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)
24 * fixed bug: metacircle resolution in cases where methods have slots
25 added before the slots from STANDARD-METHOD. (reported by Jean
27 * fixed bug: the Power PC assembly code for calling into lisp
28 sometimes computed its return address wrongly (depending on the
29 15th bit of the address, fixed at link time). This used to
30 manifest itself in a segmentation violation while building PCL.
31 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
32 many others over the years)
33 * fixed bug: ROOM no longer reports silly numbers for stack usage if
34 the stack pointer (treated as a signed integer) is negative.
35 (thanks to Peter van Eynde)
37 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
38 * new platform: experimental support for the Windows operating
39 system has been added. (thanks to Alastair Bridgewater)
40 * fixed several bugs in and robustified the PPC FFI (including
41 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
42 * bug fix: allow non-simple string symbol names (reported by Paul
44 * bug fix: interrupt handling on NetBSD (thanks to Richard M
46 * bug fix: saving a core corrupted callbacks on x86/x86-64
47 * bug fix: closed a loophole in metacircularity detection and
48 grounding in the PCL implementation of CLOS.
49 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
50 (reported by tomppa on #lisp)
51 * bug fix: clear the direction flag on Lisp -> C transitions, as
52 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
53 (reported by Andrej Grozin and Hendrik Maryns)
54 * optimization: major improvements to GC efficiency on GENCGC platforms
55 * optimization: faster implementation of EQUAL
56 * optimization: emit more efficient opcodes for some common
57 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
58 * optimization: several other minor code-generation improvements on x86-64
59 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
61 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
62 * minor incompatible change: (SETF CLASS-NAME) and (SETF
63 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
64 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
65 by ANSI as a generic function, and for consistency (SETF
66 GENERIC-FUNCTION-NAME) is treated likewise.
67 * fixed bug #233.b: make constraint propagation notice when a variable
68 value is changed after it is referenced but before it is used
69 * fixed bug #296: no more arbitrary behaviour from filenames with
70 odd characters as --load arguments. (reported by Adam Warner)
71 * fixed bug #390: :CHARACTER-SET pathname components now work as
72 expected. (reported by Tim Daly Jr)
73 * fixed bug #391: complicated :TYPE intersections in slot
74 definitions no longer cause an error in PCL internals.
75 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
76 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
77 returning the number of octets which would be written to the
78 file-stream. (thanks to Robert J. Macomber)
79 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
80 arguments to the internal stream functions. (thanks to David
82 * bug fix: stack exhaustion is now handled on the x86/SunOS
83 platform. (thanks to Daisuke Homma)
84 * bug fix: more accurate ROOM results on GENCGC platforms
85 * optimization: improved type inference for arithmetic-for
86 index variables in LOOP
87 * optimization: faster floating-point SQRT on x86-64
88 * fixed some bugs revealed by Paul Dietz' test suite:
89 ** DOCUMENTATION returns NIL instead of "" for method combinations
90 that don't have a docstring
92 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
93 * minor incompatible change: (SETF CLASS-NAME) and (SETF
94 GENERIC-FUNCTION-NAME) are no longer generic functions, and
95 therefore are not customizeable by user code (as seems to be at
96 least permitted and maybe required by AMOP). As a consolation,
97 however, the SBCL implementation of these functions now calls
98 REINITIALIZE-INSTANCE as specified by AMOP.
99 * bug fix: slot-definition documentation is propagated as per ANSI
100 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
101 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
102 argument, as do eslotd initialization methods. (from Pascal
103 Costanza's "Closer" project)
104 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
105 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
107 * bug fix: it is now possible to have more than one subclass of
108 STANDARD-GENERIC-FUNCTION without causing stack overflow.
109 (reported by Bruno Haible, Pascal Costanza and others)
110 * bug fix: the dependent update protocol now works for generic
111 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
113 * bug fix: condition-class instances corresponding to
114 DEFINE-CONDITION forms are now created eagerly. (reported by
115 Kalle Olavi Niemitalo on comp.lang.lisp)
116 * bug fix: floating point printing is more accurate in some
117 circumstances. (thanks to Simon Alexander)
118 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
119 merged with *DEFAULT-PATHNAME-DEFAULTS*.
120 * bug fix: callbacks on OS X now preserve stack-alignment.
121 (thanks to Cyrus Harmon)
122 * enhancement: the x86-64 disassembler is much better at
123 disassembling SSE instructions. (thanks to Lutz Euler)
124 * enhancement: CHECK-TYPE on a variable now allows type-inference
125 (thanks to Vincent Arkesteijn)
126 * optimization: improved performance of EUC-JP external format.
127 (thanks to NIIMI Satoshi)
128 * optimization: performance improvements to IO on file streams of
129 :ELEMENT-TYPE CHARACTER
130 * optimization: much faster memory allocation on x86-64
131 * optimization: faster garbage collections (latency and throughput)
133 * optimization: faster bignum subtraction and division on x86 and x86-64
134 (thanks to Lutz Euler)
135 * optimization: more accurate type-inference for loops that use a
136 floating point index variable or a negative step.
138 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
139 * bug fix: add a workaround to SBCL looping infinitely at startup on
140 Linux kernels with apparently buggy implementations of personality().
141 (thanks to Svein Ove Aas)
142 * bug fix: Unicode symbols are correctly printed in LDB backtraces
143 (thanks to David Lichteblau)
144 * bug fix: local bindings shadow global symbol macros for
145 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
146 * optimization: non-open coded uses of numeric comparison operators
147 (e.g. >) no longer cons when called with more than one parameter
148 on platforms supporting dynamic-extent allocation.
149 * enhancement: saving cores with foreign code loaded is now supported
150 on MIPS/Linux in addition to the previously supported platforms.
151 * bug fix: FIND-RESTART now tests for activity, not applicability when given
152 a restart object as identifier. (reported by Helmut Eller for CMUCL)
153 * bug fix: division by zero in sb-sprof when no samples were collected
154 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
155 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
156 (thanks to Alexander Kjeldaas)
157 * new feature: ignoring the return values of destructive functions
158 like NREVERSE signals a compile-time style-warning.
159 (thanks to Kevin Reid)
160 * new feature: the alignment of alien structure fields can be
161 explicitly specified. (thanks to Cyrus Harmon)
162 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
163 and dump core on SIGQUIT
165 ** incompatible change: threads do not inherit values of specials
166 from their parents (see manual)
167 ** bug fix: threads stacks belonging to dead threads are freed by the
168 next exiting thread, no need to gc to collect thread stacks anymore
169 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
170 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
171 ** bug fix: dynamic variable and thread start related gc lossage
172 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
174 * fixed some bugs revealed by Paul Dietz' test suite:
175 ** SUBTYPEP is slightly more accurate on heinously complicated
176 CONS types where some of the members have uncertain (in the
177 NIL, NIL sense) type relationships to each other.
179 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
180 no longer purified unless :PURIFY T is explicitly specified.
181 ** Non-purified cores are significantly smaller than before
183 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
184 * new feature: timers based on Zach Beane's excellent timer package
185 * added support for the following external formats: koi8-u,
186 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
187 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
188 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
189 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
190 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
191 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
192 * incompatible change: a threaded SBCL will no longer revert to
193 non-threaded mode on non-NPTL systems, but refuse to start entirely.
194 * bug fix: interrupts are disabled until startup is complete; no
195 more sigsegvs when receiving a signal to soon
196 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
198 * bug fix: add a workaround for the memory randomization features in
199 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
200 workaround will only be in effect on systems with the proc filesystem
202 * bug fix: printing objects of type HASH-TABLE signals a
203 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
205 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
206 have been running for over 50 days. (reported by Gilbert Baumann)
207 * bug fix: the logic for getting names of functions gets less
208 confused when confronded with alternate-metaclass
209 funcallable-instances. (reported by Cyrus Harmon)
210 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
211 consistent, even on internal alternate-metaclass objects.
212 * bug fix: flush closure information collected by physical
213 environment analysis prepass before the main pass. (bug reported
215 * bug fix: compiler pack phase does not modify a hashtable, which is
216 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
217 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
218 may be expanded. (reported by Alexander <malishev> on c.l.l)
219 * bug fix: no more style-warnings when compiling code using LAST
220 under high-SPEED low-SPACE settings. (reported by David Wallin
222 * fixed some incompatibilities between SBCL's MOP and the MOP
224 ** the METAOBJECT class is now implemented;
225 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
226 STANDARD-OBJECT, as required;
227 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
228 now compatible (as required by VALIDATE-SUPERCLASS); there
229 remains a consistency requirement over the presence or absence
230 of the FUNCTION class in the superclass list of finalized
231 classes; see the manual for more details;
232 ** the :AROUND method for COMPUTE-SLOTS on
233 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
234 requested slot ordering.
236 ** bug fix: parent thread now can be gc'ed even with a live
238 ** bug fix: binding a special with PROGV to no value is not
239 the same as that symbol not having been bound (thanks to
241 ** bug fix: binding specials is thread safe (thanks to
243 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
244 and SB-SPROF do not die with 'no handler for signal XX in
245 interrupt_handle_now(..)' anymore
246 ** bug fix: WITH-TIMEOUT works with multiple running threads
247 * fixed some bugs revealed by Paul Dietz' test suite:
248 ** ENSURE-GENERIC-FUNCTION should take a method class object for
249 the :method-class keyword argument.
251 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
252 * new port: the Solaris operating system on x86 processors is now
253 mostly supported, though some rough edges in the environment
254 remain. (thanks to Daisuke Homma)
255 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
256 space for its heap. (thanks to Thiemo Seufer)
257 * optimization: SBCL on MIPS platforms now supports dynamic-extent
258 closures. (thanks to Thiemo Seufer)
259 * minor incompatible change: eof selects abort in the debugger.
260 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
261 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
263 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
264 deprecated, and will go away in a future revision of SBCL.
265 * minor incompatible change: GC-ON and GC-OFF are no longer
266 implemented with a counter, it does not matter how many times gc
267 is switched on or off
268 * bug fix: discriminating functions for generic function classes
269 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
270 make invalid assumptions about method precedence order. (reported
272 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
273 platforms. (thanks to Thiemo Seufer)
274 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
275 * bug fix: correct pathname printing with printer escaping is on.
276 (thanks to Kevin Reid)
277 * bug fix: complex VOP definitions in "user-space" no longer trigger
278 package locks. (reported by Zach Beane)
279 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
280 for funcallable-instance functions, and is no different from
282 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
283 characters in the current readtable. (reported by Nicholas Neuss)
284 * bug fix: SUBTYPEP on various CONS types returns more a more
285 accurate acknowledgment of its certainty.
286 * optimizations: REMOVE-DUPLICATES now runs in linear time on
287 lists in some cases. This partially fixes bug 384.
288 * flush all standard streams before prompting in the REPL and the
290 * bug fix: signal handling and triggering gc do not conflict
291 directly anymore, in particular a high frequency sb-sprof does
292 not prevent gc from running
293 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
294 approximation for timezone and DST information between the
295 universal time epoch and the smallest negative 32-bit time_t.
296 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
297 year 1899 so long as the encoded time is non-negative.
298 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
299 the correct answer for non-constant rotate amounts when performing
300 an inline 32-bit rotation.
302 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
303 there is only one thread in the session
304 ** bug fix: memory leak for streams created in one thread and
305 written to in another
306 ** bug fix: lockup when compiled with gcc4
307 ** bug fix: race that allows the gc to be triggered when gc is
309 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
310 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
312 ** bug fix: finished threads are gc'ed properly
313 * fixed some bugs revealed by Paul Dietz' test suite:
314 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
315 has arguments with a different set of applicable methods from
316 the orignal arguments.
317 ** The type error thrown by MAP now has a correct expected-type
319 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
320 name a compiled function.
321 ** Three MISC tests where a large bignum was improperly coerced to
322 a float that couldn't represent that bignum during type
323 derivation were fixed.
324 ** SUBTYPEP can now handle types involving the negation of a
325 list-form FUNCTION type.
326 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
327 as (COMPLEX (AND RATIO (NOT FIXNUM))).
328 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
330 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
331 * New feature: Experimental support for bivalent streams: streams
332 opened with :element-type :default now allow character and binary
333 (unsigned-byte 8) I/O
334 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
335 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
337 * bug fix: no more highly sporadic "couldn't check whether ~S is
338 readable" when reading a stream and an interrupt hits in the middle
339 of a select system call
340 * compiler better recognizes complex arrays (reported by Eduardo
342 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
343 for variables on SBCL built with linkage-tables. (reported by Luis
345 * various error reporting improvements.
346 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
347 (Thanks to Thiemo Seufer)
348 * bug fix: nested reader invokations work correctly
349 * bug fix: it is possible to have simultaneous references to foreign
350 code and foreign data with the same name.
352 ** added x86-64 support
353 ** incompatible change: the threading api now works with thread
354 objects instead of thread ids
355 ** bug fix: threads are protected from signals and interruption when
356 starting up or going down
357 ** bug fix: a race where an exiting thread could lose its stack to gc
358 ** bug fix: don't halt on infinite error in threads if possible
359 ** fixed numerous gc deadlocks introduced in the pthread merge
360 ** bug fix: fixed thread safety issues in read and print
361 ** bug fix: debugger doesn't hang on session lock if interrupted at
362 an inappropriate moment
363 ** bug fix: run-program is now thread safe(r)
364 ** bug fix: inner with-recursive-lock no longer releases the mutex
365 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
366 * fixed some bugs revealed by Paul Dietz' test suite:
367 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
368 EXPECTED-TYPE slots filled.
369 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
371 ** MAKE-CONDITION accepts classes as type-designators.
372 ** COMPILE may never return NIL.
373 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
374 range before calling Unix time functions
376 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
377 * numerous signal handling fixes to increase stability
378 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
379 * minor incompatible change: we now correctly canonize default
380 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
381 as per the MOP, rather than the historical (INITARG INITFUNCTION
382 INITFORM). (reported by Bruno Haible)
383 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
384 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
385 for more information.
386 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
387 pathname is a directory pathname.
388 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
389 * dynamic space size on PPC has been increased to 768Mb. (thanks to
391 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
392 :METACLASS argument in addition to a class name. (reported by
393 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
394 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
395 (reported by Bruno Haible; patch from Gerd Moellmann)
396 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
398 * bug fix: more cleanups to the floating point exception handling on
399 x86-64 (thanks to James Knight)
400 * bug fix: the compiler does not try to constant fold calls of
401 COERCE and BIT-* functions when they return freshly constructed
402 objects. (reported by Nikodemus Siivola and Paul Dietz)
403 * optimization: improved the MIPS versions of generic (in the
404 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
405 * optimization: direct conversion of (unsigned-byte 32) to floats on
406 the PowerPC platform.
407 * optimization: structure instances with raw slots now use less
408 memory, and probably show better memory locality. (thanks to
410 * optimization: DYNAMIC-EXTENT declarations for lists and closures
411 are treated as requests for stack allocation on the x86-64,
412 Alpha-32, PPC and SPARC platforms.
413 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
414 the wrong file descriptor; implementation of SOCKET-OPEN-P.
415 (thanks to Tony Martinez)
417 ** gcing a dead thread can no longer lead to lockups
418 ** threads block signals until they are set up properly
419 ** errno is no longer shared by threads
420 ** interrupt-thread restores the eflags register on x86
421 ** fixed some lockups due to gc/thread interaction
422 * fixed some bugs revealed by Paul Dietz' test suite:
423 ** invalid dotted lists no longer raise a read error when
425 ** signal an error if a symbol that names a declaration is used
426 as the name of a type, or vice versa
427 ** allow using the (declare (typespec var*)) abbreviation for
428 (declare (type typespec var*)) with all type specifiers
429 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
430 ** always return NIL from PPRINT-POP when OBJECT is NIL
431 ** don't signal errors when pretty-printing malformed LABELS,
432 FLET or MACROLET forms
433 ** declarations in a DOLIST body are also in scope for the
435 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
437 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
440 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
441 * fixed cross-compiler leakages that prevented building a 32-bit
442 target with a 64-bit host compiler.
443 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
444 opened with :IF-EXISTS :OVERWRITE.
445 * fixed bug 281: error for an invalid qualifier in a short-form method
446 combination method is not signalled until the faulty method is called.
447 * bug fix: iteration variable type inferrer failed to deal with open
448 intervals. (reported by Alan Shields)
449 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
450 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
451 or more runtime options were provided to the sbcl binary.
452 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
453 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
454 * optimization: there's now a fast-path for fixnum arguments in the
455 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
456 * optimization: the code generated on x86-64 is more compact thanks
457 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
458 * bug fix: floating-point exception handling now partly works on
459 x86-64. (Thanks to James Knight)
460 * improvement to the MIPS backend from Thiemo Seufer: C-style
461 64-bit long long arguments and return values to alien functions
463 * fixed some bugs revealed by Paul Dietz' test suite:
464 ** the type-error signalled from WARN has a filled-in DATUM slot.
465 ** the type-error required when a stream is not associated with
466 a file has the stream as its datum.
467 ** type-errors on single-floats on x86-64 no longer have
468 :INVALID-OBJECT as the datum
469 ** the type-errors signalled for invalid function names now have
470 a correct expected type
471 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
472 for structures defined with a :TYPE. Documentation strings for
473 typed structures are no longer immediately discarded
474 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
475 broadcast streams and synonym streams. FILE-LENGTH now also works
476 on broadcast streams.
478 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
479 * incompatible change: the --noprogrammer option, deprecated since
480 version 0.7.5, has been removed. Please use the equivalent
481 --disable-debugger option instead.
482 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
484 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
485 inoperational for a while now) has been completely removed.
486 * null lexical environments are now printed as #<NULL-LEXENV>,
487 significantly reducing the amount of clutter in typical backtraces.
