1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
3 * bug fix: add a workaround to SBCL looping infinitely at startup on
4 Linux kernels with apparently buggy implementations of personality().
5 (thanks to Svein Ove Aas)
6 * bug fix: Unicode symbols are correctly printed in LDB backtraces
7 (thanks to David Lichteblau)
8 * bug fix: local bindings shadow global symbol macros for
9 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
10 * optimization: non-open coded uses of numeric comparison operators
11 (e.g. >) no longer cons when called with more than one parameter
12 on platforms supporting dynamic-extent allocation.
13 * enhancement: saving cores with foreign code loaded is now supported
14 on MIPS/Linux in addition to the previously supported platforms.
15 * bug fix: COMPUTE-RESTARTS returns all active restarts, including those
16 with :TEST, when called without a condition. (reported by Helmut Eller for
18 * bug fix: division by zero in sb-sprof when no samples were collected
19 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
20 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
21 (thanks to Alexander Kjeldaas)
22 * new feature: ignoring the return values of destructive functions
23 like NREVERSE signals a compile-time style-warning.
24 (thanks to Kevin Reid)
25 * new feature: the alignment of alien structure fields can be
26 explicitly specified. (thanks to Cyrus Harmon)
27 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
28 and dump core on SIGQUIT
30 ** incompatible change: threads do not inherit values of specials
31 from their parents (see manual)
32 ** bug fix: threads stacks belonging to dead threads are freed by the
33 next exiting thread, no need to gc to collect thread stacks anymore
34 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
35 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
36 ** bug fix: dynamic variable and thread start related gc lossage
37 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
39 * fixed some bugs revealed by Paul Dietz' test suite:
40 ** SUBTYPEP is slightly more accurate on heinously complicated
41 CONS types where some of the members have uncertain (in the
42 NIL, NIL sense) type relationships to each other.
44 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
45 no longer purified unless :PURIFY T is explicitly specified.
46 ** Non-purified cores are significantly smaller than before
48 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
49 * new feature: timers based on Zach Beane's excellent timer package
50 * added support for the following external formats: koi8-u,
51 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
52 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
53 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
54 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
55 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
56 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
57 * incompatible change: a threaded SBCL will no longer revert to
58 non-threaded mode on non-NPTL systems, but refuse to start entirely.
59 * bug fix: interrupts are disabled until startup is complete; no
60 more sigsegvs when receiving a signal to soon
61 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
63 * bug fix: add a workaround for the memory randomization features in
64 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
65 workaround will only be in effect on systems with the proc filesystem
67 * bug fix: printing objects of type HASH-TABLE signals a
68 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
70 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
71 have been running for over 50 days. (reported by Gilbert Baumann)
72 * bug fix: the logic for getting names of functions gets less
73 confused when confronded with alternate-metaclass
74 funcallable-instances. (reported by Cyrus Harmon)
75 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
76 consistent, even on internal alternate-metaclass objects.
77 * bug fix: flush closure information collected by physical
78 environment analysis prepass before the main pass. (bug reported
80 * bug fix: compiler pack phase does not modify a hashtable, which is
81 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
82 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
83 may be expanded. (reported by Alexander <malishev> on c.l.l)
84 * bug fix: no more style-warnings when compiling code using LAST
85 under high-SPEED low-SPACE settings. (reported by David Wallin
87 * fixed some incompatibilities between SBCL's MOP and the MOP
89 ** the METAOBJECT class is now implemented;
90 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
91 STANDARD-OBJECT, as required;
92 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
93 now compatible (as required by VALIDATE-SUPERCLASS); there
94 remains a consistency requirement over the presence or absence
95 of the FUNCTION class in the superclass list of finalized
96 classes; see the manual for more details;
97 ** the :AROUND method for COMPUTE-SLOTS on
98 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
99 requested slot ordering.
101 ** bug fix: parent thread now can be gc'ed even with a live
103 ** bug fix: binding a special with PROGV to no value is not
104 the same as that symbol not having been bound (thanks to
106 ** bug fix: binding specials is thread safe (thanks to
108 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
109 and SB-SPROF do not die with 'no handler for signal XX in
110 interrupt_handle_now(..)' anymore
111 ** bug fix: WITH-TIMEOUT works with multiple running threads
112 * fixed some bugs revealed by Paul Dietz' test suite:
113 ** ENSURE-GENERIC-FUNCTION should take a method class object for
114 the :method-class keyword argument.
116 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
117 * new port: the Solaris operating system on x86 processors is now
118 mostly supported, though some rough edges in the environment
119 remain. (thanks to Daisuke Homma)
120 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
121 space for its heap. (thanks to Thiemo Seufer)
122 * optimization: SBCL on MIPS platforms now supports dynamic-extent
123 closures. (thanks to Thiemo Seufer)
124 * minor incompatible change: eof selects abort in the debugger.
125 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
126 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
128 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
129 deprecated, and will go away in a future revision of SBCL.
130 * minor incompatible change: GC-ON and GC-OFF are no longer
131 implemented with a counter, it does not matter how many times gc
132 is switched on or off
133 * bug fix: discriminating functions for generic function classes
134 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
135 make invalid assumptions about method precedence order. (reported
137 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
138 platforms. (thanks to Thiemo Seufer)
139 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
140 * bug fix: correct pathname printing with printer escaping is on.
141 (thanks to Kevin Reid)
142 * bug fix: complex VOP definitions in "user-space" no longer trigger
143 package locks. (reported by Zach Beane)
144 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
145 for funcallable-instance functions, and is no different from
147 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
148 characters in the current readtable. (reported by Nicholas Neuss)
149 * bug fix: SUBTYPEP on various CONS types returns more a more
150 accurate acknowledgment of its certainty.
151 * optimizations: REMOVE-DUPLICATES now runs in linear time on
152 lists in some cases. This partially fixes bug 384.
153 * flush all standard streams before prompting in the REPL and the
155 * bug fix: signal handling and triggering gc do not conflict
156 directly anymore, in particular a high frequency sb-sprof does
157 not prevent gc from running
158 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
159 approximation for timezone and DST information between the
160 universal time epoch and the smallest negative 32-bit time_t.
161 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
162 year 1899 so long as the encoded time is non-negative.
163 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
164 the correct answer for non-constant rotate amounts when performing
165 an inline 32-bit rotation.
167 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
168 there is only one thread in the session
169 ** bug fix: memory leak for streams created in one thread and
170 written to in another
171 ** bug fix: lockup when compiled with gcc4
172 ** bug fix: race that allows the gc to be triggered when gc is
174 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
175 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
177 ** bug fix: finished threads are gc'ed properly
178 * fixed some bugs revealed by Paul Dietz' test suite:
179 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
180 has arguments with a different set of applicable methods from
181 the orignal arguments.
182 ** The type error thrown by MAP now has a correct expected-type
184 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
185 name a compiled function.
186 ** Three MISC tests where a large bignum was improperly coerced to
187 a float that couldn't represent that bignum during type
188 derivation were fixed.
189 ** SUBTYPEP can now handle types involving the negation of a
190 list-form FUNCTION type.
191 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
192 as (COMPLEX (AND RATIO (NOT FIXNUM))).
193 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
195 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
196 * New feature: Experimental support for bivalent streams: streams
197 opened with :element-type :default now allow character and binary
198 (unsigned-byte 8) I/O
199 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
200 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
202 * bug fix: no more highly sporadic "couldn't check whether ~S is
203 readable" when reading a stream and an interrupt hits in the middle
204 of a select system call
205 * compiler better recognizes complex arrays (reported by Eduardo
207 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
208 for variables on SBCL built with linkage-tables. (reported by Luis
210 * various error reporting improvements.
211 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
212 (Thanks to Thiemo Seufer)
213 * bug fix: nested reader invokations work correctly
214 * bug fix: it is possible to have simultaneous references to foreign
215 code and foreign data with the same name.
217 ** added x86-64 support
218 ** incompatible change: the threading api now works with thread
219 objects instead of thread ids
220 ** bug fix: threads are protected from signals and interruption when
221 starting up or going down
222 ** bug fix: a race where an exiting thread could lose its stack to gc
223 ** bug fix: don't halt on infinite error in threads if possible
224 ** fixed numerous gc deadlocks introduced in the pthread merge
225 ** bug fix: fixed thread safety issues in read and print
226 ** bug fix: debugger doesn't hang on session lock if interrupted at
227 an inappropriate moment
228 ** bug fix: run-program is now thread safe(r)
229 ** bug fix: inner with-recursive-lock no longer releases the mutex
230 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
231 * fixed some bugs revealed by Paul Dietz' test suite:
232 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
233 EXPECTED-TYPE slots filled.
234 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
236 ** MAKE-CONDITION accepts classes as type-designators.
237 ** COMPILE may never return NIL.
238 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
239 range before calling Unix time functions
241 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
242 * numerous signal handling fixes to increase stability
243 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
244 * minor incompatible change: we now correctly canonize default
245 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
246 as per the MOP, rather than the historical (INITARG INITFUNCTION
247 INITFORM). (reported by Bruno Haible)
248 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
249 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
250 for more information.
