231: SETQ did not check the type of a variable being set
2168 (reported by Robert E. Brown)
2169 * A new optimization for MAKE-INSTANCE has been included, fixing
2170 various bugs (including relating to :ALLOCATION :CLASS slots and
2171 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2173 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2174 ** As required by ANSI, LOOP now disallows anonymous collection
2175 clauses such as COLLECT I in conjunction with aggregate boolean
2176 clauses such as THEREIS (= I 1);
2177 ** LOOP now signals an error when any variable is reused in the
2178 same loop (including the potentially useful construct analogous
2179 to WITH A = 1 WITH A = (1+ A);
2180 ** IT is only a special loop symbol within the first clause of a
2181 conditional loop clause;
2182 ** LOOP with a typed iteration variable over a hashtable now
2183 signals a type error iff it should.
2184 * fixed some other bugs revealed by Paul Dietz' test suite:
2185 ** FILE-STREAM now names the class previously known as FD-STREAM;
2186 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2187 argument) no longer signals an error;
2188 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2189 of the slot symbol, rather than using the current package
2190 ((:CONC-NAME "") continues to intern the slot's name in the
2192 * incremented fasl file version number, because of the incompatible
2193 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2194 of the new implementation of DEFINE-COMPILER-MACRO.
2196 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2197 * Support for building SBCL for MIPS platforms running in
2198 little-endian mode has now been checked in, and basic
2199 functionality on said platforms verified.
2200 * minor incompatible change: PCL now records the pathname of a file
2201 in which methods and the like are defined, rather than its
2203 * minor incompatible change: TRUENAME now considers the truename of
2204 a file naming a directory to be the pathname with :DIRECTORY
2205 component indicating that directory.
2206 * minor incompatible change: a NAMED clause in the extended form of
2207 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2208 reason for the previous behaviour is unclear.
2209 * more systematization and improvement of CLOS and MOP conformance
2210 in PCL (thanks to Gerd Moellman and Pierre Mai):
2211 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2213 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2214 primary methods with no specializers;
2215 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2217 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2218 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2219 FORWARD-REFERENCED-CLASSes; error reporting on
2220 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2222 ** SXHASH on CLOS instances now uses a slot internal to the
2223 instance to return different numbers on distinct instances,
2224 while preserving the same return value through invocations of
2226 ** DEFMETHOD signals errors when methods with longer incongruent
2227 lambda lists are added to generic functions;
2228 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2229 CLASS, as specified in AMOP;
2230 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2231 on the order returned by the primary method for classes of
2232 class STANDARD-CLASS;
2233 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2234 * fixed some bugs shown by Paul Dietz' test suite:
2235 ** DOLIST puts its body in TAGBODY;
2236 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2238 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2239 value producing form;
2240 ** if more variables are given to PROGV than values, extra
2241 variables are bound and made to have no value;
2242 ** NSUBSTITUTE on list arguments gets the right answer with
2244 ** ELT signals an error of type TYPE-ERROR when the index argument
2245 is not a valid sequence index;
2246 ** LOOP signals (at macroexpansion time) an error of type
2247 PROGRAM-ERROR when duplicate variable names are found;
2248 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2249 ** FUNCALL of special-operators now cause an error of type
2251 ** PSETQ now works as required in the presence of side-effecting
2252 symbol-macro places;
2253 ** NCONC accepts any object as its last argument;
2254 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2256 ** loop-for-as-package does not require a package to be explicitely
2258 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2260 * fixed bug 166: compiler preserves "there is a way to go"
2261 invariant when deleting code.
2262 * fixed bug 172: macro lambda lists with required arguments after
2263 &REST arguments now cause an error to be signalled. (thanks to
2265 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2266 bug. (thanks to Matthew Danish)
2267 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2269 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2270 arguments are lexically rebound. (thanks to Gerd Moellmann and
2272 * fixed bug 194: error messages are now more informative when there
2273 is no primary method applicable in a call to a generic
2274 function. (thanks to Gerd Moellmann)
2275 * fixed bug in command line argument checking (thanks to Julian
2277 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2278 (thanks to Matthew Danish)
2279 * incremented fasl file version number, because of the
2280 SXHASH-related changes in the layout of CLOS data structures
2282 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2283 * minor incompatible change: The runtime (the Unix executable named
2284 "sbcl") is now much pickier about the .core files it will load.
2285 Essentially it now requires .core files to descend from the same
2286 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2287 as the runtime does. (The intent is to prevent the crashes which
2288 can occur, and which can even be reported as mysterious failures,
2289 when people patch the sources or change the build parameters
2290 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2291 sbcl and .core files.)
2292 * fixed bug: VALUES-LIST is no longer optimized away.
2293 * fixed bug 142: The FFI conversion of C string values to Lisp
2294 string values no longer conses excessively. (thanks to Nathan
2295 Froyd porting Raymond Toy's fix to CMU CL)
2296 * began to systematize and improve MOP conformance in PCL (thanks to
2297 Nathan Froyd, Gerd Moellman and Pierre Mai):
2298 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2300 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2301 ** FINALIZE-INHERITANCE is now called on class finalization;
2302 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2303 argument precedence order.
2304 * fixed bug 202: The compiler no longer fails on functions whose
2305 derived types contradict their declared type.
2306 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2307 so it can be non-toplevel.
2308 * The fasl file version number has changed (because of the new
2309 implementation of DEFMACRO).
2310 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2311 safe code, that any length requirement by their type-specifier
2312 argument is valid. The exceptions to this are described in bug
2314 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2315 functions now check the types of their inputs as required by ANSI.
2316 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2317 introduced symbol is DECLAREd to be SPECIAL.
2318 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2319 * fixed bug: :COUNT argument to sequence functions may be negative
2320 * fixed bug: body of DO-SYMBOLS may contain declarations
2321 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2322 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2323 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2324 symbol macro only once
2325 * fixed printing of call frame when argument list is unavailable
2326 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2327 * compiler no longer signals WARNING on unknown keyword
2330 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2331 * A beta-quality port to the mips architecture running Linux,
2332 based on the old CMUCL backend, has been made. It has been tested
2333 on a big-endian kernel, and works sufficiently well to be able to
2334 rebuild itself; it has not been tested in little-endian mode.
2335 * fixed an inconsistency between gencgc.c and purify.c which made
2336 dumping/loading .core files unreliable
2337 * fixed bug 120a: The compiler now deals correctly with IFs where
2338 the consequent is the same as the alternative, instead of
2339 misderiving the return type. (thanks to Alexey Dejneka)
2340 * fixed bug 113: Logical pathnames are now dumpable (the logical
2341 host is resolved at load-time, throwing an error if it is not
2343 * fixed bug 174: FORMAT's error message is slightly clearer when a
2344 non-printing character is used in a format directive.
2345 * fixed several bugs in compiler checking of type declarations, i.e.
2346 violations of the Python "declarations are assertions" principle
2347 (thanks to Alexey Dejneka)
2348 * fixed several bugs in PCL's error checking (thanks to Gerd
2350 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2352 * fixed bug in compilation of functions as first class values
2353 (thanks to Antonio Martinez-Shotton)
2354 * The compiler's handling TYPE-ERRORs which it can prove will
2355 inevitably happen at runtime has been cleaned up and corrected
2356 in several ways. (thanks to Alexey Dejneka)
2357 * improved argument type checking for various basic arithmetic
2358 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2359 lavished on them in the past that they can be compiled in many
2360 ways in different special cases
2361 * fixed bug 181: compiler checks validity of user supplied type
2363 * cleaned up code flushing in optimization: Function calls which
2364 should signal errors for safety purposes (e.g. which ANSI says
2365 should signal errors when their arguments are of incorrect type)
2366 are no longer optimized away.