488 * documentation on weak pointers, finalization, and after GC hooks
489 has been added to the manual.
490 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
491 as other specialized array types, is much faster. SUBSEQ and
492 COPY-SEQ on such arrays have also been sped up.
493 * optimization: EQL is now more efficient when at least other argument
494 is known to be of type (OR FIXNUM (NOT NUMBER)).
495 * fixed bug: the runtime is now less vulnerable to changes in the
496 size of the SBCL object on OS X, and virtual memory is reserved for
497 all spaces that need to be at a fixed address.
498 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
499 * fixed bug: finalizers and after GC hooks that cause consing are now
501 * fixed bug: compiler error messages and summaries are now printed to
502 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
503 * fixed inference of the upper bound of an iteration variable.
504 (reported by Rajat Datta).
505 * fixed bug 211e: calling local functions with duplicated constant
506 keyword argument no longer causes a bogus style warning about an
508 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
509 ftype declaration to be disregarded. (reported by Dave Roberts)
510 * fixed bug 373: caused by erronous compilation of references to alien
511 variables in the runtime on ppc/darwin.
512 * fixed bug 376: CONJUGATE type deriver.
513 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
514 (reported by Baughn on #lisp)
515 * a cross-compiler bug on non-x86 platforms has been identified and
516 fixed. (thanks to Bruno Haible)
517 * improvements to the MIPS runtime code for increased stability.
518 (thanks to Thiemo Seufer)
519 * increased the maximimum compact environment size to allow
520 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
521 * partial workaround for bug 135: don't name the function we're
522 calling for hairy cases of EVAL, so as not to accumulate one environment
523 entry per call to EVAL. (reported by Kevin Reid)
524 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
525 * optimization: type testing for non-vector arrays should be faster.
526 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
528 * bug fix: setting 31st element of a bit vector to zero did not work
530 * bug fix: redefining a class definition which failed due to a
531 previous accessor / function clash now works (but see BUGS entry
532 #380 for more problems in this area). (thanks to Zach Beane)
533 * the long form of DEFINE-METHOD-COMBINATION disables method group
534 checking when given a single method group with pattern *. (thanks
536 * on x86 compiler supports stack allocation of results of simple
537 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
538 * support for building and running on Mac OS X 10.4 "Tiger" has been added
539 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
540 * fixed some bugs related to Unicode integration:
541 ** the restarts for recovering from input and output encoding
542 errors only appear when there is in fact such an error to
544 * fixed some bugs revealed by Paul Dietz' test suite:
545 ** MISC.549 and similar: late transformation of unsafe type
546 assertions into derived types caused unexpected code
548 ** SCALE-FLOAT type deriver is less wrong.
549 ** type derivers for EXP, LOG and similar functions compute result
550 types for complex arguments better.
551 ** (MISC.563) CONJUGATE type deriver works for very restricted
553 ** out-of-line type testers for character strings are available.
554 ** EQUAL compiler transform understands specialness of objects
556 ** accessing double-floats stored on the stack now works on x86-64.
557 ** debugger internals could sometimes create invalid lispobjs,
558 resulting in GC crashes.
559 ** MISC.548: type check weakening can convert required type into
561 ** initialization forms for bindings are not in scope of free special
564 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
565 * incompatible change: thread support for non-NPTL systems has
566 been removed - locking is buggy and unreliable. A threaded
567 SBCL build will now warn at startup and refuse to create
568 new threads, unless futex support is detected in the kernel
569 * incompatible change: the top level REPL now has only an ABORT
570 restart associated with it, not TOPLEVEL and ABORT as it used to.
571 TOP and TOPLEVEL are now available as debugger commands for
572 returning to the top level.
573 * incompatible change: forms evaluated in the REPL now use the
574 global optimization policy.
575 * incompatible change: user- and system-initialization files are
576 no longer processed with LOAD, but by READ and EVAL; hence the
577 global optimization policy, startup package, readtable, etc,
579 * The .fasl file format number has been incremented because of
580 various incompatible changes.
581 * internal entry point details and argument counts no longer appear
582 in backtraces unless explicitly requested by setting
583 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
584 * built-in and standard functions no longer have names like "top
585 level local call to FOO".
586 * fixed bug 32: functions defined in non-null lexical environments
587 now have more legible printed representation
588 * fixed bug 33: functions defined in non-null lexical environemnts
589 are now more amenable to inspection by INSPECT.
590 * workaround for bug 354: XEPs no longer appear in backtraces unless
591 explicitly requested.
592 * fixed bug: receiving the signal which results from attempting to
593 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
594 SIGBUS on other *BSDs) on architectures where the C stack is also
595 the Lisp stack (x86 and x86-64) from foreign code no longer leads
596 to debugger confusion or wild instability. (reported by Cheuksan
598 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
599 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
600 (reported by Lutz Euler)
601 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
602 compiled code. (reported by Paul Dietz)
603 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
604 specialized parameter to a method to something that is not TYPEP
605 the specializer is now possible.
606 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
607 face of package deletion.
608 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
609 pathname. (reported by David Tolpin on c.l.l)
610 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
611 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
612 STRING-STREAM and FILE-STREAM, respectively.
613 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
614 than previously; in addition, its test suite is now run on build.
615 (thanks to Luke Gorrie)
616 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
617 * optimization: added a immediate representation for single-floats
619 * fixed some bugs revealed by Paul Dietz' test suite:
620 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
621 ** attempting to create a package with a colliding nickname causes
622 correctable errors to be signalled.
623 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
624 ** several x86-64 backend bugs related to sign-extension of immediate
627 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
628 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
629 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
631 * fixed loading of multiply forward-referenced layouts.
632 (thanks to Cheuksan Wang)
633 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
634 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
635 (see COMPILE-FILE documentation for details of :PRINT option)
636 * fixed bugs 19 and 317: fixed-format floating point printing is
637 more accurate. This also fixes a bug reported by Adam Warner
638 related to the ~@F format directive.
639 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
641 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
642 dispatch table if the from-char is a dispatch macro character.
643 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
644 by the wordsize no longer ignore the last word. (reported by Lutz
646 * fixed bug in type checking of dynamic-extent variables. (reported
648 * optimization: sequence traversal functions use their freedom to
649 coerce function designators to functions.
650 * optimization: code with many calls to CLOS methods specialized on
651 CLOS classes has had redundant type checks removed.
652 * optimization: added declarations to speed up operations that access
653 the internal character database (for example STRING-UPCASE)
654 * optimization: comparison operations between floats and sufficiently small
655 fixnums no longer create extra rationals
656 * fixed some bugs related to Unicode integration:
657 ** portions of multibyte characters at the end of buffers for
658 character-based file input are correctly transferred to the
659 start of the buffer at the next read.
660 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
661 passing it through to OPEN.
662 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
663 argument given to internal calls to OPEN.
664 ** the built-in comment readers (introduced by character sequences
665 ";" and "#|") are more forgiving to encoding errors; they will
666 STYLE-WARN and then attempt to resync the stream at a character
667 boundary. (thanks to Teemu Kalvas)
668 * fixed some bugs in the x86-64 port:
669 ** Negative short int return values from c-calls are sign-extended
671 ** The stack is aligned to 16-bytes for c-calls, as required by
672 the ABI. (thanks to Cheuksan Wang)
673 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
674 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
675 for enums. (thanks to Vincent Arkesteijn)
676 ** Multiple small optimizations and bugfixes for floating point
678 * fixed some bugs revealed by Paul Dietz' test suite:
679 ** Space, Tab, Linefeed, Return and Page have the invalid
680 secondary constituent character trait.
681 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
683 ** WITH-INPUT-FROM-STRING should only update the index place on
685 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
687 ** Bit-array operations (BIT-AND and similar) worked incorrectly
688 with one-dimensional arrays with fill pointers.
689 ** TYPE-OF failed on a complex with an integer realpart and a
691 ** compiler failure during type inference for the code of form
692 (IF (EQL X (THE ...)) ...) (MISC.535).
694 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
695 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
696 platform. The port supports 61-bit fixnums, large memory spaces
697 and reloading shared object files.
698 * enhancement: saving cores with foreign code loaded is now
699 supported on ppc/Darwin in addition to the previously supported
701 * enhancement: the statistical profiler now walks deeper into the
702 call stack for more meaningful call-graphs and accrued time
703 reports (x86/x86-64 only). It also now reports time spent in
705 * enhancement: it is now possible to trace most individual methods
706 of a generic function in addition to tracing the generic function
708 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
710 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
711 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
712 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
713 produces an error. (thanks to Vincent Arkesteijn)
714 * bug fix: NAMESTRING on pathnames with :WILD components in their
715 directories works correctly. (thanks to Artem V. Andreev)
716 * fixed bug 125: compiler preserves identity of closures. (reported
718 * bug fixed: functions with &REST arguments sometimes failed with
719 "Undefined function" when compiled with (DEBUG 3). (reported by
721 * bug fix: overflow during compiling of setting element of a bit
722 vector with constant index and value. (reported by Timmy Douglas)
723 * build fix: fixed the dependence on *LOAD-PATHNAME* and
724 *COMPILE-FILE-PATHNAME* being absolute pathnames.
725 * on x86 compiler partially supports stack allocation of dynamic-extent
727 * GO and RETURN-FROM do not check the extent of their exit points
728 when compiled with SAFETY 0.
729 * fixed some bugs related to Unicode integration:
730 ** encoding and decoding errors are now much more robustly
731 handled; it should now be possible to recover even from invalid
732 input or output to the terminal. (thanks to Teemu Kalvas)
733 ** provided a first cut at implementing STRING-TO-OCTETS and
734 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
735 ** altered the SB-MD5 contributed module slightly, changing the
736 interface just enough for it to be supportable for builds where
737 lisp characters are not eight bits.
738 * fixed some bugs revealed by Paul Dietz' test suite:
739 ** the FORMATTER-generated functions for ~V[ conditionals require
740 the correct number of arguments.
741 ** READ-FROM-STRING returns the mandated second value when applied
742 to displaced strings.
743 ** the #\Rubout and #\Backspace characters are treated as invalid
744 constituent characters by the tokenizer.
746 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
747 * new feature: reloading changed shared object files with
748 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
749 * new feature: references to foreign variables and functions
750 can now be compiled and loaded before the corresponding shared
751 object file is loaded, as long as the foreign definitions are
752 available at runtime.
753 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
754 supported. (thanks to Dan Debertin)
755 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
756 just keywords. (thanks to Vincent Arkesteijn)
757 * fixed bug #331: structure-class instances corresponding to
758 DEFSTRUCT forms are now created eagerly.
759 * fixed bug #345: backtraces from calls to undefined functions work
760 on x86 as well. Related bug #61 is now also partially fixed on x86
761 (backtraces from throws to unknown catch tags.)
762 * bug fix: lambda-list parsing is now stricter vrt. order and number
763 of lambda-list keywords.
764 * bug fix: as specified by AMOP, an error is signalled if a
765 class-option appears multiple times in a DEFCLASS form. (reported
767 * bug fix: RANDOM can be compiled when the compiler derives the type
768 of its numeric argument as a disjoint set of small integers.
769 (reported by Paul Dietz)
770 * bug fix: starting a core saved with shared objects loaded when
771 those objects are not available no longer causes threaded SBCL to
772 hang. (reported by Sean Ross)
773 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
775 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
776 argument, which must always be NIL. (reported by Kalle Niemitalo)
777 * bug fix: printing 1.0d+23 no longer results in an error.
778 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
779 * bug fix: structure slot setters preserve evaluation order. (thanks
781 * fixed some bugs related to Unicode integration:
782 ** RUN-PROGRAM can allow its child to take input from a Lisp
783 stream. (reported by Stefan Scholl)
784 ** ASDF-INSTALL successfully downloads .tar.gz files in all
785 locales. (reported by Ken Causey)
786 * fixed some bugs revealed by Paul Dietz' test suite:
787 ** INCF, DECF and REMF evaluate their place form as specified in
789 ** FORMATTER expands ~{ ~} iteration directives with V or #
790 parameters correctly.
791 ** FORMATTER deals with the ~@[ ~] conditional directive where the
792 consequent uses no arguments correctly.
793 ** the system has a partial understanding of the (COMPLEX RATIO)
795 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
798 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
799 * new feature: a build-time option (controlled by the :SB-UNICODE
800 keyword feature, enabled by default) for building the system with
801 support for the entire 21-bit character space defined by the
803 * new feature: the system now has rudimentary external-format
804 support; the primary user-visible change at this time is that
805 characters with the high bit set (such as lower-case-e-acute) will
806 print correctly to a terminal in a UTF-8 environment.
807 * minor incompatible change: BASE-CHAR no longer names a class;
808 however, CHARACTER continues to do so, as required by ANSI.
809 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
810 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
811 has been moved to the SB-EXT package (temporarily re-exported from
813 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
814 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
816 * fixed bug #351: better error-handlind and reporting for malformed
818 * fixed bug #350: bignum-printing is now more memory-efficient,
819 allowing printing of very large bignums, eg. (expt 2 10000000).
820 (reported by Bruno Haible)
821 * fixed bug #302: better primitive-type selection for intersection
823 * fixed bug #308: non-graphic characters now all have names, as
824 required. (reported by Bruno Haible)
825 * bug fix: redefining a class with different superclasses now correctly
826 removes it from the direct-subclasses of its previous superclasses.
827 (reported by David Morse)
828 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
829 new value now works. (reported by Bruno Haible)
830 * bug fix: correct canonicalization of multiple non-standard slot
831 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
832 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
833 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
834 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
835 now exists, an signals an error.
836 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
837 during macroexpansion for non-iteration variables as well. (reported
838 by Bruno Haible for CMUCL)
839 * bug fix: Cyclic structures and unprintable objects in compiler
840 messages no longer cause errors. (reported by Bruno Haible)
841 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
842 and READ-FROM-STRING all now return a primary value of NIL if
843 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
844 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
845 (reported by Bruno Haible for CMUCL)
846 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
847 do not designate a condition. (reported by Bruno Haible for
849 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
850 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
851 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
852 permitting resolution in favour of any of the conflicting symbols.
853 (reported by Bruno Haible for CMUCL)
854 * bug fix: EQUAL compiler optimizations is less aggressive on
855 strings which can potentially compare true despite having distinct
856 specialized array element types.
857 * bug fix: unit enumerations can be defined without dividing by
858 zero. (thanks to Vincent Arkesteijn)
859 * FORMAT compile-time argument count checking has been enhanced.
860 (report from Bruno Haible for CMUCL)
861 * a partial workaround for the bug 262: the compiler does not try to
862 inline-expand a local function doing RETURN-FROM from a deleted
863 BLOCK. (thanks to Peter Denno for the bug report and to David
864 Wragg for the simple test case)
865 * fixed some bugs revealed by Paul Dietz' test suite:
866 ** RENAME-PACKAGE allows all package designators as new package
868 ** constraint propagation and lambda variable substitution are
869 more cautious in dealing with partially deleted code.
870 ** compiler sometimes lost reoptimization passes.
871 ** CERROR, given a condition as condition designator, passes its
872 remaining arguments to the continue format control without
874 ** Case-altering FORMAT directives work correctly on non-ASCII
876 ** The REAL type specifier handles bounds outside the floating
877 point ranges without signalling FLOATING-POINT-OVERFLOW.
878 ** Functions with IR1-transformations can create intercomponent
879 references to global functions.
880 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
882 ** FORMAT ~R treats a nil value for its first parameter correctly.
884 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
885 * enhancement: saving cores with foreign code loaded is now
886 supported on x86/NetBSD and sparc/Linux in addition to the previously
888 * bug fix: on some platforms repeated installations caused multiple
889 copies of HTML documentation to be installed -- should not happen
890 any more. (reported by Stefan Scholl)
891 * bug fix: parsing self-recursive alien record types multiple times
892 no longer causes infinite recursion. (reported by Thomas F. Burdick,
893 original patch by Helmut Eller for CMUCL)
894 * bug fix: stack-exhaustion detection works now on NetBSD as well.
895 (thanks to Richard Kreuter)
896 * bug fix: defining classes whose accessors are methods on existing
897 generic functions in other (locked) packages no longer signals
898 bogus package lock violations. (reported by François-René Rideau)
899 * bug fix: special variables as DEFMETHOD parameters no longer have
900 associated bogus type declarations. (reported by David Wragg and
902 * bug fix: read-write consistency on streams of element-type
903 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
904 * bug fix: redefiniton of the only method of a generic function with
905 no DEFGENERIC no longer emits a full WARNING. In addition,
906 redefinition of generic functions with no DEFGENERIC to an
907 incompatible lambda list now signals an error. (thanks to Zach
909 * bug fix: DEFGENERIC now works even when there's a function of the
910 same name in an enclosing lexical environment. (thanks to Zach
912 * fixed compiler failure, caused by instrumenting code during
913 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
914 * optimization: added loop analysis and improved register allocation
915 * fixed some bugs revealed by Paul Dietz' test suite:
916 ** POSITION on displaced vectors with non-zero displacement
917 returns the right answer.
918 ** (SIMPLE-STRING) is a valid type specifier for sequence
920 ** *PRINT-LEVEL* handling for slotless structures is pedantically
922 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
923 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
925 ** The justification version of the FORMAT ~< directive treats
926 non-zero minpad parameter correctly.
928 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
929 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
930 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
931 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
932 the supported interface.
933 * new feature: Single-stepping of code compiled with DEBUG 2 or
934 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
935 * new feature: saving cores with foreign code loaded is now
936 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
937 Timothy Moore's work for CMUCL)
938 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
939 parameters to * instead of NIL if no initform is supplied.