251 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
252 pathname is a directory pathname.
253 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
254 * dynamic space size on PPC has been increased to 768Mb. (thanks to
256 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
257 :METACLASS argument in addition to a class name. (reported by
258 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
259 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
260 (reported by Bruno Haible; patch from Gerd Moellmann)
261 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
263 * bug fix: more cleanups to the floating point exception handling on
264 x86-64 (thanks to James Knight)
265 * bug fix: the compiler does not try to constant fold calls of
266 COERCE and BIT-* functions when they return freshly constructed
267 objects. (reported by Nikodemus Siivola and Paul Dietz)
268 * optimization: improved the MIPS versions of generic (in the
269 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
270 * optimization: direct conversion of (unsigned-byte 32) to floats on
271 the PowerPC platform.
272 * optimization: structure instances with raw slots now use less
273 memory, and probably show better memory locality. (thanks to
275 * optimization: DYNAMIC-EXTENT declarations for lists and closures
276 are treated as requests for stack allocation on the x86-64,
277 Alpha-32, PPC and SPARC platforms.
278 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
279 the wrong file descriptor; implementation of SOCKET-OPEN-P.
280 (thanks to Tony Martinez)
282 ** gcing a dead thread can no longer lead to lockups
283 ** threads block signals until they are set up properly
284 ** errno is no longer shared by threads
285 ** interrupt-thread restores the eflags register on x86
286 ** fixed some lockups due to gc/thread interaction
287 * fixed some bugs revealed by Paul Dietz' test suite:
288 ** invalid dotted lists no longer raise a read error when
290 ** signal an error if a symbol that names a declaration is used
291 as the name of a type, or vice versa
292 ** allow using the (declare (typespec var*)) abbreviation for
293 (declare (type typespec var*)) with all type specifiers
294 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
295 ** always return NIL from PPRINT-POP when OBJECT is NIL
296 ** don't signal errors when pretty-printing malformed LABELS,
297 FLET or MACROLET forms
298 ** declarations in a DOLIST body are also in scope for the
300 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
302 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
305 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
306 * fixed cross-compiler leakages that prevented building a 32-bit
307 target with a 64-bit host compiler.
308 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
309 opened with :IF-EXISTS :OVERWRITE.
310 * fixed bug 281: error for an invalid qualifier in a short-form method
311 combination method is not signalled until the faulty method is called.
312 * bug fix: iteration variable type inferrer failed to deal with open
313 intervals. (reported by Alan Shields)
314 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
315 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
316 or more runtime options were provided to the sbcl binary.
317 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
318 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
319 * optimization: there's now a fast-path for fixnum arguments in the
320 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
321 * optimization: the code generated on x86-64 is more compact thanks
322 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
323 * bug fix: floating-point exception handling now partly works on
324 x86-64. (Thanks to James Knight)
325 * improvement to the MIPS backend from Thiemo Seufer: C-style
326 64-bit long long arguments and return values to alien functions
328 * fixed some bugs revealed by Paul Dietz' test suite:
329 ** the type-error signalled from WARN has a filled-in DATUM slot.
330 ** the type-error required when a stream is not associated with
331 a file has the stream as its datum.
332 ** type-errors on single-floats on x86-64 no longer have
333 :INVALID-OBJECT as the datum
334 ** the type-errors signalled for invalid function names now have
335 a correct expected type
336 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
337 for structures defined with a :TYPE. Documentation strings for
338 typed structures are no longer immediately discarded
339 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
340 broadcast streams and synonym streams. FILE-LENGTH now also works
341 on broadcast streams.
343 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
344 * incompatible change: the --noprogrammer option, deprecated since
345 version 0.7.5, has been removed. Please use the equivalent
346 --disable-debugger option instead.
347 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
349 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
350 inoperational for a while now) has been completely removed.
351 * null lexical environments are now printed as #<NULL-LEXENV>,
352 significantly reducing the amount of clutter in typical backtraces.
353 * documentation on weak pointers, finalization, and after GC hooks
354 has been added to the manual.
355 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
356 as other specialized array types, is much faster. SUBSEQ and
357 COPY-SEQ on such arrays have also been sped up.
358 * optimization: EQL is now more efficient when at least other argument
359 is known to be of type (OR FIXNUM (NOT NUMBER)).
360 * fixed bug: the runtime is now less vulnerable to changes in the
361 size of the SBCL object on OS X, and virtual memory is reserved for
362 all spaces that need to be at a fixed address.
363 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
364 * fixed bug: finalizers and after GC hooks that cause consing are now
366 * fixed bug: compiler error messages and summaries are now printed to
367 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
368 * fixed inference of the upper bound of an iteration variable.
369 (reported by Rajat Datta).
370 * fixed bug 211e: calling local functions with duplicated constant
371 keyword argument no longer causes a bogus style warning about an
373 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
374 ftype declaration to be disregarded. (reported by Dave Roberts)
375 * fixed bug 373: caused by erronous compilation of references to alien
376 variables in the runtime on ppc/darwin.
377 * fixed bug 376: CONJUGATE type deriver.
378 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
379 (reported by Baughn on #lisp)
380 * a cross-compiler bug on non-x86 platforms has been identified and
381 fixed. (thanks to Bruno Haible)
382 * improvements to the MIPS runtime code for increased stability.
383 (thanks to Thiemo Seufer)
384 * increased the maximimum compact environment size to allow
385 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
386 * partial workaround for bug 135: don't name the function we're
387 calling for hairy cases of EVAL, so as not to accumulate one environment
388 entry per call to EVAL. (reported by Kevin Reid)
389 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
390 * optimization: type testing for non-vector arrays should be faster.
391 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
393 * bug fix: setting 31st element of a bit vector to zero did not work
395 * bug fix: redefining a class definition which failed due to a
396 previous accessor / function clash now works (but see BUGS entry
397 #380 for more problems in this area). (thanks to Zach Beane)
398 * the long form of DEFINE-METHOD-COMBINATION disables method group
399 checking when given a single method group with pattern *. (thanks
401 * on x86 compiler supports stack allocation of results of simple
402 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
403 * support for building and running on Mac OS X 10.4 "Tiger" has been added
404 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
405 * fixed some bugs related to Unicode integration:
406 ** the restarts for recovering from input and output encoding
407 errors only appear when there is in fact such an error to
409 * fixed some bugs revealed by Paul Dietz' test suite:
410 ** MISC.549 and similar: late transformation of unsafe type
411 assertions into derived types caused unexpected code
413 ** SCALE-FLOAT type deriver is less wrong.
414 ** type derivers for EXP, LOG and similar functions compute result
415 types for complex arguments better.
416 ** (MISC.563) CONJUGATE type deriver works for very restricted
418 ** out-of-line type testers for character strings are available.
419 ** EQUAL compiler transform understands specialness of objects
421 ** accessing double-floats stored on the stack now works on x86-64.
422 ** debugger internals could sometimes create invalid lispobjs,
423 resulting in GC crashes.
424 ** MISC.548: type check weakening can convert required type into
426 ** initialization forms for bindings are not in scope of free special
429 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
430 * incompatible change: thread support for non-NPTL systems has
431 been removed - locking is buggy and unreliable. A threaded
432 SBCL build will now warn at startup and refuse to create
433 new threads, unless futex support is detected in the kernel
434 * incompatible change: the top level REPL now has only an ABORT
435 restart associated with it, not TOPLEVEL and ABORT as it used to.
436 TOP and TOPLEVEL are now available as debugger commands for
437 returning to the top level.
438 * incompatible change: forms evaluated in the REPL now use the
439 global optimization policy.
440 * incompatible change: user- and system-initialization files are
441 no longer processed with LOAD, but by READ and EVAL; hence the
442 global optimization policy, startup package, readtable, etc,
444 * The .fasl file format number has been incremented because of
445 various incompatible changes.
446 * internal entry point details and argument counts no longer appear
447 in backtraces unless explicitly requested by setting
448 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
449 * built-in and standard functions no longer have names like "top
450 level local call to FOO".
451 * fixed bug 32: functions defined in non-null lexical environments
452 now have more legible printed representation
453 * fixed bug 33: functions defined in non-null lexical environemnts
454 are now more amenable to inspection by INSPECT.
455 * workaround for bug 354: XEPs no longer appear in backtraces unless
456 explicitly requested.
457 * fixed bug: receiving the signal which results from attempting to
458 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
459 SIGBUS on other *BSDs) on architectures where the C stack is also
460 the Lisp stack (x86 and x86-64) from foreign code no longer leads
461 to debugger confusion or wild instability. (reported by Cheuksan
463 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
464 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
465 (reported by Lutz Euler)
466 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
467 compiled code. (reported by Paul Dietz)
468 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
469 specialized parameter to a method to something that is not TYPEP
470 the specializer is now possible.
471 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
472 face of package deletion.