2367 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2368 * incremented fasl file version number, because changes in the
2369 implementation of sequence functions like COERCE caused
2370 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2371 in old inline expansions) to become undefined. (Actually these
2372 changes were later undone, so we might very well be binary
2373 compatible with 0.7.7 after all, but leaving the version number
2374 incremented seemed like the simplest and most conservative
2377 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2378 * An alpha-quality port to the parisc architecture running Linux,
2379 based on the old CMUCL backend, has been made. This, even more so
2380 than the other backends, should be considered still a work in
2381 progress; known problems include that the Linux kernel in 64-bit
2382 mode does not propagate the correct sigcontext structure to
2383 userspace, and consequently SBCL on a parisc64 kernel will not
2385 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2386 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2387 suppressed.) Also now that the compiler is looking at declarations
2388 in the environment, it checks optimization declarations as well,
2389 and suppresses inlining when (> DEBUG SPEED).
2390 * More fixes have been made to treatment of floating point exception
2391 treatment and other Unix signals. In particular, floating point
2392 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2393 * The detection and handling of control stack exhaustion (infinite
2394 or very deeply nested recursion) has changed. Stack exhaustion
2395 detection is now done by write-protecting pages at the OS level
2396 and applies at all optimization settings; when found, a
2397 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2398 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2399 be caught using IGNORE-ERRORS.
2400 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2401 correctly when transforming recursive LABELS functions to LETs.
2402 (thanks to Alexey Dejneka)
2403 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2404 that are names of constants or global variables.
2405 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2406 alien routines with docstrings.
2407 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2408 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2410 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2411 (thanks to Eric Marsden)
2412 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2413 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2414 * Bug fix: The SPARC backend can now compile functions involving
2415 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2416 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2417 to a functions expecting a PATHNAME-DESIGNATOR.
2418 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2419 lambda lists. (thanks to Alexey Dejneka)
2420 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2421 functions so that EXPORTing the name of the class doesn't cause
2422 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2423 to Antonio Martinez for reporting this)
2424 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2425 constant-folded without failing an assertion. (thanks to Einar
2426 Floystad Dorum for reporting this)
2427 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2428 types are now dealt with more correctly by the compiler.
2429 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2430 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2431 behaviour analogous to RENAME-FILE. This puts its behaviour more
2432 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2434 * The fasl file version number has changed again. (because of the
2435 bug fix involving the names of PCL MAKE-INSTANCE functions)
2437 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2438 * bug fix: Floating point exceptions are treated much more
2439 consistently on the x86/Linux and PPC/Linux platforms.
2440 * Array initialization with :INITIAL-ELEMENT is now much faster for
2441 cases when the compiler cannot open code the array creation, but
2442 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2443 array accesses have also seen a speed increase.
2444 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2445 specified. (This was at the root of some bad interactions between
2446 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2448 * bug fix: Internal error arguments for undefined functions are now
2449 computed correctly on the PPC/Linux platform.
2450 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2451 Raymond Toy's patch for CMU CL)
2452 * Support for the Solaris 9 operating environment has been included
2453 (thanks to Daniel Merritt)
2454 * A very ugly but hopefully complete draft of the missing FFI chapter
2455 of the manual has been created by reformatting the corresponding
2456 CMU CL manual chapter into (currently very ugly and incoherent)
2457 DocBook and bringing it up to date for SBCL behavior. Thus, the
2458 manual is now essentially complete, at least by my extreme
2459 once-and-only-once standards, whereby it's acceptable to refer to
2460 the doc strings of SB-EXT functions as the primary documentation.
2461 * The fasl file version number has changed again, due to cleanup of
2462 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2463 is no longer a static symbol.)
2465 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2466 * SBCL now builds with OpenMCL (version 0.12) as the
2467 cross-compilation host; also, more progress has been made toward
2468 bootstrapping under CLISP.
2469 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2471 * bug 158 fixed: The compiler can now deal with integer loop
2472 increments different from 1; fixing this turned out also to fix
2474 * bug 169 fixed: no more bogus warnings about using lexical bindings
2475 despite the presence of perfectly good SPECIAL declarations (thanks
2476 to David Lichteblau)
2477 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2478 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2479 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2481 * bug 180 fixed: Method combination specifications no longer ignore
2482 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2483 * bug fix: Structure type predicate functions now check their argument
2484 count as they should.
2485 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2486 correctly. (thanks to Pierre Mai)
2487 * minor incompatible change: The --noprogrammer option is deprecated
2488 in favor of the new --disable-debugger option, which is very similar.
2489 (The major difference is that it takes effect at a slightly different
2490 time at startup, causing handling of errors in --sysinit and
2491 --userinit files will be affected differently.) The
2492 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2493 been added to allow this functionality to be controlled from ordinary
2494 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2495 maintainers, who might want to run non-interactive scripts to
2496 build SBCL cores which will later be used interactively.)
2497 * minor incompatible change: The LOAD function no longer, when given
2498 a wild pathname to load, loads all files matching that pathname.
2499 Instead, an error of type FILE-ERROR is signalled.
2501 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2502 * bug 147 fixed: The compiler preserves its block link/count
2503 invariants more correctly now so that it doesn't crash. (thanks
2505 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2507 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2508 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2509 no-value-for-this-character as NIL (as specified by ANSI).
2510 * HOST-NAMESTRING on physical pathnames now returns a string that is
2511 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2512 (thanks to Christophe Rhodes)
2513 * The Alpha port handles icache flushing more correctly. (thanks to
2515 * More progress has been made toward bootstrapping under CLISP. (thanks
2516 to Christophe Rhodes)
2517 * The fasl file format has changed again, because dynamic loading
2518 on OpenBSD (which has non-ELF object files) motivated some cleanups
2519 in the way that foreign symbols are transformed and passed around.
2520 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2521 is no longer treated as whitespace by the reader, but instead as
2522 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2523 |A
\7fB|, instead of A as it used to.
2525 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2526 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2527 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2528 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2529 this as of 0.7.1.45, but was left out of the previous news section
2530 (thanks to Dan Barlow)
2531 * SBCL now runs on the Solaris operating system on SPARC architectures
2532 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2533 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2534 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2535 * SPARC backend cleanups, allowing builds of cores optimized for V8
2536 and V9 SPARCS, and also emission of code targeted to a particular
2537 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2539 * SBCL is closer to bootstrapping under CLISP, thanks to various
2540 fixes by Christophe Rhodes.
2541 * The fasl file format has changed again, to allow the compiler's
2542 INFO database to support symbol macros.
2543 * The user manual (in doc/) is formatted into HTML more nicely.
2544 (thanks to coreythomas)
2545 * The system is smarter about SUBTYPEP relationships, especially
2546 those involving NOT types (including types such as ATOM which are
2547 represented internally using NOT types). Thus SUBTYPEP is less
2548 likely to return (VALUES NIL NIL) in general, and in particular
2549 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2551 * The fasl file format has changed again, because the internal
2552 representation of types now includes a new slot to support the new
2553 SUBTYPEP-of-NOT-types logic.
2554 * (not a change in the main branch of SBCL, but a related prototype
2555 which can hopefully be merged into the main branch of SBCL in the
2556 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2557 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2558 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2559 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2560 dispatch macro character combinations. (thanks to Alexey Dejneka)
2561 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2562 that superficially look like logical namestrings correctly.