940 (reported by Johan Bockgaard)
941 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
942 correctly when FUNCALL appears as the car of the form. Note:
943 despite this FUNCALL forms are not currently subject to
944 compiler-macro expansion. (port of Raymond Toy's fix for the
945 same from CMUCL, reported by Johan Bockgaard)
946 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
947 (thanks for Teemu Kalvas)
948 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
949 case instead of terminating the iteration (thanks for Julian
950 Squires, Sean Champ and Raymond Toy)
951 * bug fix: incorrect expansion of defgeneric that caused a style
952 warning. (thanks for Zach Beane)
953 * bug fix: slot accessor effective method computation works properly
954 for classes with multiple non-standard applicable methods on
955 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
956 * on x86 compiler supports stack allocation of results of LIST and
957 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
958 CMUCL implementation by Gerd Moellmann)
959 * fixed some bugs revealed by Paul Dietz' test suite:
960 ** FORMAT strings with both the ~<~:;~> form of the justification
961 directive and pretty-printing directives cause an error.
963 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
964 * incompatible change: the internal functions
965 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
966 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
967 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
968 instead of the old functions.
969 * new feature: on platforms where "dladdr" is available foreign
970 function names now appear in backtraces. (based on Helmut Eller's
972 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
973 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
975 * documentation: ASDF manual is now distributed with SBCL.
976 * bug fix: SBCL can now load its contributed modules with REQUIRE
977 even if the system-provided entries have been removed from
978 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
980 * bug fix: backtraces involving undefined functions or assembly
981 routines are more informative. (thanks to Brian Downing)
982 * bug fix: mutually referent alien structures now work correctly.
983 (reported by Rick Taube)
984 * bug fix: structures defined by WITH-ALIEN can be referred to
985 within other definitions in the same WITH-ALIEN.
986 * bug fix: division operators (MOD, TRUNCATE and the like) with
987 constant zero divisors and integer dividends no longer generate
989 * bug fix: provide default methods for INPUT-STREAM-P and
990 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
991 * bug fix: improve the detection and resolution of MOP metacycles
992 (where effective method computation depends on the generic function
993 having its effective method computed). (reported by Bruno Haible)
994 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
995 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
996 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
997 * bug fix: pathnames with relative directory components can be
998 represented relative to default pathnames.
999 * optimization: in taking the GCD of bignums, reduce the two bignums
1000 to approximately the same size (using Euclid's algorithm) before
1001 applying the more sophisticated binary GCD. (thanks to Juho
1003 * optimization: COUNT on bitvectors now operates word-at-a-time.
1004 * optimization: ASH with a positive, but not necessarily constant,
1005 (leftwards) shift, when in a modular context, is compiled to a
1007 * fixed some bugs revealed by Paul Dietz' test suite:
1008 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1009 if the corresponding argument is NIL.
1010 ** FORMAT directives accepting a minpad parameter treat negative
1012 ** literal commas as character parameters to FORMAT directives are
1014 ** literal spaces directly after ~<Newline> directives within a
1015 format-logical-block (~:< ~@:>) do not induce :FILL-style
1016 conditional newlines.
1017 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1018 as specified: it no longer includes conditional newlines).
1019 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1021 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1022 :PER-LINE-PREFIX argument does not evaluate to a string.
1024 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1025 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1026 the manual for details; for now, package locks can be disabled by
1027 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1028 if no major problems are found then it is likely that they will be
1029 compiled in unconditionally.
1030 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1031 unsupported operators on all platforms. To load a shared library
1032 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1033 object file, link it into a shared library outside of SBCL and
1034 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1036 * bug workaround: ROOM T and the GC were not working together
1037 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1038 reporting facility aren't preserved. That reporting has been
1039 disabled, so now until and unless someone figures out how to make it
1040 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1041 * minor incompatible change: as threatened around sbcl-0.8.0, the
1042 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1043 The symbols are also exported from SB-PCL for backwards
1044 compatibility, but more so than before SB-PCL should be treated as
1045 an implementation-internal package.
1046 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1048 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1049 It is known as of this release not to work on the Alpha, however.
1050 * fixed bug #167: errors signalled due to illegal syntax in method
1051 bodies are now more legible.
1052 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1053 designators and can hence be used with TYPEP.
1054 * fixed bug #333: CHECK-TYPE now ensures that the type error
1055 signalled, if any, has the right object to be accessed by
1056 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1057 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1058 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1060 * fixed bug #334: programmatic addition of slots using specialized
1061 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1062 and :ALLOCATION :CLASS slots.
1063 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1064 float exponents. (rereported by Peter Seibel)
1065 * fixed a bug: #\Space (and other whitespace characters) are no
1066 longer considered to be macro characters in standard syntax by
1067 GET-MACRO-CHARACTER.
1068 * fixed bug: initialization of condition class metaobjects no longer
1069 causes an instance of the condition to be created. (reported by
1071 * fixed bug: it is now possible to have slots such that
1072 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1073 description is neither :INSTANCE nor :CLASS.
1074 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1075 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1076 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1077 non-standard SLOT-DEFINITION-ALLOCATION values.
1078 * fixed bugs: various race conditions handling exiting threads.
1079 CL-PPCRE's thread torture test now passes.
1080 * fixed arguably-a-bug: GC time proportional to number of threads in
1081 system even when most of them are idle
1082 * optimization: improved performance of BIT and SBIT on bit-vectors.
1083 * .fasl file incompatibility: The fasl file version number has
1084 been incremented because of changes associated with package locks.
1086 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1087 * minor incompatible change: the system no longer provides
1088 optimization hints (conditions which are TYPEP
1089 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1090 mode; these hints are still emitted when higher SPEED optimization
1092 * new contrib module: a sampling profiler (profiling by statistical
1093 sampling, rather than by instrumenting functions) is available as
1094 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1095 * the behaviour of the standard function ED is now customizeable by
1096 third parties through a hook variable: see ED's documentation
1097 string for information on the protocol.
1098 * the compiler no longer emits efficiency notes for (FUNCALL X)
1099 when the type of X is uncertain under default optimization
1101 * fixed bug 276: mutating a binding of a specialized parameter to a
1102 method to something that is not TYPEP the specializer is now
1104 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1105 exhibit the required behaviour on the x86 platform. (thanks to
1106 Peter van Eynde, Eric Marsden and Bruno Haible)
1107 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1108 even for difficult arguments. (reported by Peter Graves)
1109 * fixed bug 141a: the backquote printer now descends quoted
1111 * fixed another bug in backquote printing: no more destructive
1112 modification of the form's list structure. (reported by Brian
1114 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1115 move between its address being taken and the call to
1116 interrupt_thread, fixing a crashing race condition.
1117 * the SB-POSIX contrib implementation has been adjusted so that it
1118 no longer exhibits ridiculously poor performance when constructing
1119 instances corresponding to C structs.
1121 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1122 * minor incompatible change: the sb-grovel contrib now treats C
1123 structures as alien (in the sense of SB-ALIEN) objects rather than
1124 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1125 has implications for memory management of client code
1126 (sb-grovel-returned objects must now be manually managed) and for
1127 type safety (alien objects now have full types).
1128 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1129 used to control emission of compiler diagnostics, rather than the
1130 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1131 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1132 quality should be considered deprecated.
1133 * install.sh now installs the user manual as well
1134 * (not quite a new documentable feature, but worth considering in
1135 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1136 beginnings of a semantically meaningful condition hierarchy is
1137 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1139 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1140 designator as the defaults argument.
1141 * fixed bug: Displaced arrays whose displaced-to array has become
1142 too small now cause ARRAY-DIMENSION to signal an error, providing
1143 for safer bounds-checking. (reported by Bruno Haible)
1144 * fixed bug: DEFCLASS slot definitions with identical :READER and
1145 :WRITER names now signal a reasonable error. (reported by Thomas
1147 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1149 * fixed bug: Invalid slot specification errors now print correctly.
1150 (thanks to Zach Beane)
1151 * fixed bug 320: Shared to local slot value transfers in class
1152 redefinitions now happen corrently from superclasses as
1153 well. (reported by Bruno Haible)
1154 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1156 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1157 improper lists correctly. (reported by Bruno Haible)
1158 * fixed bug 313: source-transform for <fun-name> was erroneously
1159 applied to a call of a value of a variable with name <fun-name>.
1160 (reported by Antonio Menezes Leitao)
1161 * fixed bug 307: The obsolete instance protocol ensures that
1162 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1163 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1164 remove dead unknown-values globs from the middle of the stack.
1165 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1166 easier to resume long computations after using *BREAK-ON-SIGNALS*
1167 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1168 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1169 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1170 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1171 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1172 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1174 * fixed a (fairly theoretical) bug in string printing: if
1175 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1176 does not have array-element-type equal to the most general string
1178 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1179 function name. (thanks to Nikodemus Siivola)
1180 * fixed bug: compile-time format string checker failed on
1181 non-closed ~{. (reported by Thomas F Burdick)
1182 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1183 don't warn on legal loop code involving more than one aggregate
1184 boolean. (thanks to Nikodemus Siivola)
1185 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1186 in keyword position to MAKE-PACKAGE was not being reported as
1188 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1189 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1191 * fixed bugs 280 and 312: the checking for multiple definitions in a
1192 file is less likely to become confused by uses of inline
1194 * fixed bug: the #S reader macro performs the keyword coercion
1195 specified for slot names. (reported by Kalle Niemitalo)
1196 * fixed bug: lambda lists may contain symbols whose names start with
1197 & but are not lambda-list-keywords; their occurrence triggers a
1199 * fixed bug 321: define-method-combination argument lambda lists do
1200 not cause invalid code to be generated when &optional or &aux
1201 variables have default values. (reported by Bruno Haible)
1202 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1203 have CLOS classes; this appears also to have fixed reported
1204 instability in stack exhaustion detection.
1205 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1206 MUFFLE-WARNING when no associated restart is present is now
1208 * optimization: rearranged the expansion of various defining macros
1209 so that each expands into only one top-level form in a
1210 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1212 * optimization: used a previously little-used slot in symbols to
1213 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1215 * fixed some bugs revealed by Paul Dietz' test suite:
1216 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1218 ** Fixed an optimization bug related to inheritance of initforms
1219 from local to shared slots.
1220 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1221 ** CAST optimizer forgot to flush argument derived type.
1222 ** print/read consistency on floats is now orders of magnitude
1223 more likely. (thanks also to Bruno Haible for a similar report
1225 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1226 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1227 for objects of type REAL. Make it so.
1228 ** SXHASH is specified (infelicitously) to respect similarity,
1229 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1230 it so. (thanks to Markus Ziegler)
1231 ** on the Alpha, the compiler succeeds in compiling functions
1232 returning a known number of arguments greater than 63.
1233 ** fixed handling of invalid NIL arguments in keyword position
1235 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1237 ** provided a readably-printable representation for RANDOM-STATE
1239 ** ensured that pathnames, where they have a namestring, always
1240 print using #P"..." syntax.
1242 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1243 * Support for the forthcoming 2.0 version of the NetBSD kernel
1244 running on x86 hardware has been added. (thanks to Perry
1245 E. Metzger most immediately, and others for their past work)
1246 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1247 current release version; SBCL's OpenBSD support had been broken
1248 since about the time of OpenBSD's switch to ELF binary format.)
1249 * [placeholder for DX summary]
1250 ** user code with &REST lists declared dynamic-extent, under high
1251 speed or space and low safety and debug optimization policy.
1252 * The manual has been converted to Texinfo format and the debugger
1253 chapter from the cmucl manual has been added.
1254 * A facility has been added to extract documentation strings from
1255 sbcl and store them as Texinfo-formatted snippets for inclusion in
1256 the manual (via Texinfo's @include directive)
1257 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1258 the test case to Dave Roberts)
1259 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1260 pointers. (reported by Sean Ross)
1261 * bug fix: PROFILE output is printed nicely even for large numerical
1262 values. (thanks to Zach Beane)
1263 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1264 greater than 32 handle EOF correctly.
1265 * bug fix: on X86 an immediate argument of the IMUL instruction is
1266 correctly printed in disassembly. (thanks to Lutz Euler)
1267 * bug fix: class slots in redefined classes preserve their old
1268 values. (thanks to Bruno Haible and Nikodemus Siivola)
1269 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1270 succeeds. (reported by Marco Baringer)
1271 * fixed some bugs revealed by Paul Dietz' test suite:
1272 ** READ-SEQUENCE now works on ECHO-STREAMs.
1273 ** RATIONALIZE works more according to its specification. (thanks
1276 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1277 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1278 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1279 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1280 should matter to you only if you rebind the printer control
1281 variables and then find you want different bindings in the
1282 debugger than in the ordinary execution of your program.)
1283 * The runtime build system has been tweaked to support building
1284 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1285 assembler and linker. (thanks to Nikodemus Siivola)
1286 * Unbound, undefined, undeclared variables now trigger full
1287 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1288 more often programmer error than deliberate exploitation of undefined
1290 * optimization: the hash algorithm for strings has changed to one
1291 that is less vulnerable to spurious collisions. (thanks to Juho
1293 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1294 needless bounds checking. (thanks to Juho Snellman)
1295 * optimization: implemented multiplication as a modular
1296 (UNSIGNED-BYTE 32) operation on the PPC backend.
1297 * fixed some bugs revealed by Paul Dietz' test suite:
1298 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1300 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1301 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1302 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1304 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1306 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1308 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1309 their output stream on EOF from read.
1310 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1311 have been read to end-of-file.
1312 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1314 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1315 description of determination of which consecutive characters
1317 ** Printing the "Space" character with escaping on now yields "#\\ ",
1318 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1319 ** Reading floating-point numbers with *READ-BASE* set to a number
1320 less than 10 works correctly.
1321 ** Reading floating-point numbers with *READ-BASE* set to a number
1322 more than 10 works correctly.
1323 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1324 the readtable currently in effect.
1326 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1327 * minor incompatible change: parsing of namestrings on a physical
1328 (Unix) host has changed; numbers after the final #\. in a
1329 namestring are no longer interpreted as a version field. This is
1330 intented to be largely invisible to the user, except that the
1331 meaning of the namestring "*.*.*" has changed: it now refers to a
1332 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1333 should usually be replaced by
1334 (make-pathname :name :wild :type :wild :version :wild)
1335 with the added benefit that this is more likely to be portable.
1336 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1337 option to OPEN now signals an error if the file being opened
1338 exists; this may have an impact on existing code.
1339 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1340 signals received, on the PowerPC platforms (both Linux and
1341 Darwin). (thanks to Pierre Mai for pointing out the location of
1343 * several fixes on OS X: The system now builds and runs cleanly on
1344 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1345 (thanks to Brian Mastenbrook)
1346 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1347 second-resolution: integer multiples of 1/3600 between -24 and 24.
1348 (thanks to Vincent Arkesteijn)
1349 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1350 type when called with 1 argument; PEEK-CHAR checked type of
1351 PEEK-TYPE only after having read first character from a
1352 stream. (reported by Peter Graves)
1353 * bug fix: the garbage collector now has much better locality
1354 behaviour, and in particular no longer treats all memory as being
1355 exhausted when many small objects point to each other in a deeply
1357 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1358 recognized as being TYPEP their class.
1359 * bug fix: the PUSHNEW documentation string has been corrected.
1360 (thanks to Vincent Arkesteijn)
1361 * bug fix: defaulting of the value for the last of an atypically
1362 large number of multiple values being bound was not being
1363 performed correctly on the Alpha or PPC platforms
1364 * optimization: implemented multiplication as a modular
1365 (UNSIGNED-BYTE 32) operation on the x86 backend.
1366 * optimization: SEARCH on simple-base-strings can now be open-coded.
1367 (see also contrib/compiler-extras.lisp for inspiration for
1368 teaching the compiler about the Boyer-Moore algorithm).
1369 * value, returned by MAX (and MIN) called with several EQUALP, but
1370 not EQL, arguments now does not depend on compiler settings.
1371 * fixed some bugs revealed by Paul Dietz' test suite:
1372 ** in stack analysis liveness information is propagated from
1373 non-local entry points.
1374 ** pathwise CAST removing failed when the CAST node did not start
1376 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1377 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1379 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1380 host is already defined.
1381 ** RENAME-FILE works on streams instead of signalling an internal
1383 ** PEEK-CHAR uses the current readtable when determining whether
1384 or not a character is whitespace.
1385 ** MERGE-PATHNAMES handles the case when the pathname does not
1386 specify a name while the default-pathname specifies a version.
1387 ** Pathnames now stand a better chance of respecting print/read
1389 ** Attempting to use standardized file system operators with a
1390 pathname with invalid :DIRECTORY components signals a
1392 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1394 ** DIRECTORY on logical pathnames is more correct.
1395 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1396 signal a TYPE-ERROR if their argument is not a stream.
1397 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1398 designator argument does not designate a stream.
1399 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1400 examining the synonym.
1401 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1403 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1404 element-type larger than ([UN]SIGNED-BYTE 32).
1406 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1407 * When built with the :SB-FUTEX feature, threaded builds now take
1408 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1409 for faster/more reliable mutex and condition variable support.
1410 * Incompatible change (but one you probably shouldn't have been using
1411 anyway): the interface and code for arbitrating between multiple
1412 threads in the same user session has been redesigned.
1413 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1414 error when called without an explicit environment argument.
1415 (thanks to Nikodemus Siivola)
1416 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1417 confused about their position. (thanks to Adam Warner and Gerd
1419 * bug fix: Pretty printing backquoted forms with unquotations in the
1420 argument list position of various code constructs such as LAMBDA
1421 now works correctly. (reported by Paul Dietz)
1422 * bug fix: Pretty printing unquotations no longer loses all
1423 stream position information.
1424 * optimization: performance of string output streams is now less
1425 poor for multiple small sequence writes.