473 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
474 pathname. (reported by David Tolpin on c.l.l)
475 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
476 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
477 STRING-STREAM and FILE-STREAM, respectively.
478 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
479 than previously; in addition, its test suite is now run on build.
480 (thanks to Luke Gorrie)
481 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
482 * optimization: added a immediate representation for single-floats
484 * fixed some bugs revealed by Paul Dietz' test suite:
485 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
486 ** attempting to create a package with a colliding nickname causes
487 correctable errors to be signalled.
488 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
489 ** several x86-64 backend bugs related to sign-extension of immediate
492 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
493 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
494 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
496 * fixed loading of multiply forward-referenced layouts.
497 (thanks to Cheuksan Wang)
498 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
499 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
500 (see COMPILE-FILE documentation for details of :PRINT option)
501 * fixed bugs 19 and 317: fixed-format floating point printing is
502 more accurate. This also fixes a bug reported by Adam Warner
503 related to the ~@F format directive.
504 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
506 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
507 dispatch table if the from-char is a dispatch macro character.
508 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
509 by the wordsize no longer ignore the last word. (reported by Lutz
511 * fixed bug in type checking of dynamic-extent variables. (reported
513 * optimization: sequence traversal functions use their freedom to
514 coerce function designators to functions.
515 * optimization: code with many calls to CLOS methods specialized on
516 CLOS classes has had redundant type checks removed.
517 * optimization: added declarations to speed up operations that access
518 the internal character database (for example STRING-UPCASE)
519 * optimization: comparison operations between floats and sufficiently small
520 fixnums no longer create extra rationals
521 * fixed some bugs related to Unicode integration:
522 ** portions of multibyte characters at the end of buffers for
523 character-based file input are correctly transferred to the
524 start of the buffer at the next read.
525 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
526 passing it through to OPEN.
527 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
528 argument given to internal calls to OPEN.
529 ** the built-in comment readers (introduced by character sequences
530 ";" and "#|") are more forgiving to encoding errors; they will
531 STYLE-WARN and then attempt to resync the stream at a character
532 boundary. (thanks to Teemu Kalvas)
533 * fixed some bugs in the x86-64 port:
534 ** Negative short int return values from c-calls are sign-extended
536 ** The stack is aligned to 16-bytes for c-calls, as required by
537 the ABI. (thanks to Cheuksan Wang)
538 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
539 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
540 for enums. (thanks to Vincent Arkesteijn)
541 ** Multiple small optimizations and bugfixes for floating point
543 * fixed some bugs revealed by Paul Dietz' test suite:
544 ** Space, Tab, Linefeed, Return and Page have the invalid
545 secondary constituent character trait.
546 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
548 ** WITH-INPUT-FROM-STRING should only update the index place on
550 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
552 ** Bit-array operations (BIT-AND and similar) worked incorrectly
553 with one-dimensional arrays with fill pointers.
554 ** TYPE-OF failed on a complex with an integer realpart and a
556 ** compiler failure during type inference for the code of form
557 (IF (EQL X (THE ...)) ...) (MISC.535).
559 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
560 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
561 platform. The port supports 61-bit fixnums, large memory spaces
562 and reloading shared object files.
563 * enhancement: saving cores with foreign code loaded is now
564 supported on ppc/Darwin in addition to the previously supported
566 * enhancement: the statistical profiler now walks deeper into the
567 call stack for more meaningful call-graphs and accrued time
568 reports (x86/x86-64 only). It also now reports time spent in
570 * enhancement: it is now possible to trace most individual methods
571 of a generic function in addition to tracing the generic function
573 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
575 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
576 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
577 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
578 produces an error. (thanks to Vincent Arkesteijn)
579 * bug fix: NAMESTRING on pathnames with :WILD components in their
580 directories works correctly. (thanks to Artem V. Andreev)
581 * fixed bug 125: compiler preserves identity of closures. (reported
583 * bug fixed: functions with &REST arguments sometimes failed with
584 "Undefined function" when compiled with (DEBUG 3). (reported by
586 * bug fix: overflow during compiling of setting element of a bit
587 vector with constant index and value. (reported by Timmy Douglas)
588 * build fix: fixed the dependence on *LOAD-PATHNAME* and
589 *COMPILE-FILE-PATHNAME* being absolute pathnames.
590 * on x86 compiler partially supports stack allocation of dynamic-extent
592 * GO and RETURN-FROM do not check the extent of their exit points
593 when compiled with SAFETY 0.
594 * fixed some bugs related to Unicode integration:
595 ** encoding and decoding errors are now much more robustly
596 handled; it should now be possible to recover even from invalid
597 input or output to the terminal. (thanks to Teemu Kalvas)
598 ** provided a first cut at implementing STRING-TO-OCTETS and
599 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
600 ** altered the SB-MD5 contributed module slightly, changing the
601 interface just enough for it to be supportable for builds where
602 lisp characters are not eight bits.
603 * fixed some bugs revealed by Paul Dietz' test suite:
604 ** the FORMATTER-generated functions for ~V[ conditionals require
605 the correct number of arguments.
606 ** READ-FROM-STRING returns the mandated second value when applied
607 to displaced strings.
608 ** the #\Rubout and #\Backspace characters are treated as invalid
609 constituent characters by the tokenizer.
611 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
612 * new feature: reloading changed shared object files with
613 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
614 * new feature: references to foreign variables and functions
615 can now be compiled and loaded before the corresponding shared
616 object file is loaded, as long as the foreign definitions are
617 available at runtime.
618 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
619 supported. (thanks to Dan Debertin)
620 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
621 just keywords. (thanks to Vincent Arkesteijn)
622 * fixed bug #331: structure-class instances corresponding to
623 DEFSTRUCT forms are now created eagerly.
624 * fixed bug #345: backtraces from calls to undefined functions work
625 on x86 as well. Related bug #61 is now also partially fixed on x86
626 (backtraces from throws to unknown catch tags.)
627 * bug fix: lambda-list parsing is now stricter vrt. order and number
628 of lambda-list keywords.
629 * bug fix: as specified by AMOP, an error is signalled if a
630 class-option appears multiple times in a DEFCLASS form. (reported
632 * bug fix: RANDOM can be compiled when the compiler derives the type
633 of its numeric argument as a disjoint set of small integers.
634 (reported by Paul Dietz)
635 * bug fix: starting a core saved with shared objects loaded when
636 those objects are not available no longer causes threaded SBCL to
637 hang. (reported by Sean Ross)
638 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
640 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
641 argument, which must always be NIL. (reported by Kalle Niemitalo)
642 * bug fix: printing 1.0d+23 no longer results in an error.
643 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
644 * bug fix: structure slot setters preserve evaluation order. (thanks
646 * fixed some bugs related to Unicode integration:
647 ** RUN-PROGRAM can allow its child to take input from a Lisp
648 stream. (reported by Stefan Scholl)
649 ** ASDF-INSTALL successfully downloads .tar.gz files in all
650 locales. (reported by Ken Causey)
651 * fixed some bugs revealed by Paul Dietz' test suite:
652 ** INCF, DECF and REMF evaluate their place form as specified in
654 ** FORMATTER expands ~{ ~} iteration directives with V or #
655 parameters correctly.
656 ** FORMATTER deals with the ~@[ ~] conditional directive where the
657 consequent uses no arguments correctly.
658 ** the system has a partial understanding of the (COMPLEX RATIO)
660 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
663 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
664 * new feature: a build-time option (controlled by the :SB-UNICODE
665 keyword feature, enabled by default) for building the system with
666 support for the entire 21-bit character space defined by the
668 * new feature: the system now has rudimentary external-format
669 support; the primary user-visible change at this time is that
670 characters with the high bit set (such as lower-case-e-acute) will
671 print correctly to a terminal in a UTF-8 environment.
672 * minor incompatible change: BASE-CHAR no longer names a class;
673 however, CHARACTER continues to do so, as required by ANSI.
674 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
675 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
676 has been moved to the SB-EXT package (temporarily re-exported from
678 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
679 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
681 * fixed bug #351: better error-handlind and reporting for malformed
683 * fixed bug #350: bignum-printing is now more memory-efficient,
684 allowing printing of very large bignums, eg. (expt 2 10000000).
685 (reported by Bruno Haible)
686 * fixed bug #302: better primitive-type selection for intersection
688 * fixed bug #308: non-graphic characters now all have names, as
689 required. (reported by Bruno Haible)
690 * bug fix: redefining a class with different superclasses now correctly
691 removes it from the direct-subclasses of its previous superclasses.
692 (reported by David Morse)
693 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
694 new value now works. (reported by Bruno Haible)
695 * bug fix: correct canonicalization of multiple non-standard slot
696 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
697 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
698 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
699 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
700 now exists, an signals an error.