2563 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2565 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2566 default supplied-p) for &optional and &key arguments. (thanks to
2569 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2570 * incompatible change: The compiler is now less aggressive about
2571 tail call optimization, doing it only when (> SPACE DEBUG) or
2572 (> SPEED DEBUG). (This is an incompatible change because there are
2573 programs which relied on the old CMU-CL-style behavior to optimize
2574 away their unbounded recursion which will now die of stack overflow.)
2575 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2576 for non-GENCGC systems has been increased to 20M (since that
2577 seems much closer to the likely performance optimum for modern
2578 systems than the old 4M value was)
2579 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2580 and *DEBUG-PRINT-LEVEL*
2581 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2582 of CMU CL's support for SPARC, and various endianness and other
2583 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2584 * new syntactic sugar for the Unix command line: --load foo.bar is now
2585 an alternate notation for --eval '(load "foo.bar")'.
2587 ** The system now detects stack overflow and handles it gracefully,
2588 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2589 optimization settings. (This is a good thing in general, and
2590 its introduction in this version should be particularly timely
2591 for anyone whose code fails because of suppression of tail
2593 ** The system now hunts for the C variable "environ" in a more
2594 devious way, to avoid segfaults when the C library version differs
2595 between compile time and run time. (thanks to Christophe Rhodes)
2596 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2597 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2598 ** The compiler no longer issues bogus style warnings for undefined
2599 classes in the same source file as the DEFCLASSes which defined
2600 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2602 ** fixes in CONDITION class precedence list for undefined function
2603 errors (thanks to Alexei Dejneka)
2604 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2605 correctly. (thanks to Dan Barlow)
2606 ** portability fixes aiming at bootstrapping under CLISP (thanks
2607 to Dave McDonald and Christophe Rhodes)
2608 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2609 ** fixes in type translation and and type inference (thanks to
2611 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2612 ** various fixes in the new ports (thanks to Dan Barlow)
2613 * several changes related to debugging:
2614 ** suppression of tail recursion, as noted above
2615 ** stack overflow detection, as noted above
2616 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2617 is now the default. (For some time encapsulation has been more
2618 reliable than the breakpoint-based :ENCAPSULATE NIL
2619 implementation, at least on X86 systems; and I just noticed that
2620 encapsulation also seems closer to the spirit of the ANSI
2623 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2625 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2626 up properly again. (There was a packaging bug in 0.7.0 which
2627 left their definitions in SB-SYS::LOAD-FOREIGN and
2628 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2629 vital for most things which interface to C-level interfaces,
2630 like extensions working with sockets or databases or
2631 Perl-compatible regexes or whatever, and the need to fix
2632 this bug is the main reason that 0.7.1 was released so
2634 ** DEFGENERIC is now choosier about the methods it redefines, so that
2635 reLOADing a previously-LOADed file containing DEFGENERICs does
2636 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2637 cycle now works as it should. (thanks to Alexey Dejneka)
2638 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2639 fixed by Christophe Rhodes. (He actually submitted this patch
2640 months ago, and I delayed until after 0.7.0.)
2641 ** Bug 111 (internal compiler confusion about runtime checks on
2642 FUNCTION types) was fixed by Alexey Dejneka.
2643 * Some internal cleanups (getting rid of variables which aren't
2644 needed now that the byte interpreter is gone) caused the fasl
2645 file format number to change again.
2647 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2648 * major incompatible change: The default fasl file extension, i.e. the
2649 default extension for files produced by COMPILE-FILE, has changed
2650 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2652 ** There are many changes in the implementation of the compiler.
2653 SBCL is now essentially a compiler-only implementation of ANSI
2654 Common Lisp. EVAL still "interprets" a few special cases, but
2655 almost all the interesting cases are handled by creating
2656 a LAMBDA expression, calling COMPILE on it, then calling
2657 FUNCALL on the result.
2658 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2659 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2660 Since the code is newer, there might still be some new bugs
2661 (though not as many as before Martin Atzmueller's fixes:-). But
2662 the new code is substantially simpler and clearer, and hopefully
2663 any remaining bugs will be simpler, less fundamental, and more
2664 fixable then the bugs in the old code.
2665 ** The revised compiler is still a little unsteady on its feet.
2667 *** The debugging information it produces (particularly the names
2668 of FUNCTION objects) is sometimes much less useful than what
2669 the old compiler produced.
2670 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2671 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2672 within a MACROLET) has been temporarily weakened.
2673 ** There are new compiler optimizations for various functions:
2674 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2675 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2676 *** the math functions TRUNCATE, FLOOR, and CEILING
2677 *** the function-of-all-trades COERCE
2678 Mostly these should be transparent, but there's one
2679 potentially-annoying problem (bug 117): when the compiler
2680 inline-expands a function and does type analysis on the result,
2681 it can create control paths which have type mismatches, and
2682 when it can't prove that those control paths aren't taken,
2683 it will issue WARNINGs about the type mismatches. This is
2684 a particular problem in practice for the new sequence functions.
2685 It's not clear how this should be fixed, and for now, a
2686 workaround is given in the entry for 117 in the BUGS file.
2687 ** (Because of the interaction between the two previous items --
2688 occasional inlining problems and new inline expansions -- some
2689 of the new sequence function optimizations won't really kick in
2690 completely until debugging information, and then inlining, are
2691 straightened out in some future version.)
2692 * minor incompatible changes:
2693 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2694 DIRECTORY no longer implicitly promotes NIL slots of its
2695 pathname argument to :WILD. In particular, when you ask for the
2696 contents of a directory (which you used to be able to do without
2697 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2698 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2699 ** changes in behavior that ANSI explicitly defines to be
2700 implementation dependent:
2701 *** The new compiler-only implementation still conforms with ANSI,
2702 but acts a little different than before. Besides the obvious
2703 changes in performance tradeoffs (that the cost per form passed
2704 to EVAL has gone up, and the cost per form executed by EVAL
2705 has gone down), the behavior of the system changes a little
2706 because there are no longer any interpreted function objects.
2707 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2708 e.g. doing COMPILE on the output of interactive DEFUN is
2710 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2712 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2713 has changed from (:CL) to NIL.
2714 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2715 e.g. #\Newline and #\Tab, don't have names specified in the
2716 ANSI Common Lisp standard, is now based on their ASCII symbolic
2717 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2718 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2719 are no longer used for output.
2720 ** changes in internal implementation constants:
2721 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2722 4 million. (If your application spends a lot of time GCing and
2723 you have a lot of RAM, you might want to experiment with
2724 increasing it even more.)
2725 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2726 However, almost all old code should still continue to work without
2727 immediate update, as SB-C-CALL is now a (deprecated) nickname
2729 ** Old operator names in the style DEF-FOO are now deprecated in
2730 favor of new corresponding names DEFINE-FOO, for consistency with
2731 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2732 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2733 internal symbols, but a few supported extensions like
2734 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2735 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2736 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2737 etc. as you get deeper into recursive calls to the debugger
2738 command loop, instead of the old "5]", "5]]", "5]]]"
2739 sequence. (I was motivated to do this when squabbles between
2740 ILISP and SBCL left me very deeply nested in the debugger. In the
2741 short term, this change will probably provoke more ILISP/SBCL
2742 squabbles, but hopefully it will be an improvement in the long run.)