1426 * optimization: performance of CSUBTYPEP in the presence of complex
1427 expressions involving CONS and NOT many times has been improved.
1428 (reported by Paul Dietz)
1429 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1431 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1432 streams for the REPL, for the debugger, and so on. (thanks to
1434 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1435 * fixed some bugs revealed by Paul Dietz' test suite:
1436 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1437 is ignored if it is too small, rather than propagating through
1438 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1439 ** extremely complex negations of CONS types were not being
1440 sufficiently canonicalized, leading to inconsistencies in
1442 ** VALUES tranformer lost derived type.
1444 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1445 * fixed a bootstrapping bug: the build process no longer assumes
1446 that the various BOOLE-related constants have the same value in
1447 host and target lisps. (noted by Paul Dietz' test suite on an
1448 SBCL binary built from CLISP)
1449 * The system can now be dynamically linked on the MIPS platform,
1450 which enables dynamic loading of foreign code from Lisp. (thanks
1451 to Ralf Baechle for discussions on the MIPS ABI)
1452 * The system now records debugging information for its own source
1453 files in a filesystem-position-independent manner, relative to
1454 the "SYS" logical host.
1455 * fixed a compiler bug: MV-LET convertion did not check references
1456 to the "max args" entry point. (reported by Brian Downing)
1457 * tweaked disassembly notes to be less confident about proclaiming
1458 some instruction as an LRA. (thanks to Brian Downing)
1459 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1460 now each have their own history, command character, and other
1461 characteristics. (thanks to David Lichteblau)
1462 * fixed some bugs revealed by Paul Dietz' test suite:
1463 ** compiler failure in compiling LOGAND expressions including a
1465 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1466 shift greater than 32.
1467 ** FLUSH-DEST did not mark blocks for type check regeneration.
1468 ** HANDLER-CASE failed to accept declarations in handler clauses
1469 in some circumstances.
1471 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1472 * New code in contrib/sb-introspect (still probably not entirely
1473 stable yet) provides some support for smart Lisp development
1474 environments like SLIME.
1475 * The conditions signalled for errors occurring when loading .fasl
1476 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1477 in a way which should help ASDF recover gracefully.
1478 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1479 stuff has been cleaned up. If you code contrib/ stuff, this might
1480 affect you, and you can look at contrib/README, contrib/STANDARDS,
1481 and/or the 0.8.4.27 diff to check.
1482 * In full calls the compiler now does not generate checks for declared
1483 argument types for all arguments.
1484 * various threading fixes
1485 ** and some experimental patches which didn't make it into
1486 the main tree for this release, but which are shipped in
1487 contrib/experimental-thread.patch as a possible fix for some
1488 failures (deadlock, spinning...) in GC-intensive multithreaded
1490 * fixed PPC build problem (source code incompatibility of different
1491 library versions): added offsetof() hackery which attempts to divine
1492 where glibc maintainers put uc_mcontext today
1493 * fixed bug 282: compiler does not trust type assertions while passing
1494 arguments to a full call.
1495 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1496 &OPTIONAL VALUES type parameter.
1497 * fix bug 214: algorithm for noting rejected templates is now more
1498 similar to that of template seletion. (also reported by rydis on
1500 * fixed bug 141b: printing backquoted information readably and prettily
1501 inserts a space where necessary.
1502 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1503 (CHARACTER 10) are now reported as errors, rather than propagated
1504 as unknown types. (reported by piso on #lisp)
1505 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1506 with values NIL and :ERROR. (thanks to Milan Zamazal)
1507 * fixed bug 191c: CLOS now does proper keyword argument checking as
1508 described in CLHS 7.6.5 and 7.6.5.1.
1509 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1510 longer raise an error; further, using a list as a for-as-arithmetic
1511 counter now raises a meaningful error.
1512 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1513 now understood by sequence creation functions such as MAKE-SEQUENCE
1515 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1516 a STRING-INPUT-STREAM.
1517 * compiler enhancement: SIGNUM is now better able to derive the type
1519 * type declarations inside WITH-SLOTS are checked. (reported by
1521 * fixed some bugs revealed by Paul Dietz' test suite:
1522 ** incorrect optimization of TRUNCATE for a positive first
1523 argument and negative second.
1524 ** compiler failure in let-convertion during flushing dead code.
1525 ** compiler failure while deriving type of TRUNCATE on an
1526 interval, containing 0.
1527 ** ASH of a negative bignum by a negative bignum count now returns
1529 ** intersection of CONS types now canonicalizes properly, fixing
1530 inconsistencies in SUBTYPEP.
1532 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1533 * incompatible change: The --disable-debugger command line
1534 option now clobbers the debugger at a more fundamental
1535 level, by redefining #'INVOKE-DEBUGGER instead of by
1536 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1537 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1538 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1539 enter the debugger REPL and then suffer recursive errors
1540 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1541 while under the new system, BREAK in --disable-debugger mode
1542 terminates the system just as an unhandled error would.
1543 * fixed compiler performance when processing loops with a step >1;
1544 * bug fix: DOCUMENTATION now retrieves generic function
1545 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1546 support has been systematized, and now supports the methods
1547 specified by ANSI, along with a default method and a method for
1548 slot documentation. (reported by Nathan Froyd)
1549 * bug fix: effective methods associated with a generic function are
1550 no longer cached over a change of that generic function's method
1551 combination. (reported by Andreas Fuchs)
1552 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1553 names a non-existent directory. (thanks to Andreas Fuchs)
1554 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1555 the PPC platform, lead to stack corruption; this has been fixed.
1556 (reported by Rainer Joswig)
1557 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1558 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1559 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1560 arithmetic in general, is now much more reliable on the PPC
1562 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1563 platform now returns the right answer.
1564 * optimization: restored some effective method precomputation in
1565 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1566 precomputation is now tunable.
1567 * optimization: compiler-internal data structure use has been
1568 reviewed, and changes have been made that should improve the
1569 performance of the compiler by about 20%.
1570 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1571 simple-base-strings and simple-bit-vectors is improved.
1572 * optimization: the optimization of 32-bit logical and arithmetic
1573 functions introduced in version 0.8.3 on the x86 has been
1574 implemented on the mips, ppc and sparc platforms; an
1575 implementation of the same facility, but for 64-bit arithmetic,
1576 has been added for the alpha.
1577 * microoptimization: the compiler is better able to make use of the
1578 x86 LEA instruction for multiplication by constants.
1579 * bug fix: in some situations compiler did not report usage of
1580 generic arithmetic in (SPEED 3) policy.
1581 * bug 145b fix: compiler used wrong type specifier while converting
1582 MEMBER-types to numeric.
1583 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1585 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1586 index argument now works properly on the Alpha platform.
1587 * bug fix: floating point exception treatment on the Alpha platform
1589 * bug fix: FILE-POSITION works much better on string input and
1590 output streams. (thanks to Nikodemus Siivola)
1591 * bug fix: many threading/garbage collection symptoms sorted.
1592 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1593 might be pseudo-atomic.
1594 * internal change: Stopping for GC is now done with signals not
1595 ptrace. GC is now done in whichever thread wanted it, instead of
1597 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1599 * bug fix: result form in DO is not contained in the implicit
1601 * incompatible change: ICR structure is changed; the value part of
1602 CONTINUATION is now called LVAR; corresponding functions are
1603 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1604 * added type deriver for ISQRT (thanks to Robert E. Brown).
1605 * bug fix: better support for loading from the command line when an
1606 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1608 * fixed some bugs revealed by Paul Dietz' test suite:
1609 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1610 ** ROUND and FROUND now give the right answer when given very
1611 small float arguments.
1612 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1614 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1615 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1616 ** LCM should return a non-negative integer.
1617 ** PARSE-INTEGER returned the index of a terminator instead of the
1618 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1619 ** PARSE-INTEGER returned an incorrect index being applied to a
1621 ** LCM with two arguments of 0 returns 0 rather than signalling
1623 ** unsigned addition of a 32-bit constant with the high bit set no
1624 longer causes an internal compiler error on the x86.
1625 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1626 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1627 with negative last argument.
1628 ** byte specifiers with zero size and position no longer cause
1629 an error during type derivation.
1630 ** bignum multiplication on the Alpha platform now returns the
1632 * porting: The system now builds on SuSE AMD64, although it still
1633 generates a 32-bit binary.
1634 * .fasl file incompatibility: The fasl file version number has
1635 been incremented (because of the changes to internal compiler
1636 data structures referred to above).
1638 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1639 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1640 more accurately, on the Darwin kernel running on PowerPC hardware.
1641 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1642 * Compiler code deletion notes now signal a condition of type
1643 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1644 an associated MUFFLE-WARNING restart.
1645 * The compiler now performs limited argument count validation of
1646 constant format strings in FORMAT, and where appropriate in ERROR,
1647 CERROR and WARN. (thanks to Gerd Moellmann)
1648 * New ASDF-INSTALL contrib can be used for automatic download and
1649 installation of third-party Lisp code from CCLAN or other sites
1651 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1652 which forces another thread to execute a function supplied by the
1654 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1655 now accept and act upon their :ELEMENT-TYPE keyword argument.
1656 (reported by Martin Atzmueller, Edi Weitz)
1657 * bug fix: FILE-POSITION now accepts position designators up to
1658 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1659 is the greater. (thanks to Patrik Nordebo)
1660 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1661 Kalle Olavi Niemitalo)
1662 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1663 printable. (reported by Eric Marsden)
1664 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1665 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1666 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1667 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1668 (thanks to Stig Sandoe)
1669 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1670 UNREAD-CHAR now correctly includes the unread character in the
1671 target sequence. (thanks to Gerd Moellmann)
1672 * bug fix in threaded builds: the system can now be suspended and
1673 resumed by shell job control with minimal disruption.
1674 * bug fixes in times and timezones >2038AD
1675 * better handling of "where is GNU make?" problem in build scripts
1676 (thanks to Nikodemus Siivola)
1677 * new optimization: inside a named function any reference to a
1678 function with the same name is considered to be a self-reference;
1679 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1680 optimization quality.
1681 * new optimization on x86: logical functions and + now have
1682 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1683 used when the result is truncated to 32 bits.
1684 * VALUES declaration is partially enabled.
1685 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1686 * fixed some bugs revealed by Paul Dietz' test suite:
1687 ** The system now obeys the constraint imposed by
1688 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1689 types form a lattice under type intersection.
1690 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1691 ** ASSOC now ignores NIL elements in an alist.
1692 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1693 and (1+ MOST-POSITIVE-FIXNUM) answers.
1694 ** The addition of a method with invalid qualifiers to a generic
1695 function does not cause an error to be signalled immediately;
1696 a warning is signalled, and the error is generated only on
1697 calling the generic function.
1698 * changed .fasl file version number, in part to add type codes for
1699 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1700 obscure ANSI requirements
1702 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1703 * fixed bug 148: failure to inline-expand a local function left
1704 garbage, confusing the compiler.
1705 * fixed bugs 3cd: structure slot readers perform type check if the
1706 slot can have an invalid value (i.e. it is either not initialized
1707 or can be written with a less specific slot writer).
1708 * bug fix: the compiler now traps array references to elements off
1709 the end of an array; previously, the bounds checking in some
1710 circumstances could go off-by-one.
1711 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1713 * type declarations for array element types now obey the description
1714 on the CLHS page "Declaration TYPE", as per discussions on
1715 sbcl-help around 2003-05-08. This means that a declaration
1716 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1717 declaration, all references to BAR will be asserted or assumed
1718 (with THE, so dependent on compiler policy) to involve objects of
1719 type FOO. Note that no such declaration is implied in
1720 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1721 * declared types of functions from the "Conditions"
1722 chapter. (reported by Paul Dietz)
1723 * bug fix: CERROR accepts a function as its first argument.
1724 * bug fix: NTH an NTHCDR accept a bignum as index
1725 arguments. (reported by Adam Warner)
1726 * optimization: character compare routines now optimize comparing
1727 against a constant character. (reported by Gilbert Baumann)
1728 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1729 argument now works properly on the MIPS platform.
1730 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1731 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1732 declared to have a functional type, does not cause an error
1734 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1735 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1736 * fixed bug: if last continuation of a deleted block has a
1737 destination, this destination should be deleted too. (reported by
1739 * fixed a bug in the bootstrap process: the host compiler's values
1740 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1741 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1742 test case from Patrik Nordebo)
1743 * improved the ability of the disassembler on the PPC platform to
1744 provide helpful disassembly notes.
1745 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1746 the class in more cases than previously.
1747 * bug fix: FILE-POSITION now understands :START and :END for
1748 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1749 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1750 8) data. (thanks to David Lichteblau)
1751 * bug fix: it is possible to add a method to a generic function
1752 without lambda list.
1753 * bug fix: reader failed to signal END-OF-FILE inside an
1754 object representation. (reported by Nikodemus Siivola)
1755 * fixed some bugs revealed by Paul Dietz' test suite:
1756 ** LAST and [N]BUTLAST should accept a bignum.
1757 ** condition slot accessors are methods.
1758 ** (VECTOR NIL) is a subtype of STRING.
1760 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1761 * minor incompatible change: some nonsensical specialized lambda
1762 lists (used in DEFMETHOD) which were previously ignored now signal
1764 * minor incompatible change: the system is now aware of the types of
1765 variables in the COMMON-LISP package, and will signal errors for
1766 most violations of these type constraints (where previously they
1767 were silently accepted).
1768 * minor incompatible change: COMPILE-FILE now uses the freedom
1769 afforded (ANSI 3.2.2.3) to use derived function types for
1770 functions defined in the same file. This also permits the system
1771 to warn on static type mismatches and function
1772 redefinition. (Currently it does not work with high DEBUG level.)
1773 * minor incompatible change: VALUES declaration is disabled.
1774 * When issuing notes, the compiler now signals a condition of type
1775 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1776 restart for use in user handlers. It is expected that the
1777 COMPILER-NOTE condition will eventually become a condition
1778 supertype to a hierarchy of note types, which will then be
1779 handleable in a similar fashion. However, at the moment, no such
1780 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1781 but it's an implementation detail, not a classification for the
1783 * Changes in type checking closed the following bugs:
1784 ** type checking of unused values (192b, 194d, 203);
1785 ** template selection based on unsafe type assertions (192c, 236);
1786 ** type checking in branches (194bc).
1787 * A short form of VALUES type specifier has ANSI meaning (it has
1788 increased the number of situations when SBCL cannot perform type
1790 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1791 keywords or constants is permissible.
1792 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1793 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1794 classes. (thanks to Antonio Martinez)
1795 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1796 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1797 argument to a pretty-print stream. Instead, it leaves any such
1798 operations to DESCRIBE-OBJECT methods.
1799 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1801 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1802 lambda lists. (thanks to David Lichteblau)
1803 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1804 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1805 (thanks to Gerd Moellmann)
1806 * fixed bug 246: increased compilation speed of long
1807 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1809 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1811 * DEFCONSTANT now throws a condition of type
1812 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1813 constant to a non-EQL value; CONTINUE and ABORT restarts
1814 respectively change and preserve the value.
1815 * fixed bug 63: The code walker, part of the implementation of CLOS,
1816 is now better at handling symbol macros.
1817 * bug fix: There is no longer an internal implementation type named
1818 CL:LENGTH. (reported by Raymond Toy)
1819 * bug fix: In macro-like defining macros/special operators the
1820 implicit block does not enclose the lambda list.
1821 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1822 enclosed in parentheses) are not suitable as type specifiers, and
1823 their use properly signals an error now.
1824 * bug fix: An argument count mismatch for a type specifier in code
1825 being compiled no longer causes an unhandled error at compile
1826 time, but signals a compile-time warning.
1827 * fixed simple vector readable printing
1828 * bug fix: DESCRIBE takes more care over whether the class
1829 precedence list slot of a class is bound before accessing it.
1830 (reported by Markus Krummenacker)
1831 * bug fix: FORMATTER can successfully compile pretty-printer format
1832 strings which use variants of the ~* directive inside.
1833 * bug fix: SEARCH now applies its TEST predicate to the elements of
1834 the arguments in the correct order. (thanks to Wolfhard Buss)
1835 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1836 in type inference. (thanks to Robert E. Brown)
1837 * bug fix: Reading in symbols with an explicit package name of ""
1838 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1839 * fixed some bugs revealed by Paul Dietz' test suite:
1840 ** NIL is now allowed as a structure slot name.
1841 ** Arbitrary numbers, not just REALs, are allowed in certain
1842 circumstances in LOOP for-as-arithmetic clauses.
1843 ** Multiple class redefinitions before slot access no longer
1844 causes a type error.
1845 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1846 association between the name and a class.
1847 ** Generic functions with non-standard method-combination and over
1848 five methods all of which return constants no longer return NIL
1849 after the first few invocations. (thanks to Gerd Moellmann)
1850 ** CALL-NEXT-METHOD with no arguments now passes the original
1851 values of the arguments, even in the presence of assignment.
1852 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1854 ** DISASSEMBLE works with closures and funcallable instances.
1855 ** ADD-METHOD now returns the generic function, not the new method.
1856 ** FIND-METHOD signals an error if the lengths of the specializers
1857 is incompatible with the generic function, even if the ERRORP
1859 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1860 which its argument is a member.
1861 ** DEFCLASS only redefines the class named by its class-name
1862 argument if that name is the proper name of the class;
1863 otherwise, it creates a new class.
1864 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1865 of the UNBOUND-SLOT condition to the name of the slot.
1866 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1867 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1868 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1869 SLOT-MAKUNBOUND in the specified fashion.
1871 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1872 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1873 cross-compilation host. As a consequence, we can now bootstrap our
1874 way up to SBCL starting with a bare gcc toolchain and human-readable
1875 source code (first the source to CLISP, then the source to SBCL).