701 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
702 during macroexpansion for non-iteration variables as well. (reported
703 by Bruno Haible for CMUCL)
704 * bug fix: Cyclic structures and unprintable objects in compiler
705 messages no longer cause errors. (reported by Bruno Haible)
706 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
707 and READ-FROM-STRING all now return a primary value of NIL if
708 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
709 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
710 (reported by Bruno Haible for CMUCL)
711 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
712 do not designate a condition. (reported by Bruno Haible for
714 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
715 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
716 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
717 permitting resolution in favour of any of the conflicting symbols.
718 (reported by Bruno Haible for CMUCL)
719 * bug fix: EQUAL compiler optimizations is less aggressive on
720 strings which can potentially compare true despite having distinct
721 specialized array element types.
722 * bug fix: unit enumerations can be defined without dividing by
723 zero. (thanks to Vincent Arkesteijn)
724 * FORMAT compile-time argument count checking has been enhanced.
725 (report from Bruno Haible for CMUCL)
726 * a partial workaround for the bug 262: the compiler does not try to
727 inline-expand a local function doing RETURN-FROM from a deleted
728 BLOCK. (thanks to Peter Denno for the bug report and to David
729 Wragg for the simple test case)
730 * fixed some bugs revealed by Paul Dietz' test suite:
731 ** RENAME-PACKAGE allows all package designators as new package
733 ** constraint propagation and lambda variable substitution are
734 more cautious in dealing with partially deleted code.
735 ** compiler sometimes lost reoptimization passes.
736 ** CERROR, given a condition as condition designator, passes its
737 remaining arguments to the continue format control without
739 ** Case-altering FORMAT directives work correctly on non-ASCII
741 ** The REAL type specifier handles bounds outside the floating
742 point ranges without signalling FLOATING-POINT-OVERFLOW.
743 ** Functions with IR1-transformations can create intercomponent
744 references to global functions.
745 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
747 ** FORMAT ~R treats a nil value for its first parameter correctly.
749 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
750 * enhancement: saving cores with foreign code loaded is now
751 supported on x86/NetBSD and sparc/Linux in addition to the previously
753 * bug fix: on some platforms repeated installations caused multiple
754 copies of HTML documentation to be installed -- should not happen
755 any more. (reported by Stefan Scholl)
756 * bug fix: parsing self-recursive alien record types multiple times
757 no longer causes infinite recursion. (reported by Thomas F. Burdick,
758 original patch by Helmut Eller for CMUCL)
759 * bug fix: stack-exhaustion detection works now on NetBSD as well.
760 (thanks to Richard Kreuter)
761 * bug fix: defining classes whose accessors are methods on existing
762 generic functions in other (locked) packages no longer signals
763 bogus package lock violations. (reported by François-René Rideau)
764 * bug fix: special variables as DEFMETHOD parameters no longer have
765 associated bogus type declarations. (reported by David Wragg and
767 * bug fix: read-write consistency on streams of element-type
768 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
769 * bug fix: redefiniton of the only method of a generic function with
770 no DEFGENERIC no longer emits a full WARNING. In addition,
771 redefinition of generic functions with no DEFGENERIC to an
772 incompatible lambda list now signals an error. (thanks to Zach
774 * bug fix: DEFGENERIC now works even when there's a function of the
775 same name in an enclosing lexical environment. (thanks to Zach
777 * fixed compiler failure, caused by instrumenting code during
778 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
779 * optimization: added loop analysis and improved register allocation
780 * fixed some bugs revealed by Paul Dietz' test suite:
781 ** POSITION on displaced vectors with non-zero displacement
782 returns the right answer.
783 ** (SIMPLE-STRING) is a valid type specifier for sequence
785 ** *PRINT-LEVEL* handling for slotless structures is pedantically
787 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
788 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
790 ** The justification version of the FORMAT ~< directive treats
791 non-zero minpad parameter correctly.
793 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
794 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
795 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
796 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
797 the supported interface.
798 * new feature: Single-stepping of code compiled with DEBUG 2 or
799 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
800 * new feature: saving cores with foreign code loaded is now
801 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
802 Timothy Moore's work for CMUCL)
803 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
804 parameters to * instead of NIL if no initform is supplied.
805 (reported by Johan Bockgaard)
806 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
807 correctly when FUNCALL appears as the car of the form. Note:
808 despite this FUNCALL forms are not currently subject to
809 compiler-macro expansion. (port of Raymond Toy's fix for the
810 same from CMUCL, reported by Johan Bockgaard)
811 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
812 (thanks for Teemu Kalvas)
813 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
814 case instead of terminating the iteration (thanks for Julian
815 Squires, Sean Champ and Raymond Toy)
816 * bug fix: incorrect expansion of defgeneric that caused a style
817 warning. (thanks for Zach Beane)
818 * bug fix: slot accessor effective method computation works properly
819 for classes with multiple non-standard applicable methods on
820 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
821 * on x86 compiler supports stack allocation of results of LIST and
822 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
823 CMUCL implementation by Gerd Moellmann)
824 * fixed some bugs revealed by Paul Dietz' test suite:
825 ** FORMAT strings with both the ~<~:;~> form of the justification
826 directive and pretty-printing directives cause an error.
828 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
829 * incompatible change: the internal functions
830 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
831 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
832 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
833 instead of the old functions.
834 * new feature: on platforms where "dladdr" is available foreign
835 function names now appear in backtraces. (based on Helmut Eller's
837 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
838 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
840 * documentation: ASDF manual is now distributed with SBCL.
841 * bug fix: SBCL can now load its contributed modules with REQUIRE
842 even if the system-provided entries have been removed from
843 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
845 * bug fix: backtraces involving undefined functions or assembly
846 routines are more informative. (thanks to Brian Downing)
847 * bug fix: mutually referent alien structures now work correctly.
848 (reported by Rick Taube)
849 * bug fix: structures defined by WITH-ALIEN can be referred to
850 within other definitions in the same WITH-ALIEN.
851 * bug fix: division operators (MOD, TRUNCATE and the like) with
852 constant zero divisors and integer dividends no longer generate
854 * bug fix: provide default methods for INPUT-STREAM-P and
855 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
856 * bug fix: improve the detection and resolution of MOP metacycles
857 (where effective method computation depends on the generic function
858 having its effective method computed). (reported by Bruno Haible)
859 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
860 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
861 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
862 * bug fix: pathnames with relative directory components can be
863 represented relative to default pathnames.
864 * optimization: in taking the GCD of bignums, reduce the two bignums
865 to approximately the same size (using Euclid's algorithm) before
866 applying the more sophisticated binary GCD. (thanks to Juho
868 * optimization: COUNT on bitvectors now operates word-at-a-time.
869 * optimization: ASH with a positive, but not necessarily constant,
870 (leftwards) shift, when in a modular context, is compiled to a
872 * fixed some bugs revealed by Paul Dietz' test suite:
873 ** FORMAT variable parameters ("~V<char>") are defaulted properly
874 if the corresponding argument is NIL.
875 ** FORMAT directives accepting a minpad parameter treat negative
877 ** literal commas as character parameters to FORMAT directives are
879 ** literal spaces directly after ~<Newline> directives within a
880 format-logical-block (~:< ~@:>) do not induce :FILL-style
881 conditional newlines.
882 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
883 as specified: it no longer includes conditional newlines).
884 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
886 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
887 :PER-LINE-PREFIX argument does not evaluate to a string.
889 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
890 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
891 the manual for details; for now, package locks can be disabled by
892 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
893 if no major problems are found then it is likely that they will be
894 compiled in unconditionally.
895 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
896 unsupported operators on all platforms. To load a shared library
897 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
898 object file, link it into a shared library outside of SBCL and
899 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
901 * bug workaround: ROOM T and the GC were not working together
902 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
903 reporting facility aren't preserved. That reporting has been
904 disabled, so now until and unless someone figures out how to make it
905 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
906 * minor incompatible change: as threatened around sbcl-0.8.0, the
907 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
908 The symbols are also exported from SB-PCL for backwards
909 compatibility, but more so than before SB-PCL should be treated as
910 an implementation-internal package.
911 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
913 * the SB-SPROF contrib now works on (most) non-x86 architectures.
914 It is known as of this release not to work on the Alpha, however.
915 * fixed bug #167: errors signalled due to illegal syntax in method
916 bodies are now more legible.
917 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
918 designators and can hence be used with TYPEP.
919 * fixed bug #333: CHECK-TYPE now ensures that the type error
920 signalled, if any, has the right object to be accessed by
921 TYPE-ERROR-DATUM. (reported by Tony Martinez)
922 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
923 5.1.2.3 for multiple-value place subforms. (reported by Kalle
925 * fixed bug #334: programmatic addition of slots using specialized
926 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
927 and :ALLOCATION :CLASS slots.
928 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
929 float exponents. (rereported by Peter Seibel)
930 * fixed a bug: #\Space (and other whitespace characters) are no
931 longer considered to be macro characters in standard syntax by
933 * fixed bug: initialization of condition class metaobjects no longer
934 causes an instance of the condition to be created. (reported by
936 * fixed bug: it is now possible to have slots such that
937 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
938 description is neither :INSTANCE nor :CLASS.