2743 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2744 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2745 function, since declaiming return types involving aliens is
2746 (1) annoyingly messy to do by hand and (2) vital to efficient
2747 compilation of code which calls such functions.
2748 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2749 longer reexported by the SB-EXT package. They're solely useful
2750 for alien code, so it seems more logical that you should get
2751 them from the SB-ALIEN package, not in SB-EXT.
2752 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2753 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2754 features. Instead, the code that they used to control is always
2755 built into the system.
2756 * many other bug fixes
2757 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2758 advantage of the new EVAL-WHEN stuff and to clean them up in
2759 general, and they are now more ANSI-compliant in a number of
2760 ways. Martin Atzmueller is responsible for a lot of this.
2761 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2763 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2764 *** correct ERROR type for various file operations
2765 *** some fixes for Lisp streams
2766 *** DEFMETHOD syntax checking
2767 *** changing old weird representation of debug information as
2768 strings (which, among their other deficiencies, don't transform
2769 correctly when you rename packages, and don't change their
2770 print representation when you change things like *PACKAGE*
2771 and *PRINT-LENGTH*) to symbols and lists of symbols
2772 He also made several improvements and fixed several bugs in DESCRIBE.
2773 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2775 *** misbehavior of WRITE-STRING/WRITE-LINE
2776 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2777 and several other LOOP problems as well
2778 *** DIRECTORY when similar filenames are present
2779 *** DEFGENERIC with :METHOD options
2780 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2781 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2782 *** argument ordering in FIND with :TEST option
2783 *** mishandled package designator argument in APROPOS-LIST
2784 *** various problems in the backquote readmacro
2785 *** a bug in APROPOS
2786 *** probably some others that I'm not describing very well here,
2787 since the CVS log documents them by reference to sbcl-devel
2788 messages, and the SourceForge archives aren't working well.:-(
2789 ** Dan Barlow improved the Alpha port (and is making progress on the
2790 PPC port, for those of you who think different).
2791 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2792 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2793 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2794 continues to work on the SPARC port (for those of you in a position
2795 to look down upon our little PC-compatible boxes from a great height).
2796 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2797 and uses that copy, rather than the current dynamic value, when
2798 it's trying to decide whether to truncate output. Thus e.g.
2799 (let ((*print-lines* 50))
2800 (pprint-logical-block (stream nil)
2802 (let ((*print-lines* 8))
2803 (print (aref possiblybigthings i) stream)))))
2804 should now truncate the logical block only at 50 lines, instead of
2805 often truncating it at 8 lines, as it did before.
2806 * The doc/cmucl/ directory, containing old CMU CL documentation
2807 from the time of the fork, is no longer part of the base system.
2808 SourceForge has shut down its anonymous FTP service, and with it
2809 my original plan for distributing the old CMU CL documentation
2810 there. For now, if you need these files you can download an old
2811 SBCL source release and extract them from it.
2812 * The fasl file version number changed again, for dozens of reasons,
2813 some of which are apparent above.
2815 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2816 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2817 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2818 patch, so that DISASSEMBLE output is much nicer.
2819 * The code in the SB-PROFILE package now seems reasonably stable.
2820 I still haven't decided what the final interface should look like
2821 (I'd like PROFILE to interact cleanly with TRACE, since both
2822 facilities use function encapsulation) but if you have a need
2823 for profiling now, you can probably use it successfully with
2824 the current CMU-CL-style interface.
2825 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2826 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2827 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2828 link now returns the dangling link itself, and for similar
2829 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2830 link itself. (In these cases the old code signalled an error and
2831 looped endlessly, respectively.) Thus, DIRECTORY now works even
2832 in the presence of dangling and cyclic symbolic links.
2833 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2834 is now a supported extension again, since the consensus on
2835 sbcl-devel was that it can be useful for ordinary development
2836 work, not just for debugging SBCL itself.
2837 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2838 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2839 that currently-defined functions might be redefined later with
2840 different return types.
2841 * Hash tables can be printed readably, as inspired by CMU CL code
2842 of Eric Marsden and SBCL code of Martin Atzmueller.
2843 * better error handling in CLOS method combination, thanks to
2844 Martin Atzmueller porting Pierre Mai's CMU CL patches
2845 * more overflow fixes for >16Mbyte I/O buffers
2846 * A bug in READ has been fixed, so that now a single Ctrl-D
2847 character suffices to cause end-of-file on character streams.
2848 In particular, now you only need one Ctrl-D at the command
2849 line (not two) to exit SBCL.
2850 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2852 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2853 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2854 does the right thing.
2855 * The implementation of some type tests, especially for CONDITION
2856 types, is now tidier and maybe faster, due to CMU CL code
2857 originally by Douglas Crosher, ported by Martin Atzmueller.
2858 * Some math functions have been fixed, and there are new
2859 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2860 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2861 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2862 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2863 likely to be built into sbcl-0.7.0.)
2864 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2865 CL, ported to SBCL by Martin Atzmueller)
2866 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2867 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2868 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2869 * There's a new slam.sh hack to shorten the edit/compile/debug
2870 cycle for low-level changes to SBCL itself, and a new
2871 :SB-AFTER-XC-CORE target feature to control the generation of
2872 the after-xc.core file needed by slam.sh.
2873 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2874 COMPILE-FILE is no longer supported, so that now every function
2875 gets an entry point, so that block compilation looks a little
2876 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2877 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2878 returns the number of bytes consed since the system started,
2879 rather than the number consed since the first time the function
2880 was called. (The new definition parallels ANSI functions like
2881 CL:GET-INTERNAL-RUN-TIME.)
2882 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2883 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2884 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2885 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2886 * Fasl file version numbers are now independent of the target CPU,
2887 since historically most system changes which required version
2888 number changes have affected all CPUs equally. Similarly,
2889 the byte fasl file version is now equal to the ordinary
2892 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2893 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2894 is no longer recognized. I apologize for this, because it was
2895 listed in SB-EXT as a supported extension, but I found that
2896 its existing behavior was poorly specified, as well as incorrectly
2897 specified, and it looked like too much of a mess to straighten it
2898 out. I have enough on my hands trying to get ANSI stuff to work..
2899 * many patches ported from CMU CL by Martin Atzmueller, with
2900 half a dozen bug fixes in pretty-printing and the debugger, and
2901 half a dozen others elsewhere
2902 * fixed bug 13: Floating point infinities are now supported again.
2903 They might still be a little bit flaky, but thanks to bug reports
2904 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2905 as flaky as they were.
2906 * The --noprogrammer command line option is now supported. (Its
2907 behavior is slightly different in detail from what the old man
2908 page claimed it would do, but it's still appropriate under the
2909 same circumstances that the man page talks about.)
2910 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2911 are now supported, and enabled by default. Thus, the compiler can
2912 handle many floating point and complex operations much less
2913 inefficiently. (Thus e.g. you can implement a complex FFT
2915 * The compiler now detects type mismatches between DECLAIM FTYPE
2916 and DEFUN better, and implements CHECK-TYPE more correctly, and
2917 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2918 thanks to patches from Martin Atzmueller.
2919 * various fixes to make the cross-compiler more portable to
2920 ANSI-conforming-but-different cross-compilation hosts (notably
2921 Lispworks for Windows, following bug reports from Arthur Lemmens)
2922 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2923 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2924 fixed thanks to Pierre Mai's CMU CL patches.
2925 * fixes to keep the system from overflowing internal counters when
2926 it tries to use i/o buffers larger than 16M bytes
2927 * fixed bug 45a: Various internal functions required to support
2928 complex special functions have been merged from CMU CL sources.