1876 * A contributed module containing a partial implementation of the
1877 simple-streams interface has been included. (thanks to Rudi
1879 * A contributed module implementing the RFC1321 Message Digest
1880 Algorithm, known as MD5, has been included.
1881 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1882 no longer has any effect, as the code controlled by this feature
1883 has been deleted. (As far as we know, no-one has ever built using
1884 this feature, and its semantics were confused in any case).
1885 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1886 work on conditions (as required by the ANSI specification),
1887 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1888 expected behaviour on conditions. Users should note, however,
1889 that such behaviour is not required by the ANSI specification,
1890 and so use of this behaviour may render their code unportable.
1891 * fixed some bugs revealed by Paul Dietz' test suite:
1892 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1894 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1895 ** SXHASH on condition objects no longer returns NIL.
1896 ** :ALLOCATION :CLASS slots are better treated; their values are
1897 updated on class redefinition, and initforms inherited from
1898 superclasses are applied.
1899 ** REMOVE-METHOD returns its generic function argument even when
1900 no method was removed.
1901 ** SHARED-INITIALIZE now initializes the values of the requested
1902 slots, including those with :ALLOCATION :CLASS.
1903 ** ALLOCATE-INSTANCE now works on structure classes defined via
1904 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1906 ** SLOT-EXISTS-P now works on conditions, as well as structures
1908 ** MAKE-LOAD-FORM now has the required methods on
1909 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1910 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1911 keyword, and now implements the SLOT-NAMES argument.
1912 ** methods with &OPTIONAL arguments no longer allow too many
1913 arguments to be passed in the call without error.
1914 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1915 option is consistent with the required arguments of the generic
1916 function lambda list.
1917 * bug fix: REQUIRE accepts a string designator. (Thanks to
1919 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1920 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1921 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1922 * bug fix: The compiler now checks for duplicated variables in macro
1924 * bug fix: SETQ on globals returns the correct value.
1925 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1926 not a COMPILER-ERROR (followed by some other strange error on
1927 choosing the CONTINUE restart).
1928 * bug fix: make.sh and friends are now more consistent in the way that
1929 they look for GNU "make".
1931 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1932 * experimental native threads support (on x86 Linux >=2.4 only).
1933 This is not compiled in by default: you need to add :SB-THREAD to
1934 the target features. See the "Beyond ANSI" chapter of the manual
1936 * fix for longstanding nonANSIism: The old distinction between
1937 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1938 The return value from CL:FIND-CLASS is now a CLOS class, and
1939 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1940 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1941 * An interface to the MetaObject Protocol, as described in Kiczales,
1942 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1943 MIT Press, 1991, is available from the SB-MOP package.
1944 * incompatible change: the SB-PCL package should now be considered
1945 a private implementation detail, and no longer a semi-private MOP
1947 * minor incompatible change: due to rearrangement for threads, the
1948 control stack and binding stack are now allocated at arbitrary
1949 addresses instead of being hardcoded per-port. Users affected by
1950 this probably have to be doing advanced things with shared
1951 libraries, and will know who they are.
1952 * minor incompatible change: Previously, all --eval forms used were
1953 processed with READ before any of them were processed with EVAL.
1954 Now each --eval form is processed with both READ and EVAL before
1955 the next --eval form is processed. (Thus package operations like
1956 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1957 work as the user might reasonably expect.)
1958 * minor incompatible change: *STANDARD-INPUT* is now only an
1959 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1961 * minor incompatible change: Y-OR-N-P is now character-oriented,
1962 not line oriented. Also, YES-OR-NO-P now works without errors.
1963 (thanks to Antonio Martinez)
1964 * sb-aclrepl module improvements: an integrated inspector, added
1965 repl features, and a bug fix to :trace command.
1966 * Known functions, which cannot be open coded by the backend, are
1967 considered to be able to check types of their arguments. (fixing
1968 a bug report by Nathan J. Froyd)
1969 * fixed a bug in computing method discriminating functions: It is
1970 now possible to define methods specialized on classes which have
1971 forward-referenced superclasses. (thanks to Gerd Moellmann)
1972 * fixed evaluation order in optional entries (reported by Gilbert
1974 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1975 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1976 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1977 required-by-AMOP NAME argument, as well as CLASS and
1978 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1979 * fixed bug 20: DEFMETHOD can define methods using names that are
1980 not the proper names of classes to designate class specializers.
1981 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1982 Unix file descriptors, instead of blowing up. (thanks to Antonio
1984 * Garbage collection refactoring: user-visible change is that a
1985 call to the GC function during WITHOUT-GCING will not do garbage
1986 collection until the end of the WITHOUT-GCING. If you were doing
1987 this you were probably losing anyway.
1988 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1989 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1990 TYPEP the latter but not the former.
1991 * The compiler issues a full WARNING for calls to undefined functions
1992 with names from the CL package.
1993 * MAP-INTO for a vector destination is open coded. (reported by
1994 Brian Downing on c.l.l)
1995 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1996 documentation string.
1997 * fixed some bugs revealed by Paul Dietz' test suite:
1998 ** COPY-ALIST now signals an error if its argument is a dotted
2000 ** Condition slots are now accessed more correctly in the presence
2001 of multiple initargs for a given slot.
2002 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2003 exclude restarts of the same name associated with a different
2005 ** DEFCLASS of forward-referenced classes with another
2006 forward-referenced class in the superclasses list no longer
2008 ** Condition slots are now initialized once each, not multiple
2009 times. (thanks to Gerd Moellmann)
2010 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2012 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2013 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2014 arguments contain duplicated elements.
2015 ** RESTART-CASE understands local macros.
2016 ** RESTART-CASE associates exactly its own restarts with a condition.
2017 ** ENDP in safe mode checks its argument to be of type LIST.
2018 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2019 in question is unbound.
2020 ** Optimized MAKE-INSTANCE functions no longer cause internal
2021 assertion failures in the presence of duplicate initargs.
2022 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2023 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2024 propagated even in the ctor optimized implementation of
2026 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2028 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2029 * a better implementation of SXHASH on (simple) bit vectors,
2030 measured both in execution speed and in distribution of results
2031 over the positive fixnums, has been installed. Likewise, a better
2032 implementation of EQUAL for simple bit vectors is now available.
2033 * fixed CEILING optimization for a divisor of form 2^k.
2034 * fixed bug 240 (emitting extra style warnings "using the lexical
2035 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2036 by Antonio Martinez)
2037 * fixed SXHASH, giving different results for NIL depending on type
2038 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2039 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2040 variable. (found by Rolf Wester)
2041 * SBCL does not ignore type declarations for special
2042 variables. (reported by rif on c.l.l 2003-03-05)
2043 * some bug fixes in contrib/sb-aclrepl/
2044 * fixed some bugs revealed by Paul Dietz' test suite:
2045 ** a bug in the CONS type specifier, whereby the CAR and CDR
2046 types got intertwined, has been fixed;
2047 ** the type system is now able to reason about the interaction
2048 between INTEGER and RATIO types more completely;
2049 ** APPEND, [N]REVERSE and NRECONC check that those their
2050 arguments, which must be proper lists, are really so;
2051 ** An array specialized to be unable to hold elements has been
2052 implemented, as required -- yes, really -- by ANSI;
2053 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2054 on malformed property lists;
2056 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2057 * incompatible packaging change: in line with Unix convention,
2058 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2059 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2060 * REQUIRE and PROVIDE are now optionally capable of doing something
2061 useful. See the documentation string for REQUIRE.
2062 * infrastructure for a managed SBCL contrib system: contributed
2063 modules in this release include:
2064 ** the ASDF system definition facility;
2065 ** an interface to the BSD Sockets API;
2066 ** an ACL-like convenience interface to the repl;
2067 (thanks to Kevin Rosenberg)
2068 ** an implementation of ROTATE-BYTE, with efficient implementation
2070 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2071 gives a closer approximation to the right answer than previously.
2072 (thanks to Raymond Toy)
2073 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2074 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2075 optional environment arguments, as required by ANSI.
2076 * fixed bugs in other functions taking environment objects, allowing
2077 calls with an explicit NIL environment argument to be compiled
2079 * fixed bug 228: primary return values from
2080 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2081 COMPILE or FUNCTION.
2082 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2083 :TYPE will no longer signal errors on innocuous objects.
2084 * fixed bug 231b: SETQ is better at respecting type declarations in
2085 the lexical environment.
2086 * fixed a bug in DEFCLASS: classes named by symbols with no or
2087 unprintable packages can now be defined.
2088 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2089 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2090 * fixed compiler failure related to checking types of functions.
2091 (reported by Robert E. Brown)
2092 * the compiler is now much more consistent in its error-checking
2093 treatment of bounding index arguments to sequence functions: in
2094 (SAFETY 3) code, errors will be signalled in almost all cases if
2095 invalid sequence bounding indices are passed to functions defined
2096 by ANSI to operate on sequences.
2097 * fixed a bug in the build procedure: documentation of SBCL-specific
2098 packages is now preserved and available in the final Lisp image.
2099 * lifted FDEFINITION lookup out of loops in the implementation of
2100 many list operations. (thanks to Robert E. Brown)
2101 * fixed a bug in the reader: the #n# reader macro now works for
2102 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2103 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2104 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2105 E. Brown and Tony Martinez respectively)
2106 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2107 SLOT-BOUNDP: in particular, it is now able to optimize them much
2108 better, and is now not vulnerable to having packages renamed.
2109 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2110 always called when a slot is not present in an instance. (thanks
2112 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2113 longer signal an error on unreading a character following EOF on
2114 the previous constituent stream. (thanks to Tony Martinez)
2115 * fixed some bugs revealed by Paul Dietz' test suite:
2116 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2117 not just nonnegative fixnums;
2118 ** the logical bit-array operators such as BIT-AND now accept an
2119 explicit NIL for their "opt-arg" argument (to indicate a
2120 freshly-consed result bit-array);
2121 ** ELT now signals an error on an invalid sequence index in safe
2123 ** the type system is now cleverer about negations of numeric
2124 types, and consequently understands the BIGNUM and RATIO types
2126 ** the type system is now cleverer about the interaction between
2127 INTEGER and RATIO types: while bugs still remain, many more
2128 cases are accurately computed;
2129 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2130 if it is in the last clause;
2131 ** CONSTANTLY now correctly returns a side-effect-free function in
2133 ** DECLARE is no longer treated as a special-operator; in
2134 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2135 * incremented fasl file version number due to the change in the
2136 DEFSTRUCT-SLOT-DESCRIPTION structure.
2138 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2139 * minor incompatible change: code processed by the "interpreter" or
2140 EVAL now has a compilation optimization policy of (DEBUG 2)
2141 (changed from (DEBUG 1)) to improve debuggability of interactive
2142 development, and to allow the use of the debug RETURN command in
2144 * an experimental implementation of the RETURN command for the
2145 debugger has been included. (thanks to Frederik Kuivinen)
2146 * fixed bug 62: constraints were not propagated into a loop.
2147 * fixed bug in embedded calls of SORT (reported and investigated by
2149 * fixed some bugs revealed by Paul Dietz' test suite:
2150 ** printing and reading of arrays with some dimensions having
2151 length 0 (thanks to Gerd Moellmann);
2152 ** BOA constructor with &AUX argument without a default value does
2153 not cause a type error;
2154 ** CONSTANTP now returns true for all self-evaluating objects.
2156 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2157 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2158 accessors that are related by inheritance, as specified in the
2159 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2160 Valtteri Vuorikoski)
2161 * The compiler is now able to inline functions that were defined in
2162 a complex lexical environment (e.g. inside a MACROLET).
2163 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2165 * Support for the upcoming FreeBSD-5.0 release has been included.
2166 (thanks to Dag-Erling Smorgrav)
2167 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2168 effect when it is not in a toplevel context.
2169 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2170 stand a better chance of being correct. (thanks to Gerd
2172 * fixed bug in COERCE, which now signals an error on coercing a
2173 rational to a bounded real type which excludes the expected
2175 * The compiler is now able to derive types more accurately from the
2176 COERCE and COMPILE functions.
2177 * fixed bug 223: functional binding is considered to be constant
2178 only for symbols in the CL package.
2179 * fixed bug 231: SETQ did not check the type of a variable being set
2180 (reported by Robert E. Brown)
2181 * A new optimization for MAKE-INSTANCE has been included, fixing
2182 various bugs (including relating to :ALLOCATION :CLASS slots and
2183 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2185 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2186 ** As required by ANSI, LOOP now disallows anonymous collection
2187 clauses such as COLLECT I in conjunction with aggregate boolean
2188 clauses such as THEREIS (= I 1);
2189 ** LOOP now signals an error when any variable is reused in the
2190 same loop (including the potentially useful construct analogous
2191 to WITH A = 1 WITH A = (1+ A);
2192 ** IT is only a special loop symbol within the first clause of a
2193 conditional loop clause;
2194 ** LOOP with a typed iteration variable over a hashtable now
2195 signals a type error iff it should.
2196 * fixed some other bugs revealed by Paul Dietz' test suite:
2197 ** FILE-STREAM now names the class previously known as FD-STREAM;
2198 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2199 argument) no longer signals an error;
2200 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2201 of the slot symbol, rather than using the current package
2202 ((:CONC-NAME "") continues to intern the slot's name in the
2204 * incremented fasl file version number, because of the incompatible
2205 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2206 of the new implementation of DEFINE-COMPILER-MACRO.
2208 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2209 * Support for building SBCL for MIPS platforms running in
2210 little-endian mode has now been checked in, and basic
2211 functionality on said platforms verified.
2212 * minor incompatible change: PCL now records the pathname of a file
2213 in which methods and the like are defined, rather than its
2215 * minor incompatible change: TRUENAME now considers the truename of
2216 a file naming a directory to be the pathname with :DIRECTORY
2217 component indicating that directory.
2218 * minor incompatible change: a NAMED clause in the extended form of
2219 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2220 reason for the previous behaviour is unclear.
2221 * more systematization and improvement of CLOS and MOP conformance
2222 in PCL (thanks to Gerd Moellman and Pierre Mai):
2223 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2225 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2226 primary methods with no specializers;
2227 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2229 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2230 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2231 FORWARD-REFERENCED-CLASSes; error reporting on
2232 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2234 ** SXHASH on CLOS instances now uses a slot internal to the
2235 instance to return different numbers on distinct instances,
2236 while preserving the same return value through invocations of
2238 ** DEFMETHOD signals errors when methods with longer incongruent
2239 lambda lists are added to generic functions;
2240 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2241 CLASS, as specified in AMOP;
2242 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2243 on the order returned by the primary method for classes of
2244 class STANDARD-CLASS;
2245 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2246 * fixed some bugs shown by Paul Dietz' test suite:
2247 ** DOLIST puts its body in TAGBODY;
2248 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2250 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2251 value producing form;
2252 ** if more variables are given to PROGV than values, extra
2253 variables are bound and made to have no value;
2254 ** NSUBSTITUTE on list arguments gets the right answer with
2256 ** ELT signals an error of type TYPE-ERROR when the index argument
2257 is not a valid sequence index;
2258 ** LOOP signals (at macroexpansion time) an error of type
2259 PROGRAM-ERROR when duplicate variable names are found;
2260 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2261 ** FUNCALL of special-operators now cause an error of type
2263 ** PSETQ now works as required in the presence of side-effecting
2264 symbol-macro places;
2265 ** NCONC accepts any object as its last argument;
2266 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2268 ** loop-for-as-package does not require a package to be explicitely
2270 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2272 * fixed bug 166: compiler preserves "there is a way to go"
2273 invariant when deleting code.
2274 * fixed bug 172: macro lambda lists with required arguments after
2275 &REST arguments now cause an error to be signalled. (thanks to
2277 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2278 bug. (thanks to Matthew Danish)
2279 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2281 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2282 arguments are lexically rebound. (thanks to Gerd Moellmann and
2284 * fixed bug 194: error messages are now more informative when there
2285 is no primary method applicable in a call to a generic
2286 function. (thanks to Gerd Moellmann)
2287 * fixed bug in command line argument checking (thanks to Julian
2289 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2290 (thanks to Matthew Danish)
2291 * incremented fasl file version number, because of the
2292 SXHASH-related changes in the layout of CLOS data structures
2294 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2295 * minor incompatible change: The runtime (the Unix executable named
2296 "sbcl") is now much pickier about the .core files it will load.
2297 Essentially it now requires .core files to descend from the same
2298 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2299 as the runtime does. (The intent is to prevent the crashes which
2300 can occur, and which can even be reported as mysterious failures,
2301 when people patch the sources or change the build parameters
2302 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2303 sbcl and .core files.)
2304 * fixed bug: VALUES-LIST is no longer optimized away.
2305 * fixed bug 142: The FFI conversion of C string values to Lisp
2306 string values no longer conses excessively. (thanks to Nathan
2307 Froyd porting Raymond Toy's fix to CMU CL)
2308 * began to systematize and improve MOP conformance in PCL (thanks to
2309 Nathan Froyd, Gerd Moellman and Pierre Mai):
2310 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2312 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2313 ** FINALIZE-INHERITANCE is now called on class finalization;
2314 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2315 argument precedence order.
2316 * fixed bug 202: The compiler no longer fails on functions whose
2317 derived types contradict their declared type.
2318 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2319 so it can be non-toplevel.
2320 * The fasl file version number has changed (because of the new
2321 implementation of DEFMACRO).
2322 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2323 safe code, that any length requirement by their type-specifier
2324 argument is valid. The exceptions to this are described in bug
2326 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2327 functions now check the types of their inputs as required by ANSI.
2328 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2329 introduced symbol is DECLAREd to be SPECIAL.