939 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
940 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
941 SLOT-VALUE-USING-CLASS), and no longer causes errors with
942 non-standard SLOT-DEFINITION-ALLOCATION values.
943 * fixed bugs: various race conditions handling exiting threads.
944 CL-PPCRE's thread torture test now passes.
945 * fixed arguably-a-bug: GC time proportional to number of threads in
946 system even when most of them are idle
947 * optimization: improved performance of BIT and SBIT on bit-vectors.
948 * .fasl file incompatibility: The fasl file version number has
949 been incremented because of changes associated with package locks.
951 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
952 * minor incompatible change: the system no longer provides
953 optimization hints (conditions which are TYPEP
954 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
955 mode; these hints are still emitted when higher SPEED optimization
957 * new contrib module: a sampling profiler (profiling by statistical
958 sampling, rather than by instrumenting functions) is available as
959 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
960 * the behaviour of the standard function ED is now customizeable by
961 third parties through a hook variable: see ED's documentation
962 string for information on the protocol.
963 * the compiler no longer emits efficiency notes for (FUNCALL X)
964 when the type of X is uncertain under default optimization
966 * fixed bug 276: mutating a binding of a specialized parameter to a
967 method to something that is not TYPEP the specializer is now
969 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
970 exhibit the required behaviour on the x86 platform. (thanks to
971 Peter van Eynde, Eric Marsden and Bruno Haible)
972 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
973 even for difficult arguments. (reported by Peter Graves)
974 * fixed bug 141a: the backquote printer now descends quoted
976 * fixed another bug in backquote printing: no more destructive
977 modification of the form's list structure. (reported by Brian
979 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
980 move between its address being taken and the call to
981 interrupt_thread, fixing a crashing race condition.
982 * the SB-POSIX contrib implementation has been adjusted so that it
983 no longer exhibits ridiculously poor performance when constructing
984 instances corresponding to C structs.
986 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
987 * minor incompatible change: the sb-grovel contrib now treats C
988 structures as alien (in the sense of SB-ALIEN) objects rather than
989 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
990 has implications for memory management of client code
991 (sb-grovel-returned objects must now be manually managed) and for
992 type safety (alien objects now have full types).
993 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
994 used to control emission of compiler diagnostics, rather than the
995 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
996 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
997 quality should be considered deprecated.
998 * install.sh now installs the user manual as well
999 * (not quite a new documentable feature, but worth considering in
1000 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1001 beginnings of a semantically meaningful condition hierarchy is
1002 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1004 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1005 designator as the defaults argument.
1006 * fixed bug: Displaced arrays whose displaced-to array has become
1007 too small now cause ARRAY-DIMENSION to signal an error, providing
1008 for safer bounds-checking. (reported by Bruno Haible)
1009 * fixed bug: DEFCLASS slot definitions with identical :READER and
1010 :WRITER names now signal a reasonable error. (reported by Thomas
1012 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1014 * fixed bug: Invalid slot specification errors now print correctly.
1015 (thanks to Zach Beane)
1016 * fixed bug 320: Shared to local slot value transfers in class
1017 redefinitions now happen corrently from superclasses as
1018 well. (reported by Bruno Haible)
1019 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1021 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1022 improper lists correctly. (reported by Bruno Haible)
1023 * fixed bug 313: source-transform for <fun-name> was erroneously
1024 applied to a call of a value of a variable with name <fun-name>.
1025 (reported by Antonio Menezes Leitao)
1026 * fixed bug 307: The obsolete instance protocol ensures that
1027 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1028 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1029 remove dead unknown-values globs from the middle of the stack.
1030 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1031 easier to resume long computations after using *BREAK-ON-SIGNALS*
1032 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1033 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1034 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1035 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1036 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1037 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1039 * fixed a (fairly theoretical) bug in string printing: if
1040 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1041 does not have array-element-type equal to the most general string
1043 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1044 function name. (thanks to Nikodemus Siivola)
1045 * fixed bug: compile-time format string checker failed on
1046 non-closed ~{. (reported by Thomas F Burdick)
1047 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1048 don't warn on legal loop code involving more than one aggregate
1049 boolean. (thanks to Nikodemus Siivola)
1050 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1051 in keyword position to MAKE-PACKAGE was not being reported as
1053 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1054 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1056 * fixed bugs 280 and 312: the checking for multiple definitions in a
1057 file is less likely to become confused by uses of inline
1059 * fixed bug: the #S reader macro performs the keyword coercion
1060 specified for slot names. (reported by Kalle Niemitalo)
1061 * fixed bug: lambda lists may contain symbols whose names start with
1062 & but are not lambda-list-keywords; their occurrence triggers a
1064 * fixed bug 321: define-method-combination argument lambda lists do
1065 not cause invalid code to be generated when &optional or &aux
1066 variables have default values. (reported by Bruno Haible)
1067 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1068 have CLOS classes; this appears also to have fixed reported
1069 instability in stack exhaustion detection.
1070 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1071 MUFFLE-WARNING when no associated restart is present is now
1073 * optimization: rearranged the expansion of various defining macros
1074 so that each expands into only one top-level form in a
1075 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1077 * optimization: used a previously little-used slot in symbols to
1078 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1080 * fixed some bugs revealed by Paul Dietz' test suite:
1081 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1083 ** Fixed an optimization bug related to inheritance of initforms
1084 from local to shared slots.
1085 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1086 ** CAST optimizer forgot to flush argument derived type.
1087 ** print/read consistency on floats is now orders of magnitude
1088 more likely. (thanks also to Bruno Haible for a similar report
1090 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1091 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1092 for objects of type REAL. Make it so.
1093 ** SXHASH is specified (infelicitously) to respect similarity,
1094 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1095 it so. (thanks to Markus Ziegler)
1096 ** on the Alpha, the compiler succeeds in compiling functions
1097 returning a known number of arguments greater than 63.
1098 ** fixed handling of invalid NIL arguments in keyword position
1100 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1102 ** provided a readably-printable representation for RANDOM-STATE
1104 ** ensured that pathnames, where they have a namestring, always
1105 print using #P"..." syntax.
1107 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1108 * Support for the forthcoming 2.0 version of the NetBSD kernel
1109 running on x86 hardware has been added. (thanks to Perry
1110 E. Metzger most immediately, and others for their past work)
1111 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1112 current release version; SBCL's OpenBSD support had been broken
1113 since about the time of OpenBSD's switch to ELF binary format.)
1114 * [placeholder for DX summary]
1115 ** user code with &REST lists declared dynamic-extent, under high
1116 speed or space and low safety and debug optimization policy.
1117 * The manual has been converted to Texinfo format and the debugger
1118 chapter from the cmucl manual has been added.
1119 * A facility has been added to extract documentation strings from
1120 sbcl and store them as Texinfo-formatted snippets for inclusion in
1121 the manual (via Texinfo's @include directive)
1122 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1123 the test case to Dave Roberts)
1124 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1125 pointers. (reported by Sean Ross)
1126 * bug fix: PROFILE output is printed nicely even for large numerical
1127 values. (thanks to Zach Beane)
1128 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1129 greater than 32 handle EOF correctly.
1130 * bug fix: on X86 an immediate argument of the IMUL instruction is
1131 correctly printed in disassembly. (thanks to Lutz Euler)
1132 * bug fix: class slots in redefined classes preserve their old
1133 values. (thanks to Bruno Haible and Nikodemus Siivola)
1134 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1135 succeeds. (reported by Marco Baringer)
1136 * fixed some bugs revealed by Paul Dietz' test suite:
1137 ** READ-SEQUENCE now works on ECHO-STREAMs.
1138 ** RATIONALIZE works more according to its specification. (thanks
1141 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1142 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1143 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1144 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1145 should matter to you only if you rebind the printer control
1146 variables and then find you want different bindings in the
1147 debugger than in the ordinary execution of your program.)
1148 * The runtime build system has been tweaked to support building
1149 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1150 assembler and linker. (thanks to Nikodemus Siivola)
1151 * Unbound, undefined, undeclared variables now trigger full
1152 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1153 more often programmer error than deliberate exploitation of undefined
1155 * optimization: the hash algorithm for strings has changed to one
1156 that is less vulnerable to spurious collisions. (thanks to Juho
1158 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1159 needless bounds checking. (thanks to Juho Snellman)
1160 * optimization: implemented multiplication as a modular
1161 (UNSIGNED-BYTE 32) operation on the PPC backend.
1162 * fixed some bugs revealed by Paul Dietz' test suite:
1163 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1165 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1166 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1167 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1169 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1171 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1173 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1174 their output stream on EOF from read.
1175 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1176 have been read to end-of-file.
1177 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1179 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1180 description of determination of which consecutive characters
1182 ** Printing the "Space" character with escaping on now yields "#\\ ",
1183 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1184 ** Reading floating-point numbers with *READ-BASE* set to a number
1185 less than 10 works correctly.
1186 ** Reading floating-point numbers with *READ-BASE* set to a number
1187 more than 10 works correctly.