2929 (When I was first setting up SBCL, I misunderstood a compile-time
2930 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2931 * improved support for type intersection and union, fixing bug 12
2932 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2933 more obscure bugs as well
2934 * some steps toward byte-compiling non-performance-critical
2935 parts of the system, courtesy of patches from Martin Atzmueller
2936 * Christophe Rhodes has made some debian packages of sbcl at
2937 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2938 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2939 stable, but are nonetheless usable. When he's ready, I'd be happy
2940 to add them to the SourceForge "File Releases" section. (And if
2941 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2942 * new fasl file format version number (because of changes in
2943 internal representation of (OR ..) types to accommodate the new
2944 support for (AND ..) types, among other things)
2946 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2947 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2949 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2951 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2952 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2953 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2954 * DESCRIBE now works on structure objects again.
2955 * Most function call argument type mismatches are now handled as
2956 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2957 know whether the function will be redefined before the call is
2958 executed. (The compiler could flag local calls with full WARNINGs,
2959 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2960 it doesn't keep track of enough information to know whether calls
2961 are local in this sense.)
2962 * Compiler output is now more verbose, with messages truncated
2963 later than before. (There should be some supported way for users
2964 to override the default verbosity, but I haven't decided how to
2965 provide it yet, so this behavior is still controlled by the internal
2966 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2967 src/compiler/ir1util.lisp.)
2968 * Fasl file format version numbers have increased again, because
2969 support for the Gray streams extension changes the layout of the
2970 system's STREAM objects.
2971 * The Gray subclassable streams extension now works, thanks to a
2972 patch from Martin Atzmueller.
2973 * The full LOAD-FOREIGN extension (not just the primitive
2974 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2975 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2976 but like most other programs, it defaults to copying the Unix
2977 environment from the original process instead of starting the
2978 new process in an empty environment.
2979 * Extensions which manipulate the Unix environment now support
2980 an :ENVIRONMENT keyword option which doesn't smash case or
2981 do other bad things. The CMU-CL-style :ENV option is retained
2982 for porting convenience.
2983 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2984 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2986 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2988 * A patch from Martin Atzmueller seems to have solved the SIGINT
2989 problem, and as far as we know, signal-handling now works cleanly.
2990 (If you find any new bugs, please report them!)
2991 * The system no longer defaults Lisp source file names to types
2992 ".l", ".cl", or ".lsp", but only to ".lisp".
2993 * The compiler no longer uses special default file extensions for
2994 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2995 seems to expect a single default extension for all compiled code,
2996 and there's no compelling reason to try to stretch the standard
2997 to allow two different extensions.) Instead, byte-compiled files
2998 default to the same extension as native-compiled files.
2999 * Fasl file format version numbers have increased again, because
3000 a rearrangement of internal implementation packages made some
3001 dumped symbols in old fasl files unreadable in new cores.
3002 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3003 many fewer weird special cases.
3004 * Bug #17 (differing COMPILE-FILE behavior between logical and
3005 physical pathnames) has been fixed, and some related misbehavior too,
3006 thanks to a patch from Martin Atzmueller.
3007 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3008 by Tim Moore, ported to SBCL by Martin Atzmueller.
3009 * Martin Atzmueller fixed several filesystem-related problems,
3010 including bug #36, in part by porting CMU CL patches, which were
3011 written in part by Paul Werkowski.
3012 * More compiler warnings in src/runtime/ are gone, thanks to
3013 more patches from Martin Atzmueller.
3014 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3017 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3019 * DESCRIBE now works on CONDITION objects.
3020 * The debugger now handles errors which arise when trying to print
3021 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3023 * The build system now uses an additional file, customize-target-features.lisp,
3024 to allow local modifications to the target *FEATURES* list. (The point of
3025 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3026 debugging features enabled, without having to worry about propagating it
3027 into everyone's system when I do a "cvs update".) When no
3028 customize-target-features.lisp file exists, the target *FEATURES* list
3029 should be constructed the same way as before.
3030 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3031 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3032 warn about behavior which is undefined under the ANSI spec. Note: This
3033 is specified by ANSI, but it's not very popular with programmers.
3034 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3035 macro in the SBCL sources for an example of a workaround which you
3036 might use to make portable ANSI-standard code which does what you want.
3037 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3038 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3039 when it's used as a non-toplevel form. (This is required in order
3040 to implement the DEFCONSTANT-EQX macro.)
3041 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3042 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3043 funny things, due to bugs in the implementation of EVAL-WHEN
3044 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3045 non-toplevel DEFCONSTANTs are uncommon.)
3046 * The core file version number and fasl file version number have been
3047 incremented, because the old noncompliant DEFCONSTANT behavior involved
3048 calling functions which no longer exist, and because I also took the
3049 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3050 * fixed bug 1 (error handling before read-eval-print loop starts), and
3051 redid debugger restarts and related debugger commands somewhat while
3053 ** The QUIT debugger command is gone, since it did something
3054 rather different than the SB-EXT:QUIT command, and since it never
3055 worked properly outside the main toplevel read/eval/print loop.
3056 Invoking the new TOPLEVEL restart provides the same functionality.
3057 ** The GO debugger command is also gone, since you can just invoke
3058 the CONTINUE restart directly instead.
3059 ** The TOP debugger command is also gone, since it's redundant with the
3060 FRAME 0 command, and since it interfered with abbreviations for the
3062 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3064 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3065 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3066 some porting work by Martin Atzmueller)
3067 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3068 with a boatload of patches which he ported to SBCL. Now that those
3070 ** The system tries to make sure that its low-priority messages
3071 are prefixed by semicolons, to help people who like to use
3072 syntax highlighting in their ILISP buffer. (This patch
3073 was originally due to Raymond Toy.)
3074 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3075 patches originally written by Raymond Toy.
3076 ** The compiler understands coercion between single-value and
3077 multiple-VALUES type expressions better, getting rid of some very
3078 weird behavior, thanks to patches originally by Robert MacLachlan
3079 and Douglas Crosher.
3080 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3081 lambda lists, thanks to a patch originally by Pierre Mai.
3082 ** The system no longer bogusly warns about "abbreviated type
3084 ** The compiler gets less confused by inlining and RETURN-FROM,
3085 thanks to some patches originally by Tim Moore.
3086 ** The system no longer hangs when dumping circular lists to fasl
3087 files, thanks to a patch originally from Douglas Crosher.
3088 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3089 undefined function error.
3090 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3091 in structure slot initforms) for now, documented workaround instead:-|
3092 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3093 * fixed bug 5: added stubs for various Gray stream functions called
3094 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3095 installed, at least appropriate type errors are generated
3096 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3097 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3098 consistently in DEFMETHOD forms.
3099 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3100 it doesn't seem to affect SBCL after all
3101 * The C runtime system now builds with better optimization and many
3102 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3104 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3106 * The system is now under CVS at SourceForge (instead of the
3107 CVS repository on my home machine).
3108 * The new signal handling code has been tweaked to treat register
3109 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3110 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3111 downstream, e.g. in debug-int.lisp, has implicit dependencies
3112 on the unsignedness of integer representation of machine words,
3113 and that caused the system to bomb out with infinite regress
3114 when trying to recover from type errors involving signed values,
3115 e.g. (BUTLAST '(1 2 3) -1).)
3116 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3117 (This was one of the bugs Peter Van Eynde reported back in July.)