2330 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2331 * fixed bug: :COUNT argument to sequence functions may be negative
2332 * fixed bug: body of DO-SYMBOLS may contain declarations
2333 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2334 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2335 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2336 symbol macro only once
2337 * fixed printing of call frame when argument list is unavailable
2338 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2339 * compiler no longer signals WARNING on unknown keyword
2342 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2343 * A beta-quality port to the mips architecture running Linux,
2344 based on the old CMUCL backend, has been made. It has been tested
2345 on a big-endian kernel, and works sufficiently well to be able to
2346 rebuild itself; it has not been tested in little-endian mode.
2347 * fixed an inconsistency between gencgc.c and purify.c which made
2348 dumping/loading .core files unreliable
2349 * fixed bug 120a: The compiler now deals correctly with IFs where
2350 the consequent is the same as the alternative, instead of
2351 misderiving the return type. (thanks to Alexey Dejneka)
2352 * fixed bug 113: Logical pathnames are now dumpable (the logical
2353 host is resolved at load-time, throwing an error if it is not
2355 * fixed bug 174: FORMAT's error message is slightly clearer when a
2356 non-printing character is used in a format directive.
2357 * fixed several bugs in compiler checking of type declarations, i.e.
2358 violations of the Python "declarations are assertions" principle
2359 (thanks to Alexey Dejneka)
2360 * fixed several bugs in PCL's error checking (thanks to Gerd
2362 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2364 * fixed bug in compilation of functions as first class values
2365 (thanks to Antonio Martinez-Shotton)
2366 * The compiler's handling TYPE-ERRORs which it can prove will
2367 inevitably happen at runtime has been cleaned up and corrected
2368 in several ways. (thanks to Alexey Dejneka)
2369 * improved argument type checking for various basic arithmetic
2370 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2371 lavished on them in the past that they can be compiled in many
2372 ways in different special cases
2373 * fixed bug 181: compiler checks validity of user supplied type
2375 * cleaned up code flushing in optimization: Function calls which
2376 should signal errors for safety purposes (e.g. which ANSI says
2377 should signal errors when their arguments are of incorrect type)
2378 are no longer optimized away.
2379 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2380 * incremented fasl file version number, because changes in the
2381 implementation of sequence functions like COERCE caused
2382 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2383 in old inline expansions) to become undefined. (Actually these
2384 changes were later undone, so we might very well be binary
2385 compatible with 0.7.7 after all, but leaving the version number
2386 incremented seemed like the simplest and most conservative
2389 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2390 * An alpha-quality port to the parisc architecture running Linux,
2391 based on the old CMUCL backend, has been made. This, even more so
2392 than the other backends, should be considered still a work in
2393 progress; known problems include that the Linux kernel in 64-bit
2394 mode does not propagate the correct sigcontext structure to
2395 userspace, and consequently SBCL on a parisc64 kernel will not
2397 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2398 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2399 suppressed.) Also now that the compiler is looking at declarations
2400 in the environment, it checks optimization declarations as well,
2401 and suppresses inlining when (> DEBUG SPEED).
2402 * More fixes have been made to treatment of floating point exception
2403 treatment and other Unix signals. In particular, floating point
2404 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2405 * The detection and handling of control stack exhaustion (infinite
2406 or very deeply nested recursion) has changed. Stack exhaustion
2407 detection is now done by write-protecting pages at the OS level
2408 and applies at all optimization settings; when found, a
2409 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2410 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2411 be caught using IGNORE-ERRORS.
2412 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2413 correctly when transforming recursive LABELS functions to LETs.
2414 (thanks to Alexey Dejneka)
2415 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2416 that are names of constants or global variables.
2417 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2418 alien routines with docstrings.
2419 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2420 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2422 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2423 (thanks to Eric Marsden)
2424 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2425 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2426 * Bug fix: The SPARC backend can now compile functions involving
2427 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2428 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2429 to a functions expecting a PATHNAME-DESIGNATOR.
2430 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2431 lambda lists. (thanks to Alexey Dejneka)
2432 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2433 functions so that EXPORTing the name of the class doesn't cause
2434 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2435 to Antonio Martinez for reporting this)
2436 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2437 constant-folded without failing an assertion. (thanks to Einar
2438 Floystad Dorum for reporting this)
2439 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2440 types are now dealt with more correctly by the compiler.
2441 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2442 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2443 behaviour analogous to RENAME-FILE. This puts its behaviour more
2444 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2446 * The fasl file version number has changed again. (because of the
2447 bug fix involving the names of PCL MAKE-INSTANCE functions)
2449 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2450 * bug fix: Floating point exceptions are treated much more
2451 consistently on the x86/Linux and PPC/Linux platforms.
2452 * Array initialization with :INITIAL-ELEMENT is now much faster for
2453 cases when the compiler cannot open code the array creation, but
2454 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2455 array accesses have also seen a speed increase.
2456 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2457 specified. (This was at the root of some bad interactions between
2458 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2460 * bug fix: Internal error arguments for undefined functions are now
2461 computed correctly on the PPC/Linux platform.
2462 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2463 Raymond Toy's patch for CMU CL)
2464 * Support for the Solaris 9 operating environment has been included
2465 (thanks to Daniel Merritt)
2466 * A very ugly but hopefully complete draft of the missing FFI chapter
2467 of the manual has been created by reformatting the corresponding
2468 CMU CL manual chapter into (currently very ugly and incoherent)
2469 DocBook and bringing it up to date for SBCL behavior. Thus, the
2470 manual is now essentially complete, at least by my extreme
2471 once-and-only-once standards, whereby it's acceptable to refer to
2472 the doc strings of SB-EXT functions as the primary documentation.
2473 * The fasl file version number has changed again, due to cleanup of
2474 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2475 is no longer a static symbol.)
2477 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2478 * SBCL now builds with OpenMCL (version 0.12) as the
2479 cross-compilation host; also, more progress has been made toward
2480 bootstrapping under CLISP.
2481 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2483 * bug 158 fixed: The compiler can now deal with integer loop
2484 increments different from 1; fixing this turned out also to fix
2486 * bug 169 fixed: no more bogus warnings about using lexical bindings
2487 despite the presence of perfectly good SPECIAL declarations (thanks
2488 to David Lichteblau)
2489 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2490 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2491 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2493 * bug 180 fixed: Method combination specifications no longer ignore
2494 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2495 * bug fix: Structure type predicate functions now check their argument
2496 count as they should.
2497 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2498 correctly. (thanks to Pierre Mai)
2499 * minor incompatible change: The --noprogrammer option is deprecated
2500 in favor of the new --disable-debugger option, which is very similar.
2501 (The major difference is that it takes effect at a slightly different
2502 time at startup, causing handling of errors in --sysinit and
2503 --userinit files will be affected differently.) The
2504 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2505 been added to allow this functionality to be controlled from ordinary
2506 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2507 maintainers, who might want to run non-interactive scripts to
2508 build SBCL cores which will later be used interactively.)
2509 * minor incompatible change: The LOAD function no longer, when given
2510 a wild pathname to load, loads all files matching that pathname.
2511 Instead, an error of type FILE-ERROR is signalled.
2513 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2514 * bug 147 fixed: The compiler preserves its block link/count
2515 invariants more correctly now so that it doesn't crash. (thanks
2517 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2519 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2520 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2521 no-value-for-this-character as NIL (as specified by ANSI).
2522 * HOST-NAMESTRING on physical pathnames now returns a string that is
2523 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2524 (thanks to Christophe Rhodes)
2525 * The Alpha port handles icache flushing more correctly. (thanks to
2527 * More progress has been made toward bootstrapping under CLISP. (thanks
2528 to Christophe Rhodes)
2529 * The fasl file format has changed again, because dynamic loading
2530 on OpenBSD (which has non-ELF object files) motivated some cleanups
2531 in the way that foreign symbols are transformed and passed around.
2532 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2533 is no longer treated as whitespace by the reader, but instead as
2534 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2535 |A
\7fB|, instead of A as it used to.
2537 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2538 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2539 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2540 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2541 this as of 0.7.1.45, but was left out of the previous news section
2542 (thanks to Dan Barlow)
2543 * SBCL now runs on the Solaris operating system on SPARC architectures
2544 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2545 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2546 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2547 * SPARC backend cleanups, allowing builds of cores optimized for V8
2548 and V9 SPARCS, and also emission of code targeted to a particular
2549 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2551 * SBCL is closer to bootstrapping under CLISP, thanks to various
2552 fixes by Christophe Rhodes.
2553 * The fasl file format has changed again, to allow the compiler's
2554 INFO database to support symbol macros.
2555 * The user manual (in doc/) is formatted into HTML more nicely.
2556 (thanks to coreythomas)
2557 * The system is smarter about SUBTYPEP relationships, especially
2558 those involving NOT types (including types such as ATOM which are
2559 represented internally using NOT types). Thus SUBTYPEP is less
2560 likely to return (VALUES NIL NIL) in general, and in particular
2561 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2563 * The fasl file format has changed again, because the internal
2564 representation of types now includes a new slot to support the new
2565 SUBTYPEP-of-NOT-types logic.
2566 * (not a change in the main branch of SBCL, but a related prototype
2567 which can hopefully be merged into the main branch of SBCL in the
2568 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2569 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2570 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2571 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2572 dispatch macro character combinations. (thanks to Alexey Dejneka)
2573 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2574 that superficially look like logical namestrings correctly.
2575 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2577 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2578 default supplied-p) for &optional and &key arguments. (thanks to
2581 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2582 * incompatible change: The compiler is now less aggressive about
2583 tail call optimization, doing it only when (> SPACE DEBUG) or
2584 (> SPEED DEBUG). (This is an incompatible change because there are
2585 programs which relied on the old CMU-CL-style behavior to optimize
2586 away their unbounded recursion which will now die of stack overflow.)
2587 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2588 for non-GENCGC systems has been increased to 20M (since that
2589 seems much closer to the likely performance optimum for modern
2590 systems than the old 4M value was)
2591 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2592 and *DEBUG-PRINT-LEVEL*
2593 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2594 of CMU CL's support for SPARC, and various endianness and other
2595 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2596 * new syntactic sugar for the Unix command line: --load foo.bar is now
2597 an alternate notation for --eval '(load "foo.bar")'.
2599 ** The system now detects stack overflow and handles it gracefully,
2600 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2601 optimization settings. (This is a good thing in general, and
2602 its introduction in this version should be particularly timely
2603 for anyone whose code fails because of suppression of tail
2605 ** The system now hunts for the C variable "environ" in a more
2606 devious way, to avoid segfaults when the C library version differs
2607 between compile time and run time. (thanks to Christophe Rhodes)
2608 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2609 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2610 ** The compiler no longer issues bogus style warnings for undefined
2611 classes in the same source file as the DEFCLASSes which defined
2612 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2614 ** fixes in CONDITION class precedence list for undefined function
2615 errors (thanks to Alexei Dejneka)
2616 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2617 correctly. (thanks to Dan Barlow)
2618 ** portability fixes aiming at bootstrapping under CLISP (thanks
2619 to Dave McDonald and Christophe Rhodes)
2620 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2621 ** fixes in type translation and and type inference (thanks to
2623 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2624 ** various fixes in the new ports (thanks to Dan Barlow)
2625 * several changes related to debugging:
2626 ** suppression of tail recursion, as noted above
2627 ** stack overflow detection, as noted above
2628 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2629 is now the default. (For some time encapsulation has been more
2630 reliable than the breakpoint-based :ENCAPSULATE NIL
2631 implementation, at least on X86 systems; and I just noticed that
2632 encapsulation also seems closer to the spirit of the ANSI
2635 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2637 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2638 up properly again. (There was a packaging bug in 0.7.0 which
2639 left their definitions in SB-SYS::LOAD-FOREIGN and
2640 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2641 vital for most things which interface to C-level interfaces,
2642 like extensions working with sockets or databases or
2643 Perl-compatible regexes or whatever, and the need to fix
2644 this bug is the main reason that 0.7.1 was released so
2646 ** DEFGENERIC is now choosier about the methods it redefines, so that
2647 reLOADing a previously-LOADed file containing DEFGENERICs does
2648 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2649 cycle now works as it should. (thanks to Alexey Dejneka)
2650 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2651 fixed by Christophe Rhodes. (He actually submitted this patch
2652 months ago, and I delayed until after 0.7.0.)
2653 ** Bug 111 (internal compiler confusion about runtime checks on
2654 FUNCTION types) was fixed by Alexey Dejneka.
2655 * Some internal cleanups (getting rid of variables which aren't
2656 needed now that the byte interpreter is gone) caused the fasl
2657 file format number to change again.
2659 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2660 * major incompatible change: The default fasl file extension, i.e. the
2661 default extension for files produced by COMPILE-FILE, has changed
2662 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2664 ** There are many changes in the implementation of the compiler.
2665 SBCL is now essentially a compiler-only implementation of ANSI
2666 Common Lisp. EVAL still "interprets" a few special cases, but
2667 almost all the interesting cases are handled by creating
2668 a LAMBDA expression, calling COMPILE on it, then calling
2669 FUNCALL on the result.
2670 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2671 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2672 Since the code is newer, there might still be some new bugs
2673 (though not as many as before Martin Atzmueller's fixes:-). But
2674 the new code is substantially simpler and clearer, and hopefully
2675 any remaining bugs will be simpler, less fundamental, and more
2676 fixable then the bugs in the old code.
2677 ** The revised compiler is still a little unsteady on its feet.
2679 *** The debugging information it produces (particularly the names
2680 of FUNCTION objects) is sometimes much less useful than what
2681 the old compiler produced.
2682 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2683 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2684 within a MACROLET) has been temporarily weakened.
2685 ** There are new compiler optimizations for various functions:
2686 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2687 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2688 *** the math functions TRUNCATE, FLOOR, and CEILING
2689 *** the function-of-all-trades COERCE
2690 Mostly these should be transparent, but there's one
2691 potentially-annoying problem (bug 117): when the compiler
2692 inline-expands a function and does type analysis on the result,
2693 it can create control paths which have type mismatches, and
2694 when it can't prove that those control paths aren't taken,
2695 it will issue WARNINGs about the type mismatches. This is
2696 a particular problem in practice for the new sequence functions.
2697 It's not clear how this should be fixed, and for now, a
2698 workaround is given in the entry for 117 in the BUGS file.
2699 ** (Because of the interaction between the two previous items --
2700 occasional inlining problems and new inline expansions -- some
2701 of the new sequence function optimizations won't really kick in
2702 completely until debugging information, and then inlining, are
2703 straightened out in some future version.)
2704 * minor incompatible changes:
2705 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2706 DIRECTORY no longer implicitly promotes NIL slots of its
2707 pathname argument to :WILD. In particular, when you ask for the
2708 contents of a directory (which you used to be able to do without
2709 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2710 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2711 ** changes in behavior that ANSI explicitly defines to be
2712 implementation dependent:
2713 *** The new compiler-only implementation still conforms with ANSI,
2714 but acts a little different than before. Besides the obvious
2715 changes in performance tradeoffs (that the cost per form passed
2716 to EVAL has gone up, and the cost per form executed by EVAL
2717 has gone down), the behavior of the system changes a little
2718 because there are no longer any interpreted function objects.
2719 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2720 e.g. doing COMPILE on the output of interactive DEFUN is
2722 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2724 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2725 has changed from (:CL) to NIL.
2726 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2727 e.g. #\Newline and #\Tab, don't have names specified in the
2728 ANSI Common Lisp standard, is now based on their ASCII symbolic
2729 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2730 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2731 are no longer used for output.
2732 ** changes in internal implementation constants:
2733 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2734 4 million. (If your application spends a lot of time GCing and
2735 you have a lot of RAM, you might want to experiment with
2736 increasing it even more.)
2737 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2738 However, almost all old code should still continue to work without
2739 immediate update, as SB-C-CALL is now a (deprecated) nickname
2741 ** Old operator names in the style DEF-FOO are now deprecated in
2742 favor of new corresponding names DEFINE-FOO, for consistency with
2743 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2744 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2745 internal symbols, but a few supported extensions like
2746 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2747 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2748 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2749 etc. as you get deeper into recursive calls to the debugger
2750 command loop, instead of the old "5]", "5]]", "5]]]"
2751 sequence. (I was motivated to do this when squabbles between
2752 ILISP and SBCL left me very deeply nested in the debugger. In the
2753 short term, this change will probably provoke more ILISP/SBCL
2754 squabbles, but hopefully it will be an improvement in the long run.)
2755 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2756 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2757 function, since declaiming return types involving aliens is
2758 (1) annoyingly messy to do by hand and (2) vital to efficient
2759 compilation of code which calls such functions.
2760 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2761 longer reexported by the SB-EXT package. They're solely useful
2762 for alien code, so it seems more logical that you should get
2763 them from the SB-ALIEN package, not in SB-EXT.
2764 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2765 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2766 features. Instead, the code that they used to control is always
2767 built into the system.
2768 * many other bug fixes
2769 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2770 advantage of the new EVAL-WHEN stuff and to clean them up in
2771 general, and they are now more ANSI-compliant in a number of
2772 ways. Martin Atzmueller is responsible for a lot of this.
2773 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2775 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2776 *** correct ERROR type for various file operations
2777 *** some fixes for Lisp streams
2778 *** DEFMETHOD syntax checking
2779 *** changing old weird representation of debug information as
2780 strings (which, among their other deficiencies, don't transform
2781 correctly when you rename packages, and don't change their
2782 print representation when you change things like *PACKAGE*
2783 and *PRINT-LENGTH*) to symbols and lists of symbols
2784 He also made several improvements and fixed several bugs in DESCRIBE.
2785 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2787 *** misbehavior of WRITE-STRING/WRITE-LINE
2788 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2789 and several other LOOP problems as well
2790 *** DIRECTORY when similar filenames are present
2791 *** DEFGENERIC with :METHOD options
2792 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2793 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2794 *** argument ordering in FIND with :TEST option
2795 *** mishandled package designator argument in APROPOS-LIST
2796 *** various problems in the backquote readmacro
2797 *** a bug in APROPOS
2798 *** probably some others that I'm not describing very well here,
2799 since the CVS log documents them by reference to sbcl-devel
2800 messages, and the SourceForge archives aren't working well.:-(
2801 ** Dan Barlow improved the Alpha port (and is making progress on the
2802 PPC port, for those of you who think different).
2803 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2804 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2805 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2806 continues to work on the SPARC port (for those of you in a position
2807 to look down upon our little PC-compatible boxes from a great height).