1188 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1189 the readtable currently in effect.
1191 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1192 * minor incompatible change: parsing of namestrings on a physical
1193 (Unix) host has changed; numbers after the final #\. in a
1194 namestring are no longer interpreted as a version field. This is
1195 intented to be largely invisible to the user, except that the
1196 meaning of the namestring "*.*.*" has changed: it now refers to a
1197 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1198 should usually be replaced by
1199 (make-pathname :name :wild :type :wild :version :wild)
1200 with the added benefit that this is more likely to be portable.
1201 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1202 option to OPEN now signals an error if the file being opened
1203 exists; this may have an impact on existing code.
1204 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1205 signals received, on the PowerPC platforms (both Linux and
1206 Darwin). (thanks to Pierre Mai for pointing out the location of
1208 * several fixes on OS X: The system now builds and runs cleanly on
1209 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1210 (thanks to Brian Mastenbrook)
1211 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1212 second-resolution: integer multiples of 1/3600 between -24 and 24.
1213 (thanks to Vincent Arkesteijn)
1214 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1215 type when called with 1 argument; PEEK-CHAR checked type of
1216 PEEK-TYPE only after having read first character from a
1217 stream. (reported by Peter Graves)
1218 * bug fix: the garbage collector now has much better locality
1219 behaviour, and in particular no longer treats all memory as being
1220 exhausted when many small objects point to each other in a deeply
1222 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1223 recognized as being TYPEP their class.
1224 * bug fix: the PUSHNEW documentation string has been corrected.
1225 (thanks to Vincent Arkesteijn)
1226 * bug fix: defaulting of the value for the last of an atypically
1227 large number of multiple values being bound was not being
1228 performed correctly on the Alpha or PPC platforms
1229 * optimization: implemented multiplication as a modular
1230 (UNSIGNED-BYTE 32) operation on the x86 backend.
1231 * optimization: SEARCH on simple-base-strings can now be open-coded.
1232 (see also contrib/compiler-extras.lisp for inspiration for
1233 teaching the compiler about the Boyer-Moore algorithm).
1234 * value, returned by MAX (and MIN) called with several EQUALP, but
1235 not EQL, arguments now does not depend on compiler settings.
1236 * fixed some bugs revealed by Paul Dietz' test suite:
1237 ** in stack analysis liveness information is propagated from
1238 non-local entry points.
1239 ** pathwise CAST removing failed when the CAST node did not start
1241 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1242 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1244 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1245 host is already defined.
1246 ** RENAME-FILE works on streams instead of signalling an internal
1248 ** PEEK-CHAR uses the current readtable when determining whether
1249 or not a character is whitespace.
1250 ** MERGE-PATHNAMES handles the case when the pathname does not
1251 specify a name while the default-pathname specifies a version.
1252 ** Pathnames now stand a better chance of respecting print/read
1254 ** Attempting to use standardized file system operators with a
1255 pathname with invalid :DIRECTORY components signals a
1257 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1259 ** DIRECTORY on logical pathnames is more correct.
1260 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1261 signal a TYPE-ERROR if their argument is not a stream.
1262 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1263 designator argument does not designate a stream.
1264 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1265 examining the synonym.
1266 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1268 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1269 element-type larger than ([UN]SIGNED-BYTE 32).
1271 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1272 * When built with the :SB-FUTEX feature, threaded builds now take
1273 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1274 for faster/more reliable mutex and condition variable support.
1275 * Incompatible change (but one you probably shouldn't have been using
1276 anyway): the interface and code for arbitrating between multiple
1277 threads in the same user session has been redesigned.
1278 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1279 error when called without an explicit environment argument.
1280 (thanks to Nikodemus Siivola)
1281 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1282 confused about their position. (thanks to Adam Warner and Gerd
1284 * bug fix: Pretty printing backquoted forms with unquotations in the
1285 argument list position of various code constructs such as LAMBDA
1286 now works correctly. (reported by Paul Dietz)
1287 * bug fix: Pretty printing unquotations no longer loses all
1288 stream position information.
1289 * optimization: performance of string output streams is now less
1290 poor for multiple small sequence writes.
1291 * optimization: performance of CSUBTYPEP in the presence of complex
1292 expressions involving CONS and NOT many times has been improved.
1293 (reported by Paul Dietz)
1294 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1296 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1297 streams for the REPL, for the debugger, and so on. (thanks to
1299 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1300 * fixed some bugs revealed by Paul Dietz' test suite:
1301 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1302 is ignored if it is too small, rather than propagating through
1303 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1304 ** extremely complex negations of CONS types were not being
1305 sufficiently canonicalized, leading to inconsistencies in
1307 ** VALUES tranformer lost derived type.
1309 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1310 * fixed a bootstrapping bug: the build process no longer assumes
1311 that the various BOOLE-related constants have the same value in
1312 host and target lisps. (noted by Paul Dietz' test suite on an
1313 SBCL binary built from CLISP)
1314 * The system can now be dynamically linked on the MIPS platform,
1315 which enables dynamic loading of foreign code from Lisp. (thanks
1316 to Ralf Baechle for discussions on the MIPS ABI)
1317 * The system now records debugging information for its own source
1318 files in a filesystem-position-independent manner, relative to
1319 the "SYS" logical host.
1320 * fixed a compiler bug: MV-LET convertion did not check references
1321 to the "max args" entry point. (reported by Brian Downing)
1322 * tweaked disassembly notes to be less confident about proclaiming
1323 some instruction as an LRA. (thanks to Brian Downing)
1324 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1325 now each have their own history, command character, and other
1326 characteristics. (thanks to David Lichteblau)
1327 * fixed some bugs revealed by Paul Dietz' test suite:
1328 ** compiler failure in compiling LOGAND expressions including a
1330 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1331 shift greater than 32.
1332 ** FLUSH-DEST did not mark blocks for type check regeneration.
1333 ** HANDLER-CASE failed to accept declarations in handler clauses
1334 in some circumstances.
1336 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1337 * New code in contrib/sb-introspect (still probably not entirely
1338 stable yet) provides some support for smart Lisp development
1339 environments like SLIME.
1340 * The conditions signalled for errors occurring when loading .fasl
1341 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1342 in a way which should help ASDF recover gracefully.
1343 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1344 stuff has been cleaned up. If you code contrib/ stuff, this might
1345 affect you, and you can look at contrib/README, contrib/STANDARDS,
1346 and/or the 0.8.4.27 diff to check.
1347 * In full calls the compiler now does not generate checks for declared
1348 argument types for all arguments.
1349 * various threading fixes
1350 ** and some experimental patches which didn't make it into
1351 the main tree for this release, but which are shipped in
1352 contrib/experimental-thread.patch as a possible fix for some
1353 failures (deadlock, spinning...) in GC-intensive multithreaded
1355 * fixed PPC build problem (source code incompatibility of different
1356 library versions): added offsetof() hackery which attempts to divine
1357 where glibc maintainers put uc_mcontext today
1358 * fixed bug 282: compiler does not trust type assertions while passing
1359 arguments to a full call.
1360 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1361 &OPTIONAL VALUES type parameter.
1362 * fix bug 214: algorithm for noting rejected templates is now more
1363 similar to that of template seletion. (also reported by rydis on
1365 * fixed bug 141b: printing backquoted information readably and prettily
1366 inserts a space where necessary.
1367 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1368 (CHARACTER 10) are now reported as errors, rather than propagated
1369 as unknown types. (reported by piso on #lisp)
1370 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1371 with values NIL and :ERROR. (thanks to Milan Zamazal)
1372 * fixed bug 191c: CLOS now does proper keyword argument checking as
1373 described in CLHS 7.6.5 and 7.6.5.1.
1374 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1375 longer raise an error; further, using a list as a for-as-arithmetic
1376 counter now raises a meaningful error.
1377 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1378 now understood by sequence creation functions such as MAKE-SEQUENCE
1380 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1381 a STRING-INPUT-STREAM.
1382 * compiler enhancement: SIGNUM is now better able to derive the type
1384 * type declarations inside WITH-SLOTS are checked. (reported by
1386 * fixed some bugs revealed by Paul Dietz' test suite:
1387 ** incorrect optimization of TRUNCATE for a positive first
1388 argument and negative second.
1389 ** compiler failure in let-convertion during flushing dead code.
1390 ** compiler failure while deriving type of TRUNCATE on an
1391 interval, containing 0.
1392 ** ASH of a negative bignum by a negative bignum count now returns
1394 ** intersection of CONS types now canonicalizes properly, fixing
1395 inconsistencies in SUBTYPEP.
1397 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1398 * incompatible change: The --disable-debugger command line
1399 option now clobbers the debugger at a more fundamental
1400 level, by redefining #'INVOKE-DEBUGGER instead of by
1401 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1402 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1403 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1404 enter the debugger REPL and then suffer recursive errors
1405 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1406 while under the new system, BREAK in --disable-debugger mode
1407 terminates the system just as an unhandled error would.