3118 * The system now uses code inspired by Colin Walters' O(N)
3119 implementation of MAP (from the cmucl-imp@cons.org mailing
3120 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3121 inline the MAP operation, and there is more than one
3122 sequence argument to the MAP call (so that it can't just
3123 do ETYPECASE once and for all based on the type of the
3124 single sequence argument). (The old non-inline implementation
3125 of the general M-argument sequence-of-length-N case required
3126 O(M*N*N) time when any of the sequence arguments were LISTs.)
3127 * The QUIT :UNIX-CODE keyword argument has been renamed to
3128 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3129 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3131 * Raymond Wiker's patches to port dynamic loading from Linux to
3132 FreeBSD have been added.
3133 * The BUGS file is now more nearly up to date, thanks in large part
3134 to Martin Atzmueller's review of it.
3135 * The debugger now flushes standard output streams before it begins
3136 its output ("debugger invoked" and so forth).
3137 * The core version number and fasl file version number have both
3138 been incremented, because of incompatible changes in the layout
3140 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3141 used to not be called for a saved Lisp image.)
3142 * Martin Atzmueller's version of a patch to fix a compiler crash,
3143 as posted on sbcl-devel 13 September 2000, has been installed.
3144 * Instead of installing Martin Atzmueller's patch for the
3145 compiler transform for SUBSEQ, I deleted the compiler transform,
3146 and transforms for some similar consing operations.
3147 * A bug in signal handling which kept TRACE from working on OpenBSD
3149 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3150 away by constant folding
3151 * The system now defines its address space constants in one place
3152 (in the Lisp sources), and propagates them automatically elsewhere
3153 (through GENESIS and the sbcl.h file). Therefore, patching the
3154 address map is less unnecessarily tedious and error-prone. The
3155 Lisp names of address space constants have also been systematized.
3156 * CVS tags like dollar-Header-dollar have been removed from
3157 the sources, because they have never saved me trouble and
3158 they've been source of trouble working with patches and other
3159 diff-related operations.
3160 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3161 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3163 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3165 * The system has been ported to OpenBSD.
3166 * The system now compiles with a simple "sh make.sh" on the systems
3167 that it's supported on. I.e., now you no longer need to tweak
3168 text in the target-features.lisp-expr and symlinks in src/runtime/
3169 by hand, the make.sh takes care of it for you.
3170 * The system is no longer so grossly inefficient when compiling code
3171 involving vectors implemented as general (not simple) vectors (VECTOR T),
3172 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3173 which dares to use the various sequence functions on non-simple
3174 vectors, takes less of a performance hit.
3175 * There is now a primitive type predicate VECTOR-T-P
3176 to test for the (VECTOR T) type, so that e.g.
3177 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3178 can now be compiled with some semblance of efficiency. (The old code
3179 turned the type declaration into a full call to %TYPEP at runtime!)
3180 * AREF on (VECTOR T) is still not fast, since it's still compiled
3181 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3182 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3183 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3184 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3185 without making full calls to SUBTYPEP at runtime.
3186 (Some analogous efficiency issues for non-simple vectors specialized to
3187 element types other than T, or for non-simple multidimensional arrays,
3188 have not been addressed. They could almost certainly be handled the
3189 same way if anyone is motivated to do so.)
3190 * The changes in array handling break binary compatibility, so
3191 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3192 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3193 instead of (VALUES T T).
3194 * By following the instructions that Dan Barlow posted to sbcl-devel
3195 on 2 July 2000, I was able to enable primitive dynamic object
3196 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3197 functionality is not implemented (since it calls ld to resolve
3198 library references automatically, requiring RUN-PROGRAM for its
3199 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3200 try to resolve library references) is now supported.
3201 * The system now flushes the standard output streams when it terminates,
3202 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3203 them at several other probably-convenient times, e.g. in each pass of
3204 the toplevel read-eval-print loop, and after evaluating a form given
3205 as an "--eval" command-line option. (These changes were motivated by a
3206 discussion of stream flushing issues on cmucl-imp in August 2000.)
3207 * The source transform for TYPEP of array types no longer assumes
3208 that an array whose element type is a not-yet-defined type
3209 is implemented as an array of T, but instead punts, so that the
3210 type will be interpreted at runtime.
3211 * There is now some support for cross-compiling in make.sh: each of
3212 the phases of make.sh has its own script. (This should be transparent
3213 to people doing ordinary, non-cross-compile builds.)
3214 * Since my laptop doesn't have hundreds of megabytes of memory like
3215 my desktop machine, I became more motivated to do some items on
3216 my to-do list in order to reduce the size of the system a little:
3217 ** Arrange for various needed-only-at-cold-init things to be
3218 uninterned after cold init. To support this, those things have
3219 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3220 symbols with such names are now uninterned after cold init).
3221 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3222 fasl files for cold load.
3223 ** Remove the old compiler structure pooling code (which used to
3224 be conditional on the target feature :SB-ALLOC) completely.
3225 ** Redo the representation of some data in cold init to be more compact.
3226 (I also looked into supporting byte compiled code at bootstrap time,
3227 which would probably reduce the size of the system a lot, but that
3228 looked too complicated, so I punted for now.)
3229 * The maximum signal nesting depth in the src/runtime/ support code has
3230 been reduced from 4096 to 256. (I don't know any reason for the very
3231 large old value. If the new smaller value turns out to break something,
3232 I'll probably just bump it back up.)
3233 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3235 * Many, many bugs reported by Peter Van Eynde have been added to
3236 the BUGS list; some have even been fixed.
3237 * While enabling dynamic object file loading, I tried to make the
3238 code easier to understand, renaming various functions and variables
3239 with less ambiguous names, and changing some function calling
3240 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3241 * While trying to figure out how to do the OpenBSD port, I tried to
3242 clean up some of the code in src/runtime/. In particular, I dropped
3243 support for non-POSIX signal handling, added various comments,
3244 tweaked the code to reduce the number of compilation warnings, and
3245 renamed some files to increase consistency.
3246 * To support the new automatic configuration functionality in make.sh,
3247 the source file target-features.lisp-expr has been replaced with the
3248 source file base-target-features.lisp-expr and the machine-generated
3249 file local-target-features.lisp-expr.
3250 * fixed a stupid quoting error in make.sh so that using CMU CL
3251 "lisp -batch" as cross-compilation host works again
3253 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3255 * DESCRIBE no longer tries to call itself recursively to describe
3256 bound/fbound values, so that it no longer fails on symbols which are
3257 bound to themselves (like keywords, T, and NIL).
3258 * DESCRIBE now works on generic functions.
3259 * The printer now prints less-screwed-up representations of closures
3260 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3261 * A private symbol is used instead of the :EMPTY keyword previously
3262 used to mark empty slots in hash tables. Thus
3263 (DEFVAR *HT* (MAKE-HASH-TABLE))
3264 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3265 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3266 now does what ANSI says that it should. (You can still get
3267 similar noncompliant behavior if bang on the hash table
3268 implementation with all the symbols you get back from
3269 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3270 This breaks binary compatibility, since tests for equality to
3271 :EMPTY are wired into things like the macroexpansion of
3272 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3274 * There's now a minimal placeholder implementation for CL:STEP,
3275 as required by ANSI.
3276 * An obscure bug in the interaction of the normal compiler, the byte
3277 compiler, inlining, and structure predicates has been patched
3278 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3279 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3281 * Missing ordinary arguments in a macro call are now detected even
3282 when the macro lambda list contains &KEY or &REST.