2808 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2809 and uses that copy, rather than the current dynamic value, when
2810 it's trying to decide whether to truncate output. Thus e.g.
2811 (let ((*print-lines* 50))
2812 (pprint-logical-block (stream nil)
2814 (let ((*print-lines* 8))
2815 (print (aref possiblybigthings i) stream)))))
2816 should now truncate the logical block only at 50 lines, instead of
2817 often truncating it at 8 lines, as it did before.
2818 * The doc/cmucl/ directory, containing old CMU CL documentation
2819 from the time of the fork, is no longer part of the base system.
2820 SourceForge has shut down its anonymous FTP service, and with it
2821 my original plan for distributing the old CMU CL documentation
2822 there. For now, if you need these files you can download an old
2823 SBCL source release and extract them from it.
2824 * The fasl file version number changed again, for dozens of reasons,
2825 some of which are apparent above.
2827 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2828 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2829 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2830 patch, so that DISASSEMBLE output is much nicer.
2831 * The code in the SB-PROFILE package now seems reasonably stable.
2832 I still haven't decided what the final interface should look like
2833 (I'd like PROFILE to interact cleanly with TRACE, since both
2834 facilities use function encapsulation) but if you have a need
2835 for profiling now, you can probably use it successfully with
2836 the current CMU-CL-style interface.
2837 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2838 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2839 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2840 link now returns the dangling link itself, and for similar
2841 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2842 link itself. (In these cases the old code signalled an error and
2843 looped endlessly, respectively.) Thus, DIRECTORY now works even
2844 in the presence of dangling and cyclic symbolic links.
2845 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2846 is now a supported extension again, since the consensus on
2847 sbcl-devel was that it can be useful for ordinary development
2848 work, not just for debugging SBCL itself.
2849 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2850 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2851 that currently-defined functions might be redefined later with
2852 different return types.
2853 * Hash tables can be printed readably, as inspired by CMU CL code
2854 of Eric Marsden and SBCL code of Martin Atzmueller.
2855 * better error handling in CLOS method combination, thanks to
2856 Martin Atzmueller porting Pierre Mai's CMU CL patches
2857 * more overflow fixes for >16Mbyte I/O buffers
2858 * A bug in READ has been fixed, so that now a single Ctrl-D
2859 character suffices to cause end-of-file on character streams.
2860 In particular, now you only need one Ctrl-D at the command
2861 line (not two) to exit SBCL.
2862 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2864 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2865 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2866 does the right thing.
2867 * The implementation of some type tests, especially for CONDITION
2868 types, is now tidier and maybe faster, due to CMU CL code
2869 originally by Douglas Crosher, ported by Martin Atzmueller.
2870 * Some math functions have been fixed, and there are new
2871 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2872 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2873 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2874 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2875 likely to be built into sbcl-0.7.0.)
2876 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2877 CL, ported to SBCL by Martin Atzmueller)
2878 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2879 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2880 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2881 * There's a new slam.sh hack to shorten the edit/compile/debug
2882 cycle for low-level changes to SBCL itself, and a new
2883 :SB-AFTER-XC-CORE target feature to control the generation of
2884 the after-xc.core file needed by slam.sh.
2885 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2886 COMPILE-FILE is no longer supported, so that now every function
2887 gets an entry point, so that block compilation looks a little
2888 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2889 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2890 returns the number of bytes consed since the system started,
2891 rather than the number consed since the first time the function
2892 was called. (The new definition parallels ANSI functions like
2893 CL:GET-INTERNAL-RUN-TIME.)
2894 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2895 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2896 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2897 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2898 * Fasl file version numbers are now independent of the target CPU,
2899 since historically most system changes which required version
2900 number changes have affected all CPUs equally. Similarly,
2901 the byte fasl file version is now equal to the ordinary
2904 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2905 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2906 is no longer recognized. I apologize for this, because it was
2907 listed in SB-EXT as a supported extension, but I found that
2908 its existing behavior was poorly specified, as well as incorrectly
2909 specified, and it looked like too much of a mess to straighten it
2910 out. I have enough on my hands trying to get ANSI stuff to work..
2911 * many patches ported from CMU CL by Martin Atzmueller, with
2912 half a dozen bug fixes in pretty-printing and the debugger, and
2913 half a dozen others elsewhere
2914 * fixed bug 13: Floating point infinities are now supported again.
2915 They might still be a little bit flaky, but thanks to bug reports
2916 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2917 as flaky as they were.
2918 * The --noprogrammer command line option is now supported. (Its
2919 behavior is slightly different in detail from what the old man
2920 page claimed it would do, but it's still appropriate under the
2921 same circumstances that the man page talks about.)
2922 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2923 are now supported, and enabled by default. Thus, the compiler can
2924 handle many floating point and complex operations much less
2925 inefficiently. (Thus e.g. you can implement a complex FFT
2927 * The compiler now detects type mismatches between DECLAIM FTYPE
2928 and DEFUN better, and implements CHECK-TYPE more correctly, and
2929 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2930 thanks to patches from Martin Atzmueller.
2931 * various fixes to make the cross-compiler more portable to
2932 ANSI-conforming-but-different cross-compilation hosts (notably
2933 Lispworks for Windows, following bug reports from Arthur Lemmens)
2934 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2935 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2936 fixed thanks to Pierre Mai's CMU CL patches.
2937 * fixes to keep the system from overflowing internal counters when
2938 it tries to use i/o buffers larger than 16M bytes
2939 * fixed bug 45a: Various internal functions required to support
2940 complex special functions have been merged from CMU CL sources.
2941 (When I was first setting up SBCL, I misunderstood a compile-time
2942 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2943 * improved support for type intersection and union, fixing bug 12
2944 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2945 more obscure bugs as well
2946 * some steps toward byte-compiling non-performance-critical
2947 parts of the system, courtesy of patches from Martin Atzmueller
2948 * Christophe Rhodes has made some debian packages of sbcl at
2949 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2950 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2951 stable, but are nonetheless usable. When he's ready, I'd be happy
2952 to add them to the SourceForge "File Releases" section. (And if
2953 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2954 * new fasl file format version number (because of changes in
2955 internal representation of (OR ..) types to accommodate the new
2956 support for (AND ..) types, among other things)
2958 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2959 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2961 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2963 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2964 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2965 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2966 * DESCRIBE now works on structure objects again.
2967 * Most function call argument type mismatches are now handled as
2968 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2969 know whether the function will be redefined before the call is
2970 executed. (The compiler could flag local calls with full WARNINGs,
2971 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2972 it doesn't keep track of enough information to know whether calls
2973 are local in this sense.)
2974 * Compiler output is now more verbose, with messages truncated
2975 later than before. (There should be some supported way for users
2976 to override the default verbosity, but I haven't decided how to
2977 provide it yet, so this behavior is still controlled by the internal
2978 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2979 src/compiler/ir1util.lisp.)
2980 * Fasl file format version numbers have increased again, because
2981 support for the Gray streams extension changes the layout of the
2982 system's STREAM objects.
2983 * The Gray subclassable streams extension now works, thanks to a
2984 patch from Martin Atzmueller.
2985 * The full LOAD-FOREIGN extension (not just the primitive
2986 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2987 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2988 but like most other programs, it defaults to copying the Unix
2989 environment from the original process instead of starting the
2990 new process in an empty environment.
2991 * Extensions which manipulate the Unix environment now support
2992 an :ENVIRONMENT keyword option which doesn't smash case or
2993 do other bad things. The CMU-CL-style :ENV option is retained
2994 for porting convenience.
2995 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2996 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2998 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3000 * A patch from Martin Atzmueller seems to have solved the SIGINT
3001 problem, and as far as we know, signal-handling now works cleanly.
3002 (If you find any new bugs, please report them!)
3003 * The system no longer defaults Lisp source file names to types
3004 ".l", ".cl", or ".lsp", but only to ".lisp".
3005 * The compiler no longer uses special default file extensions for
3006 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3007 seems to expect a single default extension for all compiled code,
3008 and there's no compelling reason to try to stretch the standard
3009 to allow two different extensions.) Instead, byte-compiled files
3010 default to the same extension as native-compiled files.
3011 * Fasl file format version numbers have increased again, because
3012 a rearrangement of internal implementation packages made some
3013 dumped symbols in old fasl files unreadable in new cores.
3014 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3015 many fewer weird special cases.
3016 * Bug #17 (differing COMPILE-FILE behavior between logical and
3017 physical pathnames) has been fixed, and some related misbehavior too,
3018 thanks to a patch from Martin Atzmueller.
3019 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3020 by Tim Moore, ported to SBCL by Martin Atzmueller.
3021 * Martin Atzmueller fixed several filesystem-related problems,
3022 including bug #36, in part by porting CMU CL patches, which were
3023 written in part by Paul Werkowski.
3024 * More compiler warnings in src/runtime/ are gone, thanks to
3025 more patches from Martin Atzmueller.
3026 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3029 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3031 * DESCRIBE now works on CONDITION objects.
3032 * The debugger now handles errors which arise when trying to print
3033 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3035 * The build system now uses an additional file, customize-target-features.lisp,
3036 to allow local modifications to the target *FEATURES* list. (The point of
3037 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3038 debugging features enabled, without having to worry about propagating it
3039 into everyone's system when I do a "cvs update".) When no
3040 customize-target-features.lisp file exists, the target *FEATURES* list
3041 should be constructed the same way as before.
3042 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3043 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3044 warn about behavior which is undefined under the ANSI spec. Note: This
3045 is specified by ANSI, but it's not very popular with programmers.
3046 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3047 macro in the SBCL sources for an example of a workaround which you
3048 might use to make portable ANSI-standard code which does what you want.
3049 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3050 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3051 when it's used as a non-toplevel form. (This is required in order
3052 to implement the DEFCONSTANT-EQX macro.)
3053 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3054 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3055 funny things, due to bugs in the implementation of EVAL-WHEN
3056 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3057 non-toplevel DEFCONSTANTs are uncommon.)
3058 * The core file version number and fasl file version number have been
3059 incremented, because the old noncompliant DEFCONSTANT behavior involved
3060 calling functions which no longer exist, and because I also took the
3061 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3062 * fixed bug 1 (error handling before read-eval-print loop starts), and
3063 redid debugger restarts and related debugger commands somewhat while
3065 ** The QUIT debugger command is gone, since it did something
3066 rather different than the SB-EXT:QUIT command, and since it never
3067 worked properly outside the main toplevel read/eval/print loop.
3068 Invoking the new TOPLEVEL restart provides the same functionality.
3069 ** The GO debugger command is also gone, since you can just invoke
3070 the CONTINUE restart directly instead.
3071 ** The TOP debugger command is also gone, since it's redundant with the
3072 FRAME 0 command, and since it interfered with abbreviations for the
3074 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3076 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3077 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3078 some porting work by Martin Atzmueller)
3079 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3080 with a boatload of patches which he ported to SBCL. Now that those
3082 ** The system tries to make sure that its low-priority messages
3083 are prefixed by semicolons, to help people who like to use
3084 syntax highlighting in their ILISP buffer. (This patch
3085 was originally due to Raymond Toy.)
3086 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3087 patches originally written by Raymond Toy.
3088 ** The compiler understands coercion between single-value and
3089 multiple-VALUES type expressions better, getting rid of some very
3090 weird behavior, thanks to patches originally by Robert MacLachlan
3091 and Douglas Crosher.
3092 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3093 lambda lists, thanks to a patch originally by Pierre Mai.
3094 ** The system no longer bogusly warns about "abbreviated type
3096 ** The compiler gets less confused by inlining and RETURN-FROM,
3097 thanks to some patches originally by Tim Moore.
3098 ** The system no longer hangs when dumping circular lists to fasl
3099 files, thanks to a patch originally from Douglas Crosher.
3100 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3101 undefined function error.
3102 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3103 in structure slot initforms) for now, documented workaround instead:-|
3104 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3105 * fixed bug 5: added stubs for various Gray stream functions called
3106 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3107 installed, at least appropriate type errors are generated
3108 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3109 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3110 consistently in DEFMETHOD forms.
3111 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3112 it doesn't seem to affect SBCL after all
3113 * The C runtime system now builds with better optimization and many
3114 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3116 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3118 * The system is now under CVS at SourceForge (instead of the
3119 CVS repository on my home machine).
3120 * The new signal handling code has been tweaked to treat register
3121 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3122 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3123 downstream, e.g. in debug-int.lisp, has implicit dependencies
3124 on the unsignedness of integer representation of machine words,
3125 and that caused the system to bomb out with infinite regress
3126 when trying to recover from type errors involving signed values,
3127 e.g. (BUTLAST '(1 2 3) -1).)
3128 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3129 (This was one of the bugs Peter Van Eynde reported back in July.)
3130 * The system now uses code inspired by Colin Walters' O(N)
3131 implementation of MAP (from the cmucl-imp@cons.org mailing
3132 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3133 inline the MAP operation, and there is more than one
3134 sequence argument to the MAP call (so that it can't just
3135 do ETYPECASE once and for all based on the type of the
3136 single sequence argument). (The old non-inline implementation
3137 of the general M-argument sequence-of-length-N case required
3138 O(M*N*N) time when any of the sequence arguments were LISTs.)
3139 * The QUIT :UNIX-CODE keyword argument has been renamed to
3140 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3141 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3143 * Raymond Wiker's patches to port dynamic loading from Linux to
3144 FreeBSD have been added.
3145 * The BUGS file is now more nearly up to date, thanks in large part
3146 to Martin Atzmueller's review of it.
3147 * The debugger now flushes standard output streams before it begins
3148 its output ("debugger invoked" and so forth).
3149 * The core version number and fasl file version number have both
3150 been incremented, because of incompatible changes in the layout
3152 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3153 used to not be called for a saved Lisp image.)
3154 * Martin Atzmueller's version of a patch to fix a compiler crash,
3155 as posted on sbcl-devel 13 September 2000, has been installed.
3156 * Instead of installing Martin Atzmueller's patch for the
3157 compiler transform for SUBSEQ, I deleted the compiler transform,
3158 and transforms for some similar consing operations.
3159 * A bug in signal handling which kept TRACE from working on OpenBSD
3161 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3162 away by constant folding
3163 * The system now defines its address space constants in one place
3164 (in the Lisp sources), and propagates them automatically elsewhere
3165 (through GENESIS and the sbcl.h file). Therefore, patching the
3166 address map is less unnecessarily tedious and error-prone. The
3167 Lisp names of address space constants have also been systematized.
3168 * CVS tags like dollar-Header-dollar have been removed from
3169 the sources, because they have never saved me trouble and
3170 they've been source of trouble working with patches and other
3171 diff-related operations.
3172 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3173 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3175 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3177 * The system has been ported to OpenBSD.
3178 * The system now compiles with a simple "sh make.sh" on the systems
3179 that it's supported on. I.e., now you no longer need to tweak
3180 text in the target-features.lisp-expr and symlinks in src/runtime/
3181 by hand, the make.sh takes care of it for you.
3182 * The system is no longer so grossly inefficient when compiling code
3183 involving vectors implemented as general (not simple) vectors (VECTOR T),
3184 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3185 which dares to use the various sequence functions on non-simple
3186 vectors, takes less of a performance hit.
3187 * There is now a primitive type predicate VECTOR-T-P
3188 to test for the (VECTOR T) type, so that e.g.
3189 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3190 can now be compiled with some semblance of efficiency. (The old code
3191 turned the type declaration into a full call to %TYPEP at runtime!)
3192 * AREF on (VECTOR T) is still not fast, since it's still compiled
3193 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3194 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3195 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3196 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3197 without making full calls to SUBTYPEP at runtime.
3198 (Some analogous efficiency issues for non-simple vectors specialized to
3199 element types other than T, or for non-simple multidimensional arrays,
3200 have not been addressed. They could almost certainly be handled the
3201 same way if anyone is motivated to do so.)
3202 * The changes in array handling break binary compatibility, so
3203 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3204 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3205 instead of (VALUES T T).
3206 * By following the instructions that Dan Barlow posted to sbcl-devel
3207 on 2 July 2000, I was able to enable primitive dynamic object
3208 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3209 functionality is not implemented (since it calls ld to resolve
3210 library references automatically, requiring RUN-PROGRAM for its
3211 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3212 try to resolve library references) is now supported.
3213 * The system now flushes the standard output streams when it terminates,
3214 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3215 them at several other probably-convenient times, e.g. in each pass of
3216 the toplevel read-eval-print loop, and after evaluating a form given
3217 as an "--eval" command-line option. (These changes were motivated by a
3218 discussion of stream flushing issues on cmucl-imp in August 2000.)
3219 * The source transform for TYPEP of array types no longer assumes
3220 that an array whose element type is a not-yet-defined type
3221 is implemented as an array of T, but instead punts, so that the
3222 type will be interpreted at runtime.
3223 * There is now some support for cross-compiling in make.sh: each of
3224 the phases of make.sh has its own script. (This should be transparent
3225 to people doing ordinary, non-cross-compile builds.)