1408 * fixed compiler performance when processing loops with a step >1;
1409 * bug fix: DOCUMENTATION now retrieves generic function
1410 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1411 support has been systematized, and now supports the methods
1412 specified by ANSI, along with a default method and a method for
1413 slot documentation. (reported by Nathan Froyd)
1414 * bug fix: effective methods associated with a generic function are
1415 no longer cached over a change of that generic function's method
1416 combination. (reported by Andreas Fuchs)
1417 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1418 names a non-existent directory. (thanks to Andreas Fuchs)
1419 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1420 the PPC platform, lead to stack corruption; this has been fixed.
1421 (reported by Rainer Joswig)
1422 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1423 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1424 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1425 arithmetic in general, is now much more reliable on the PPC
1427 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1428 platform now returns the right answer.
1429 * optimization: restored some effective method precomputation in
1430 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1431 precomputation is now tunable.
1432 * optimization: compiler-internal data structure use has been
1433 reviewed, and changes have been made that should improve the
1434 performance of the compiler by about 20%.
1435 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1436 simple-base-strings and simple-bit-vectors is improved.
1437 * optimization: the optimization of 32-bit logical and arithmetic
1438 functions introduced in version 0.8.3 on the x86 has been
1439 implemented on the mips, ppc and sparc platforms; an
1440 implementation of the same facility, but for 64-bit arithmetic,
1441 has been added for the alpha.
1442 * microoptimization: the compiler is better able to make use of the
1443 x86 LEA instruction for multiplication by constants.
1444 * bug fix: in some situations compiler did not report usage of
1445 generic arithmetic in (SPEED 3) policy.
1446 * bug 145b fix: compiler used wrong type specifier while converting
1447 MEMBER-types to numeric.
1448 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1450 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1451 index argument now works properly on the Alpha platform.
1452 * bug fix: floating point exception treatment on the Alpha platform
1454 * bug fix: FILE-POSITION works much better on string input and
1455 output streams. (thanks to Nikodemus Siivola)
1456 * bug fix: many threading/garbage collection symptoms sorted.
1457 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1458 might be pseudo-atomic.
1459 * internal change: Stopping for GC is now done with signals not
1460 ptrace. GC is now done in whichever thread wanted it, instead of
1462 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1464 * bug fix: result form in DO is not contained in the implicit
1466 * incompatible change: ICR structure is changed; the value part of
1467 CONTINUATION is now called LVAR; corresponding functions are
1468 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1469 * added type deriver for ISQRT (thanks to Robert E. Brown).
1470 * bug fix: better support for loading from the command line when an
1471 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1473 * fixed some bugs revealed by Paul Dietz' test suite:
1474 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1475 ** ROUND and FROUND now give the right answer when given very
1476 small float arguments.
1477 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1479 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1480 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1481 ** LCM should return a non-negative integer.
1482 ** PARSE-INTEGER returned the index of a terminator instead of the
1483 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1484 ** PARSE-INTEGER returned an incorrect index being applied to a
1486 ** LCM with two arguments of 0 returns 0 rather than signalling
1488 ** unsigned addition of a 32-bit constant with the high bit set no
1489 longer causes an internal compiler error on the x86.
1490 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1491 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1492 with negative last argument.
1493 ** byte specifiers with zero size and position no longer cause
1494 an error during type derivation.
1495 ** bignum multiplication on the Alpha platform now returns the
1497 * porting: The system now builds on SuSE AMD64, although it still
1498 generates a 32-bit binary.
1499 * .fasl file incompatibility: The fasl file version number has
1500 been incremented (because of the changes to internal compiler
1501 data structures referred to above).
1503 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1504 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1505 more accurately, on the Darwin kernel running on PowerPC hardware.
1506 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1507 * Compiler code deletion notes now signal a condition of type
1508 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1509 an associated MUFFLE-WARNING restart.
1510 * The compiler now performs limited argument count validation of
1511 constant format strings in FORMAT, and where appropriate in ERROR,
1512 CERROR and WARN. (thanks to Gerd Moellmann)
1513 * New ASDF-INSTALL contrib can be used for automatic download and
1514 installation of third-party Lisp code from CCLAN or other sites
1516 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1517 which forces another thread to execute a function supplied by the
1519 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1520 now accept and act upon their :ELEMENT-TYPE keyword argument.
1521 (reported by Martin Atzmueller, Edi Weitz)
1522 * bug fix: FILE-POSITION now accepts position designators up to
1523 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1524 is the greater. (thanks to Patrik Nordebo)
1525 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1526 Kalle Olavi Niemitalo)
1527 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1528 printable. (reported by Eric Marsden)
1529 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1530 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1531 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1532 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1533 (thanks to Stig Sandoe)
1534 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1535 UNREAD-CHAR now correctly includes the unread character in the
1536 target sequence. (thanks to Gerd Moellmann)
1537 * bug fix in threaded builds: the system can now be suspended and
1538 resumed by shell job control with minimal disruption.
1539 * bug fixes in times and timezones >2038AD
1540 * better handling of "where is GNU make?" problem in build scripts
1541 (thanks to Nikodemus Siivola)
1542 * new optimization: inside a named function any reference to a
1543 function with the same name is considered to be a self-reference;
1544 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1545 optimization quality.
1546 * new optimization on x86: logical functions and + now have
1547 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1548 used when the result is truncated to 32 bits.
1549 * VALUES declaration is partially enabled.
1550 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1551 * fixed some bugs revealed by Paul Dietz' test suite:
1552 ** The system now obeys the constraint imposed by
1553 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1554 types form a lattice under type intersection.
1555 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1556 ** ASSOC now ignores NIL elements in an alist.
1557 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1558 and (1+ MOST-POSITIVE-FIXNUM) answers.
1559 ** The addition of a method with invalid qualifiers to a generic
1560 function does not cause an error to be signalled immediately;
1561 a warning is signalled, and the error is generated only on
1562 calling the generic function.
1563 * changed .fasl file version number, in part to add type codes for
1564 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1565 obscure ANSI requirements
1567 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1568 * fixed bug 148: failure to inline-expand a local function left
1569 garbage, confusing the compiler.
1570 * fixed bugs 3cd: structure slot readers perform type check if the
1571 slot can have an invalid value (i.e. it is either not initialized
1572 or can be written with a less specific slot writer).
1573 * bug fix: the compiler now traps array references to elements off
1574 the end of an array; previously, the bounds checking in some
1575 circumstances could go off-by-one.
1576 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1578 * type declarations for array element types now obey the description
1579 on the CLHS page "Declaration TYPE", as per discussions on
1580 sbcl-help around 2003-05-08. This means that a declaration
1581 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1582 declaration, all references to BAR will be asserted or assumed
1583 (with THE, so dependent on compiler policy) to involve objects of
1584 type FOO. Note that no such declaration is implied in
1585 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1586 * declared types of functions from the "Conditions"
1587 chapter. (reported by Paul Dietz)
1588 * bug fix: CERROR accepts a function as its first argument.
1589 * bug fix: NTH an NTHCDR accept a bignum as index
1590 arguments. (reported by Adam Warner)
1591 * optimization: character compare routines now optimize comparing
1592 against a constant character. (reported by Gilbert Baumann)
1593 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1594 argument now works properly on the MIPS platform.
1595 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1596 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1597 declared to have a functional type, does not cause an error
1599 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1600 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1601 * fixed bug: if last continuation of a deleted block has a
1602 destination, this destination should be deleted too. (reported by
1604 * fixed a bug in the bootstrap process: the host compiler's values
1605 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1606 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1607 test case from Patrik Nordebo)
1608 * improved the ability of the disassembler on the PPC platform to
1609 provide helpful disassembly notes.
1610 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1611 the class in more cases than previously.
1612 * bug fix: FILE-POSITION now understands :START and :END for
1613 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1614 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1615 8) data. (thanks to David Lichteblau)
1616 * bug fix: it is possible to add a method to a generic function
1617 without lambda list.
1618 * bug fix: reader failed to signal END-OF-FILE inside an
1619 object representation. (reported by Nikodemus Siivola)
1620 * fixed some bugs revealed by Paul Dietz' test suite:
1621 ** LAST and [N]BUTLAST should accept a bignum.
1622 ** condition slot accessors are methods.
1623 ** (VECTOR NIL) is a subtype of STRING.
1625 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1626 * minor incompatible change: some nonsensical specialized lambda
1627 lists (used in DEFMETHOD) which were previously ignored now signal
1629 * minor incompatible change: the system is now aware of the types of
1630 variables in the COMMON-LISP package, and will signal errors for
1631 most violations of these type constraints (where previously they
1632 were silently accepted).
1633 * minor incompatible change: COMPILE-FILE now uses the freedom
1634 afforded (ANSI 3.2.2.3) to use derived function types for
1635 functions defined in the same file. This also permits the system
1636 to warn on static type mismatches and function
1637 redefinition. (Currently it does not work with high DEBUG level.)