3283 * The debugger no longer complains about encountering the top of the
3284 stack when you type "FRAME 0" to explicitly instruct it to go to
3285 the top of the stack. And it now prints the frame you request even
3286 if it's the current frame (instead of saying "You are here.").
3287 * As specified by ANSI, the system now always prints keywords
3288 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3290 * The default initial SIZE of HASH-TABLEs is now smaller.
3291 * Type information from CLOS class dispatch is now propagated
3292 into DEFMETHOD bodies, so that e.g.
3293 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3295 is now basically equivalent to
3296 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3297 (DECLARE (TYPE SINGLE-FLOAT X))
3299 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3300 operation, without having to do run-time type dispatch.
3301 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3302 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3303 * Since I don't seem to be making big file reorganizations very often
3304 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3305 is overflowing my ability to conveniently back them up), I've finally
3306 checked the system into CVS. (The CVS repository is on my home system,
3307 not at SourceForge -- putting it on SourceForge might come later.)
3308 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3309 high-level GC-NOTIFY-FOO functions send their output. (There's
3310 still very little control of where low-level verbose GC functions
3311 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3312 less than it used to -- the GC-NOTIFY-FOO functions are now under
3313 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3314 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3315 in only one place in the source code, and propagates it automatically
3316 everywhere that it's needed. Thus e.g. when I bump the version from
3317 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3318 * The C source files now include boilerplate legalese and documentation
3319 at the head of each file (just as the Lisp source files already did).
3320 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3321 to his page will be replaced with a link to his new CLiki service.
3323 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3325 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3326 * The build process now looks for GNU make under the default name "gmake",
3327 instead of "make" as it used to. If GNU make is not available as "gmake"
3328 on your system, you can change this default behavior by setting the
3329 GNUMAKE environment variable.
3330 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3331 can build without error under CMU CL.
3333 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3335 * There is now a partial SBCL user manual (with some new text and some
3336 text cribbed from the CMU CL manual).
3337 * The beginnings of a profiler have been added (starting with the
3338 CMU CL profiler and simplifying and cleaning up). Eventually the
3339 main interface should be through the TRACE macro, but for now,
3340 it's still accessed through vaguely CMU-CL-style functions and macros
3341 exported from the package SB-PROFILE.
3342 * Some problems left over from porting CMU CL to the new
3343 cross-compilation bootstrap process have been cleaned up:
3344 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3345 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3346 part when getting rid of the compiler *BACKEND* stuff.)
3347 ** The value of *NULL-TYPE* was screwed up, because it was
3348 being initialized before the type system knew the final
3349 definition of the 'NULL type. This screwed up several key
3350 optimizations in the compiler, causing inefficiency in all sorts
3351 of places. (I found it because I wanted to understand why
3352 GET-INTERNAL-RUN-TIME was consing.)
3353 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3354 PROCLAIM FTYPE information. Unfortunately this broke binary
3355 compatibility again, since now the forms output by DEFGENERIC
3356 to refer to functions which didn't exist in 0.6.3.
3357 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3358 can use the new (as of 0.6.3) transform for SOME into MAP into
3360 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3361 values to (INTEGER 0 1000000), so that the time code will no longer
3362 occasionally get blown up by Linux returning 1000000 microseconds
3363 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3364 its output conform to the ANSI spec. (Alas, this makes its output
3365 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3366 * A full call to MAP NIL with a single sequence argument no longer conses.
3367 * fixes to problems pointed out by Martin Atzmueller:
3368 * The manual page no longer talks about multiprocessing as though
3369 it were currently supported.
3370 * The ILISP support patches have been removed from the distribution,
3371 because as of version 5.10.1, ILISP now supports SBCL without us
3372 having to maintain patches.
3373 * added a modified version of Raymond Toy's recent CMU CL patch for
3374 EQUALP comparison of HASH-TABLE
3376 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3378 * The system still can't cross-compile itself with
3379 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3380 entails), but at least it can compile more of itself that way
3381 than it used to be able to, and various buglets which were uncovered
3382 by trying to cross-compile itself that way have now been fixed.
3383 * This release breaks binary compatibility again. This time
3384 at least I've incremented the FASL file format version to 2, so that the
3385 problem can be detected reliably instead of just causing weird errors.
3386 * various new style warnings:
3387 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3388 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3389 ** using the lexical binding of a variable named in the *FOO* style
3390 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3391 as specified by ANSI.
3392 * *RANDOM-STATE* is no longer automatically initialized from
3393 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3394 default behavior of the system is to repeat its behavior every time
3395 it's run. If you'd like to change this behavior, you can always
3396 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3397 old convention there was no comparably easy way to get the system to
3398 repeat its behavior every time it was run.
3399 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3400 FTYPE declarations has been removed, in favor of their ANSI
3401 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3402 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3403 the types of their sequence arguments can be determined at compile time.
3404 This is done through a new open code expansion for MAP which eliminates
3405 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3406 argument types can be determined at compile time.
3407 * The optimizer now transforms COERCE into an identity operation when it
3408 can prove that the coerced object is already of the correct type. (This
3409 can be a win for machine generated code, including the output of other
3410 optimization transforms, such as the MAP transform above.)
3411 * Credit information has been moved from source file headers into CREDITS.
3412 * Source file headers have been made more standard.
3413 * The CASE macro now compiles without complaining even when it has
3416 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3418 * (Note that the way that the PCL macroexpansions were rewritten
3419 to accommodate the change in DEFGENERIC below breaks binary
3420 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3421 not run under sbcl-0.6.2. Once we get out of alpha releases,
3422 i.e. hit release 1.0.0, we'll probably try to maintain binary
3423 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3424 and sbcl-1.4.4. Until then, however, it might be fairly common
3425 for maintenance releases to break binary compatibility.)
3426 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3428 * The reporting of unhandled signals has been changed to print some
3429 explanatory text as well as the report form. (Previously only
3430 the report form was printed.)
3431 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3432 it defines, so that the compiler no longer issues undefined function
3433 warnings for compiled-but-not-yet-loaded generic functions.
3434 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3435 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3436 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3437 section of the ANSI spec are supported.
3438 * The "" nickname for the "KEYWORD" package has been removed.
3439 The reader still handles symbol tokens which begin with a package marker
3440 as keywords, but it doesn't expose its mechanism for doing so in the
3441 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3442 * The system now issues STYLE-WARNINGs for contradictory TYPE
3443 proclamations. (Warnings for contradictory FTYPE proclamations would
3444 be nice too, but those can't be done usefully unless the type system
3445 is made smarter about FUNCTION types.)
3446 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3447 systematized, so that "*target-*.lisp is supposed to exist only on the
3448 target and imply that there's a related file which exists on the
3449 host, and *host-*.lisp is supposed to exist only on the host and imply
3450 that there's a related file which exists on the target. This involves a
3451 lot of renaming. Hopefully the acute confusion caused by the renaming
3452 will be justified by the reduction in chronic confusion..