3226 * Since my laptop doesn't have hundreds of megabytes of memory like
3227 my desktop machine, I became more motivated to do some items on
3228 my to-do list in order to reduce the size of the system a little:
3229 ** Arrange for various needed-only-at-cold-init things to be
3230 uninterned after cold init. To support this, those things have
3231 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3232 symbols with such names are now uninterned after cold init).
3233 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3234 fasl files for cold load.
3235 ** Remove the old compiler structure pooling code (which used to
3236 be conditional on the target feature :SB-ALLOC) completely.
3237 ** Redo the representation of some data in cold init to be more compact.
3238 (I also looked into supporting byte compiled code at bootstrap time,
3239 which would probably reduce the size of the system a lot, but that
3240 looked too complicated, so I punted for now.)
3241 * The maximum signal nesting depth in the src/runtime/ support code has
3242 been reduced from 4096 to 256. (I don't know any reason for the very
3243 large old value. If the new smaller value turns out to break something,
3244 I'll probably just bump it back up.)
3245 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3247 * Many, many bugs reported by Peter Van Eynde have been added to
3248 the BUGS list; some have even been fixed.
3249 * While enabling dynamic object file loading, I tried to make the
3250 code easier to understand, renaming various functions and variables
3251 with less ambiguous names, and changing some function calling
3252 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3253 * While trying to figure out how to do the OpenBSD port, I tried to
3254 clean up some of the code in src/runtime/. In particular, I dropped
3255 support for non-POSIX signal handling, added various comments,
3256 tweaked the code to reduce the number of compilation warnings, and
3257 renamed some files to increase consistency.
3258 * To support the new automatic configuration functionality in make.sh,
3259 the source file target-features.lisp-expr has been replaced with the
3260 source file base-target-features.lisp-expr and the machine-generated
3261 file local-target-features.lisp-expr.
3262 * fixed a stupid quoting error in make.sh so that using CMU CL
3263 "lisp -batch" as cross-compilation host works again
3265 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3267 * DESCRIBE no longer tries to call itself recursively to describe
3268 bound/fbound values, so that it no longer fails on symbols which are
3269 bound to themselves (like keywords, T, and NIL).
3270 * DESCRIBE now works on generic functions.
3271 * The printer now prints less-screwed-up representations of closures
3272 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3273 * A private symbol is used instead of the :EMPTY keyword previously
3274 used to mark empty slots in hash tables. Thus
3275 (DEFVAR *HT* (MAKE-HASH-TABLE))
3276 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3277 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3278 now does what ANSI says that it should. (You can still get
3279 similar noncompliant behavior if bang on the hash table
3280 implementation with all the symbols you get back from
3281 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3282 This breaks binary compatibility, since tests for equality to
3283 :EMPTY are wired into things like the macroexpansion of
3284 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3286 * There's now a minimal placeholder implementation for CL:STEP,
3287 as required by ANSI.
3288 * An obscure bug in the interaction of the normal compiler, the byte
3289 compiler, inlining, and structure predicates has been patched
3290 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3291 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3293 * Missing ordinary arguments in a macro call are now detected even
3294 when the macro lambda list contains &KEY or &REST.
3295 * The debugger no longer complains about encountering the top of the
3296 stack when you type "FRAME 0" to explicitly instruct it to go to
3297 the top of the stack. And it now prints the frame you request even
3298 if it's the current frame (instead of saying "You are here.").
3299 * As specified by ANSI, the system now always prints keywords
3300 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3302 * The default initial SIZE of HASH-TABLEs is now smaller.
3303 * Type information from CLOS class dispatch is now propagated
3304 into DEFMETHOD bodies, so that e.g.
3305 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3307 is now basically equivalent to
3308 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3309 (DECLARE (TYPE SINGLE-FLOAT X))
3311 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3312 operation, without having to do run-time type dispatch.
3313 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3314 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3315 * Since I don't seem to be making big file reorganizations very often
3316 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3317 is overflowing my ability to conveniently back them up), I've finally
3318 checked the system into CVS. (The CVS repository is on my home system,
3319 not at SourceForge -- putting it on SourceForge might come later.)
3320 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3321 high-level GC-NOTIFY-FOO functions send their output. (There's
3322 still very little control of where low-level verbose GC functions
3323 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3324 less than it used to -- the GC-NOTIFY-FOO functions are now under
3325 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3326 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3327 in only one place in the source code, and propagates it automatically
3328 everywhere that it's needed. Thus e.g. when I bump the version from
3329 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3330 * The C source files now include boilerplate legalese and documentation
3331 at the head of each file (just as the Lisp source files already did).
3332 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3333 to his page will be replaced with a link to his new CLiki service.
3335 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3337 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3338 * The build process now looks for GNU make under the default name "gmake",
3339 instead of "make" as it used to. If GNU make is not available as "gmake"
3340 on your system, you can change this default behavior by setting the
3341 GNUMAKE environment variable.
3342 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3343 can build without error under CMU CL.
3345 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3347 * There is now a partial SBCL user manual (with some new text and some
3348 text cribbed from the CMU CL manual).
3349 * The beginnings of a profiler have been added (starting with the
3350 CMU CL profiler and simplifying and cleaning up). Eventually the
3351 main interface should be through the TRACE macro, but for now,
3352 it's still accessed through vaguely CMU-CL-style functions and macros
3353 exported from the package SB-PROFILE.
3354 * Some problems left over from porting CMU CL to the new
3355 cross-compilation bootstrap process have been cleaned up:
3356 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3357 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3358 part when getting rid of the compiler *BACKEND* stuff.)
3359 ** The value of *NULL-TYPE* was screwed up, because it was
3360 being initialized before the type system knew the final
3361 definition of the 'NULL type. This screwed up several key
3362 optimizations in the compiler, causing inefficiency in all sorts
3363 of places. (I found it because I wanted to understand why
3364 GET-INTERNAL-RUN-TIME was consing.)
3365 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3366 PROCLAIM FTYPE information. Unfortunately this broke binary
3367 compatibility again, since now the forms output by DEFGENERIC
3368 to refer to functions which didn't exist in 0.6.3.
3369 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3370 can use the new (as of 0.6.3) transform for SOME into MAP into
3372 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3373 values to (INTEGER 0 1000000), so that the time code will no longer
3374 occasionally get blown up by Linux returning 1000000 microseconds
3375 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3376 its output conform to the ANSI spec. (Alas, this makes its output
3377 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3378 * A full call to MAP NIL with a single sequence argument no longer conses.
3379 * fixes to problems pointed out by Martin Atzmueller:
3380 * The manual page no longer talks about multiprocessing as though
3381 it were currently supported.
3382 * The ILISP support patches have been removed from the distribution,
3383 because as of version 5.10.1, ILISP now supports SBCL without us
3384 having to maintain patches.
3385 * added a modified version of Raymond Toy's recent CMU CL patch for
3386 EQUALP comparison of HASH-TABLE
3388 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3390 * The system still can't cross-compile itself with
3391 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3392 entails), but at least it can compile more of itself that way
3393 than it used to be able to, and various buglets which were uncovered
3394 by trying to cross-compile itself that way have now been fixed.
3395 * This release breaks binary compatibility again. This time
3396 at least I've incremented the FASL file format version to 2, so that the
3397 problem can be detected reliably instead of just causing weird errors.
3398 * various new style warnings:
3399 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3400 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3401 ** using the lexical binding of a variable named in the *FOO* style
3402 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3403 as specified by ANSI.
3404 * *RANDOM-STATE* is no longer automatically initialized from
3405 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3406 default behavior of the system is to repeat its behavior every time
3407 it's run. If you'd like to change this behavior, you can always
3408 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3409 old convention there was no comparably easy way to get the system to
3410 repeat its behavior every time it was run.
3411 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3412 FTYPE declarations has been removed, in favor of their ANSI
3413 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3414 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3415 the types of their sequence arguments can be determined at compile time.
3416 This is done through a new open code expansion for MAP which eliminates
3417 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3418 argument types can be determined at compile time.
3419 * The optimizer now transforms COERCE into an identity operation when it
3420 can prove that the coerced object is already of the correct type. (This
3421 can be a win for machine generated code, including the output of other
3422 optimization transforms, such as the MAP transform above.)
3423 * Credit information has been moved from source file headers into CREDITS.
3424 * Source file headers have been made more standard.
3425 * The CASE macro now compiles without complaining even when it has
3428 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3430 * (Note that the way that the PCL macroexpansions were rewritten
3431 to accommodate the change in DEFGENERIC below breaks binary
3432 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3433 not run under sbcl-0.6.2. Once we get out of alpha releases,
3434 i.e. hit release 1.0.0, we'll probably try to maintain binary
3435 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3436 and sbcl-1.4.4. Until then, however, it might be fairly common
3437 for maintenance releases to break binary compatibility.)
3438 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3440 * The reporting of unhandled signals has been changed to print some
3441 explanatory text as well as the report form. (Previously only
3442 the report form was printed.)
3443 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3444 it defines, so that the compiler no longer issues undefined function
3445 warnings for compiled-but-not-yet-loaded generic functions.
3446 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3447 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3448 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3449 section of the ANSI spec are supported.
3450 * The "" nickname for the "KEYWORD" package has been removed.
3451 The reader still handles symbol tokens which begin with a package marker
3452 as keywords, but it doesn't expose its mechanism for doing so in the
3453 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3454 * The system now issues STYLE-WARNINGs for contradictory TYPE
3455 proclamations. (Warnings for contradictory FTYPE proclamations would
3456 be nice too, but those can't be done usefully unless the type system
3457 is made smarter about FUNCTION types.)
3458 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3459 systematized, so that "*target-*.lisp is supposed to exist only on the
3460 target and imply that there's a related file which exists on the
3461 host, and *host-*.lisp is supposed to exist only on the host and imply
3462 that there's a related file which exists on the target. This involves a
3463 lot of renaming. Hopefully the acute confusion caused by the renaming
3464 will be justified by the reduction in chronic confusion..
3465 ** runtime-type.lisp -> early-target-type.lisp
3466 ** target-type.lisp -> late-target-type.lisp
3467 ** early-host-format.lisp -> early-format.lisp
3468 ** late-host-format.lisp -> late-format.lisp
3469 ** host-error.lisp -> misc-error.lisp
3470 ** early-error.lisp -> early-target-error.lisp
3471 ** late-error.lisp -> late-target-error.lisp
3472 ** host-defboot.lisp -> early-defboot.lisp
3473 ** code/misc.lisp -> code/target-misc.lisp
3474 ** code/host-misc.lisp -> code/misc.lisp
3475 ** code/numbers.lisp -> code/target-numbers.lisp
3476 ** code/early-numbers.lisp -> numbers.lisp
3477 ** early-host-type.lisp -> early-type.lisp
3478 ** late-host-type.lisp -> late-type.lisp
3479 ** host-typep.lisp -> typep.lisp
3480 ** load.lisp -> target-load.lisp
3481 ** host-load.lisp -> load.lisp
3482 ** host-disassem.lisp -> disassem.lisp
3483 ** host-insts.lisp -> insts.lisp
3484 ** byte-comp.lisp -> target-byte-comp.lisp
3485 ** host-byte-comp.lisp -> byte-comp.lisp
3486 ** host-signal.lisp -> signal.lisp
3487 ** host-defstruct.lisp -> defstruct.lisp
3488 ** late-target-type.lisp -> deftypes-for-target.lisp
3489 Furthermore, several other previously target-only files foo.lisp (e.g.
3490 hash-table.lisp and random.lisp) have been split into a target-and-host
3491 foo.lisp file and a target-only target-foo.lisp file, with their key type
3492 definitions in the target-and-host part, so that the cross-compiler will
3493 know more about target types.
3494 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3495 gone away. In their place are various *BACKEND-FOO* variables
3496 corresponding to the slots of the old structure.
3497 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3498 into the target SBCL has been fixed.
3499 * The chill.lisp system for loading cold code into a running SBCL
3501 * Support for the CMU CL "scavenger hook" extension has been removed.
3502 (It was undocumented and unused in the CMU CL sources that SBCL was
3503 derived from, and stale in sbcl-0.6.1.)
3504 * Various errors in the cross-compiler type system were detected
3505 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3506 (enabling various consistency checks). Many of them were fixed,
3507 but some hard problems remain, so the compiler is back to
3508 running without *TYPE-SYSTEM-INITIALIZED* for now.
3509 * As part of the cross-compiler type system cleanup, I implemented
3510 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3511 * I have started adding UNCROSS calls throughout the type system
3512 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3513 on cross-compiler input files will be able to go away, and various
3515 * CONSTANTP now returns true for quoted forms (as explicitly required
3518 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3520 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3521 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3522 (MAKE-LIST -1). In the longer term, it ought to become true
3523 that declarations are assertions even at SAFETY 1. For now, it's not
3524 quite true even at SAFETY 3, but it's at least more nearly true..
3525 (Note that this change seems to increases the size of the system by
3526 O(5%) and to decrease the speed of the compiler by 20% or more.)
3527 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3528 for the problem of printing dozens of lines of distracting information
3529 about low-level system machinery as part of the top stack frame
3530 on entry to the debugger when an undefined function was called.
3531 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3532 is not reset to COMMON-LISP-USER.
3533 * Compilation of stuff related to dyncount.lisp has been made conditional
3534 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3535 smaller. The various dyncount-related symbols have been moved into
3536 a new "SB-DYNCOUNT" package.
3537 * tty-inspect.lisp has been renamed to inspect.lisp.
3538 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3539 feature has gone away. (When we eventually port to other flavors of
3540 libc and/or Unix, we'll try to make the differences between flavors
3541 invisible at the user level.)
3542 * Various other *FEATURES* tags, and/or their associated conditionals,
3543 have been removed if obsolescent, or given better documentation, or
3544 sometimes given more-mnemonic names.
3546 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3548 * tidied up "make.sh" script
3549 * tidied up system directory structure
3550 * better "clean.sh" behavior
3551 * added doc/FOR-CMUCL-DEVELOPERS
3552 * many many small tweaks to output format, e.g. removing possibly-confusing
3553 trailing #\. character in DESCRIBE-INSTANCE
3554 * (EQUALP #\A 'A) no longer signals an error.
3555 * new hashing code, including EQUALP hashing
3556 * tidied up Lisp initialization and toplevel
3557 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3558 * command line argument processing
3559 * added POSIX-GETENV function to deal with Unix-ish environment variables
3560 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3561 terminating SBCL on EOF
3562 * non-verbose GC by default
3563 * There is no more "sbcl" shell script; the sbcl file is now the C
3564 runtime executable (just like CMU CL).
3565 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3567 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3568 symbol and package information as Lisp native symbol and package objects
3569 instead of strings naming symbols and strings naming packages. This way,
3570 whenever packages are renamed (as in warm init), debug information is
3571 transformed along with everything else.
3572 * tweaked the optimization policy declarations which control the building
3573 of SBCL itself. Now, among other things, the system no longer saves
3574 source location debugging information. (This helps two problems at once
3575 by reducing SBCL size and by keeping SBCL from trying to look for its
3576 sources -- which may not exist -- when reporting errors.)
3577 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3578 debugging and testing purposes
3579 * cleaned up printing, making the printer call PRINT-OBJECT for
3580 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3581 methods, giving nearly-ANSI behavior
3582 * converted almost all special variables to use *FOO* naming convention
3583 * deleted PARSE-TIME functionality, since it can be done portably
3584 * moved some files out of cold init into warm init
3585 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3587 * regularized formatting of source files
3588 * added an install.sh script
3589 * fixed ridiculous memory usage of cross-compiler by making
3590 compiler/alloc.lisp not try to do pooling unless it can hook
3591 itself into the GC of the cross-compilation host. Now the system
3592 builds nicely on my old laptop.
3593 * added :SB-ALLOC in target-features.lisp-expr
3594 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3595 was not implemented)
3596 * re-did condition handling and note reporting in the compiler. Notes
3597 are no longer handled by signalling conditions. Style warnings
3598 and warnings are handled more correctly and reported in such a way
3599 that it's easy to find one or the other in your output (so that you
3600 can e.g. figure out which of many problems caused COMPILE-FILE to
3602 * changed the severity of several compiler warnings from full WARNING
3603 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3604 compiler note reporting so that it doesn't use the condition system
3605 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3606 COMPILE-FILE command)
3607 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3608 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3609 no longer supported, since their implementation was deeply intertwingled
3610 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3611 * removed bogus "support" for compiler macros named (SETF FOO), and
3612 removed the compiler macro for SETF INFO (but only after making a fool
3613 of myself on the cmucl-imp mailing list by posting a bogus patch for
3614 DEFINE-COMPILER-MACRO..)
3615 * Compiled files containing forms which have side effects on the Lisp
3616 reader (such as DEFPACKAGE forms) are now handled more correctly.
3617 (Compiler queuing of top level lambdas has been suppressed by setting
3618 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3619 * deleted various currently-unused source files, e.g. gengc.lisp. They
3620 may be added back at some point e.g. when porting to other architectures,
3621 but until they are it's distracting to distribute them and to try to
3623 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3624 there were so many of them they're just distractions, and UNCROSS is
3625 known to be able to handle the current sources
3626 * moved PROFILE functionality into TRACE, so that it will be clear
3627 how the wrapping and unwrapping of functions when you profile them
3628 interacts with the wrapping and unwrapping of functions when you
3629 trace them. (Actually, the functionality isn't there yet, but at least
3630 the interface specification is there. Hopefully, the functionality will
3631 arrive with some future maintenance release.)
3632 * removed host-oops.lisp
3633 * changed signature of QUIT function to allow UNIX-CODE argument
3634 * fixed READ-SEQUENCE bug
3635 * tweaked verbose GC output so that it looks more like the progress
3636 output that ANSI specifies for functions like LOAD
3637 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3638 * added <http://sbcl.sourceforge.com> to the banner information printed by