3453 ** runtime-type.lisp -> early-target-type.lisp
3454 ** target-type.lisp -> late-target-type.lisp
3455 ** early-host-format.lisp -> early-format.lisp
3456 ** late-host-format.lisp -> late-format.lisp
3457 ** host-error.lisp -> misc-error.lisp
3458 ** early-error.lisp -> early-target-error.lisp
3459 ** late-error.lisp -> late-target-error.lisp
3460 ** host-defboot.lisp -> early-defboot.lisp
3461 ** code/misc.lisp -> code/target-misc.lisp
3462 ** code/host-misc.lisp -> code/misc.lisp
3463 ** code/numbers.lisp -> code/target-numbers.lisp
3464 ** code/early-numbers.lisp -> numbers.lisp
3465 ** early-host-type.lisp -> early-type.lisp
3466 ** late-host-type.lisp -> late-type.lisp
3467 ** host-typep.lisp -> typep.lisp
3468 ** load.lisp -> target-load.lisp
3469 ** host-load.lisp -> load.lisp
3470 ** host-disassem.lisp -> disassem.lisp
3471 ** host-insts.lisp -> insts.lisp
3472 ** byte-comp.lisp -> target-byte-comp.lisp
3473 ** host-byte-comp.lisp -> byte-comp.lisp
3474 ** host-signal.lisp -> signal.lisp
3475 ** host-defstruct.lisp -> defstruct.lisp
3476 ** late-target-type.lisp -> deftypes-for-target.lisp
3477 Furthermore, several other previously target-only files foo.lisp (e.g.
3478 hash-table.lisp and random.lisp) have been split into a target-and-host
3479 foo.lisp file and a target-only target-foo.lisp file, with their key type
3480 definitions in the target-and-host part, so that the cross-compiler will
3481 know more about target types.
3482 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3483 gone away. In their place are various *BACKEND-FOO* variables
3484 corresponding to the slots of the old structure.
3485 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3486 into the target SBCL has been fixed.
3487 * The chill.lisp system for loading cold code into a running SBCL
3489 * Support for the CMU CL "scavenger hook" extension has been removed.
3490 (It was undocumented and unused in the CMU CL sources that SBCL was
3491 derived from, and stale in sbcl-0.6.1.)
3492 * Various errors in the cross-compiler type system were detected
3493 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3494 (enabling various consistency checks). Many of them were fixed,
3495 but some hard problems remain, so the compiler is back to
3496 running without *TYPE-SYSTEM-INITIALIZED* for now.
3497 * As part of the cross-compiler type system cleanup, I implemented
3498 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3499 * I have started adding UNCROSS calls throughout the type system
3500 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3501 on cross-compiler input files will be able to go away, and various
3503 * CONSTANTP now returns true for quoted forms (as explicitly required
3506 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3508 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3509 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3510 (MAKE-LIST -1). In the longer term, it ought to become true
3511 that declarations are assertions even at SAFETY 1. For now, it's not
3512 quite true even at SAFETY 3, but it's at least more nearly true..
3513 (Note that this change seems to increases the size of the system by
3514 O(5%) and to decrease the speed of the compiler by 20% or more.)
3515 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3516 for the problem of printing dozens of lines of distracting information
3517 about low-level system machinery as part of the top stack frame
3518 on entry to the debugger when an undefined function was called.
3519 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3520 is not reset to COMMON-LISP-USER.
3521 * Compilation of stuff related to dyncount.lisp has been made conditional
3522 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3523 smaller. The various dyncount-related symbols have been moved into
3524 a new "SB-DYNCOUNT" package.
3525 * tty-inspect.lisp has been renamed to inspect.lisp.
3526 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3527 feature has gone away. (When we eventually port to other flavors of
3528 libc and/or Unix, we'll try to make the differences between flavors
3529 invisible at the user level.)
3530 * Various other *FEATURES* tags, and/or their associated conditionals,
3531 have been removed if obsolescent, or given better documentation, or
3532 sometimes given more-mnemonic names.
3534 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3536 * tidied up "make.sh" script
3537 * tidied up system directory structure
3538 * better "clean.sh" behavior
3539 * added doc/FOR-CMUCL-DEVELOPERS
3540 * many many small tweaks to output format, e.g. removing possibly-confusing
3541 trailing #\. character in DESCRIBE-INSTANCE
3542 * (EQUALP #\A 'A) no longer signals an error.
3543 * new hashing code, including EQUALP hashing
3544 * tidied up Lisp initialization and toplevel
3545 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3546 * command line argument processing
3547 * added POSIX-GETENV function to deal with Unix-ish environment variables
3548 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3549 terminating SBCL on EOF
3550 * non-verbose GC by default
3551 * There is no more "sbcl" shell script; the sbcl file is now the C
3552 runtime executable (just like CMU CL).
3553 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3555 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3556 symbol and package information as Lisp native symbol and package objects
3557 instead of strings naming symbols and strings naming packages. This way,
3558 whenever packages are renamed (as in warm init), debug information is
3559 transformed along with everything else.
3560 * tweaked the optimization policy declarations which control the building
3561 of SBCL itself. Now, among other things, the system no longer saves
3562 source location debugging information. (This helps two problems at once
3563 by reducing SBCL size and by keeping SBCL from trying to look for its
3564 sources -- which may not exist -- when reporting errors.)
3565 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3566 debugging and testing purposes
3567 * cleaned up printing, making the printer call PRINT-OBJECT for
3568 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3569 methods, giving nearly-ANSI behavior
3570 * converted almost all special variables to use *FOO* naming convention
3571 * deleted PARSE-TIME functionality, since it can be done portably
3572 * moved some files out of cold init into warm init
3573 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3575 * regularized formatting of source files
3576 * added an install.sh script
3577 * fixed ridiculous memory usage of cross-compiler by making
3578 compiler/alloc.lisp not try to do pooling unless it can hook
3579 itself into the GC of the cross-compilation host. Now the system
3580 builds nicely on my old laptop.
3581 * added :SB-ALLOC in target-features.lisp-expr
3582 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3583 was not implemented)
3584 * re-did condition handling and note reporting in the compiler. Notes
3585 are no longer handled by signalling conditions. Style warnings
3586 and warnings are handled more correctly and reported in such a way
3587 that it's easy to find one or the other in your output (so that you
3588 can e.g. figure out which of many problems caused COMPILE-FILE to
3590 * changed the severity of several compiler warnings from full WARNING
3591 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3592 compiler note reporting so that it doesn't use the condition system
3593 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3594 COMPILE-FILE command)
3595 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3596 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3597 no longer supported, since their implementation was deeply intertwingled
3598 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3599 * removed bogus "support" for compiler macros named (SETF FOO), and
3600 removed the compiler macro for SETF INFO (but only after making a fool
3601 of myself on the cmucl-imp mailing list by posting a bogus patch for
3602 DEFINE-COMPILER-MACRO..)
3603 * Compiled files containing forms which have side effects on the Lisp
3604 reader (such as DEFPACKAGE forms) are now handled more correctly.
3605 (Compiler queuing of top level lambdas has been suppressed by setting
3606 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3607 * deleted various currently-unused source files, e.g. gengc.lisp. They
3608 may be added back at some point e.g. when porting to other architectures,
3609 but until they are it's distracting to distribute them and to try to
3611 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3612 there were so many of them they're just distractions, and UNCROSS is
3613 known to be able to handle the current sources
3614 * moved PROFILE functionality into TRACE, so that it will be clear
3615 how the wrapping and unwrapping of functions when you profile them
3616 interacts with the wrapping and unwrapping of functions when you
3617 trace them. (Actually, the functionality isn't there yet, but at least
3618 the interface specification is there. Hopefully, the functionality will
3619 arrive with some future maintenance release.)
3620 * removed host-oops.lisp
3621 * changed signature of QUIT function to allow UNIX-CODE argument
3622 * fixed READ-SEQUENCE bug
3623 * tweaked verbose GC output so that it looks more like the progress
3624 output that ANSI specifies for functions like LOAD
3625 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3626 * added <http://sbcl.sourceforge.com> to the banner information printed by