1638 * minor incompatible change: VALUES declaration is disabled.
1639 * When issuing notes, the compiler now signals a condition of type
1640 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1641 restart for use in user handlers. It is expected that the
1642 COMPILER-NOTE condition will eventually become a condition
1643 supertype to a hierarchy of note types, which will then be
1644 handleable in a similar fashion. However, at the moment, no such
1645 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1646 but it's an implementation detail, not a classification for the
1648 * Changes in type checking closed the following bugs:
1649 ** type checking of unused values (192b, 194d, 203);
1650 ** template selection based on unsafe type assertions (192c, 236);
1651 ** type checking in branches (194bc).
1652 * A short form of VALUES type specifier has ANSI meaning (it has
1653 increased the number of situations when SBCL cannot perform type
1655 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1656 keywords or constants is permissible.
1657 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1658 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1659 classes. (thanks to Antonio Martinez)
1660 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1661 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1662 argument to a pretty-print stream. Instead, it leaves any such
1663 operations to DESCRIBE-OBJECT methods.
1664 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1666 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1667 lambda lists. (thanks to David Lichteblau)
1668 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1669 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1670 (thanks to Gerd Moellmann)
1671 * fixed bug 246: increased compilation speed of long
1672 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1674 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1676 * DEFCONSTANT now throws a condition of type
1677 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1678 constant to a non-EQL value; CONTINUE and ABORT restarts
1679 respectively change and preserve the value.
1680 * fixed bug 63: The code walker, part of the implementation of CLOS,
1681 is now better at handling symbol macros.
1682 * bug fix: There is no longer an internal implementation type named
1683 CL:LENGTH. (reported by Raymond Toy)
1684 * bug fix: In macro-like defining macros/special operators the
1685 implicit block does not enclose the lambda list.
1686 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1687 enclosed in parentheses) are not suitable as type specifiers, and
1688 their use properly signals an error now.
1689 * bug fix: An argument count mismatch for a type specifier in code
1690 being compiled no longer causes an unhandled error at compile
1691 time, but signals a compile-time warning.
1692 * fixed simple vector readable printing
1693 * bug fix: DESCRIBE takes more care over whether the class
1694 precedence list slot of a class is bound before accessing it.
1695 (reported by Markus Krummenacker)
1696 * bug fix: FORMATTER can successfully compile pretty-printer format
1697 strings which use variants of the ~* directive inside.
1698 * bug fix: SEARCH now applies its TEST predicate to the elements of
1699 the arguments in the correct order. (thanks to Wolfhard Buss)
1700 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1701 in type inference. (thanks to Robert E. Brown)
1702 * bug fix: Reading in symbols with an explicit package name of ""
1703 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1704 * fixed some bugs revealed by Paul Dietz' test suite:
1705 ** NIL is now allowed as a structure slot name.
1706 ** Arbitrary numbers, not just REALs, are allowed in certain
1707 circumstances in LOOP for-as-arithmetic clauses.
1708 ** Multiple class redefinitions before slot access no longer
1709 causes a type error.
1710 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1711 association between the name and a class.
1712 ** Generic functions with non-standard method-combination and over
1713 five methods all of which return constants no longer return NIL
1714 after the first few invocations. (thanks to Gerd Moellmann)
1715 ** CALL-NEXT-METHOD with no arguments now passes the original
1716 values of the arguments, even in the presence of assignment.
1717 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1719 ** DISASSEMBLE works with closures and funcallable instances.
1720 ** ADD-METHOD now returns the generic function, not the new method.
1721 ** FIND-METHOD signals an error if the lengths of the specializers
1722 is incompatible with the generic function, even if the ERRORP
1724 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1725 which its argument is a member.
1726 ** DEFCLASS only redefines the class named by its class-name
1727 argument if that name is the proper name of the class;
1728 otherwise, it creates a new class.
1729 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1730 of the UNBOUND-SLOT condition to the name of the slot.
1731 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1732 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1733 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1734 SLOT-MAKUNBOUND in the specified fashion.
1736 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1737 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1738 cross-compilation host. As a consequence, we can now bootstrap our
1739 way up to SBCL starting with a bare gcc toolchain and human-readable
1740 source code (first the source to CLISP, then the source to SBCL).
1741 * A contributed module containing a partial implementation of the
1742 simple-streams interface has been included. (thanks to Rudi
1744 * A contributed module implementing the RFC1321 Message Digest
1745 Algorithm, known as MD5, has been included.
1746 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1747 no longer has any effect, as the code controlled by this feature
1748 has been deleted. (As far as we know, no-one has ever built using
1749 this feature, and its semantics were confused in any case).
1750 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1751 work on conditions (as required by the ANSI specification),
1752 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1753 expected behaviour on conditions. Users should note, however,
1754 that such behaviour is not required by the ANSI specification,
1755 and so use of this behaviour may render their code unportable.
1756 * fixed some bugs revealed by Paul Dietz' test suite:
1757 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1759 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1760 ** SXHASH on condition objects no longer returns NIL.
1761 ** :ALLOCATION :CLASS slots are better treated; their values are
1762 updated on class redefinition, and initforms inherited from
1763 superclasses are applied.
1764 ** REMOVE-METHOD returns its generic function argument even when
1765 no method was removed.
1766 ** SHARED-INITIALIZE now initializes the values of the requested
1767 slots, including those with :ALLOCATION :CLASS.
1768 ** ALLOCATE-INSTANCE now works on structure classes defined via
1769 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1771 ** SLOT-EXISTS-P now works on conditions, as well as structures
1773 ** MAKE-LOAD-FORM now has the required methods on
1774 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1775 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1776 keyword, and now implements the SLOT-NAMES argument.
1777 ** methods with &OPTIONAL arguments no longer allow too many
1778 arguments to be passed in the call without error.
1779 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1780 option is consistent with the required arguments of the generic
1781 function lambda list.
1782 * bug fix: REQUIRE accepts a string designator. (Thanks to
1784 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1785 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1786 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1787 * bug fix: The compiler now checks for duplicated variables in macro
1789 * bug fix: SETQ on globals returns the correct value.
1790 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1791 not a COMPILER-ERROR (followed by some other strange error on
1792 choosing the CONTINUE restart).
1793 * bug fix: make.sh and friends are now more consistent in the way that
1794 they look for GNU "make".
1796 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1797 * experimental native threads support (on x86 Linux >=2.4 only).
1798 This is not compiled in by default: you need to add :SB-THREAD to
1799 the target features. See the "Beyond ANSI" chapter of the manual
1801 * fix for longstanding nonANSIism: The old distinction between
1802 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1803 The return value from CL:FIND-CLASS is now a CLOS class, and
1804 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1805 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1806 * An interface to the MetaObject Protocol, as described in Kiczales,
1807 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1808 MIT Press, 1991, is available from the SB-MOP package.
1809 * incompatible change: the SB-PCL package should now be considered
1810 a private implementation detail, and no longer a semi-private MOP
1812 * minor incompatible change: due to rearrangement for threads, the
1813 control stack and binding stack are now allocated at arbitrary
1814 addresses instead of being hardcoded per-port. Users affected by
1815 this probably have to be doing advanced things with shared
1816 libraries, and will know who they are.
1817 * minor incompatible change: Previously, all --eval forms used were
1818 processed with READ before any of them were processed with EVAL.
1819 Now each --eval form is processed with both READ and EVAL before
1820 the next --eval form is processed. (Thus package operations like
1821 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1822 work as the user might reasonably expect.)
1823 * minor incompatible change: *STANDARD-INPUT* is now only an
1824 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1826 * minor incompatible change: Y-OR-N-P is now character-oriented,
1827 not line oriented. Also, YES-OR-NO-P now works without errors.
1828 (thanks to Antonio Martinez)
1829 * sb-aclrepl module improvements: an integrated inspector, added
1830 repl features, and a bug fix to :trace command.
1831 * Known functions, which cannot be open coded by the backend, are
1832 considered to be able to check types of their arguments. (fixing
1833 a bug report by Nathan J. Froyd)
1834 * fixed a bug in computing method discriminating functions: It is
1835 now possible to define methods specialized on classes which have
1836 forward-referenced superclasses. (thanks to Gerd Moellmann)
1837 * fixed evaluation order in optional entries (reported by Gilbert
1839 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1840 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1841 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1842 required-by-AMOP NAME argument, as well as CLASS and
1843 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1844 * fixed bug 20: DEFMETHOD can define methods using names that are
1845 not the proper names of classes to designate class specializers.
1846 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1847 Unix file descriptors, instead of blowing up. (thanks to Antonio
1849 * Garbage collection refactoring: user-visible change is that a
1850 call to the GC function during WITHOUT-GCING will not do garbage
1851 collection until the end of the WITHOUT-GCING. If you were doing
1852 this you were probably losing anyway.
1853 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1854 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1855 TYPEP the latter but not the former.
1856 * The compiler issues a full WARNING for calls to undefined functions
1857 with names from the CL package.
1858 * MAP-INTO for a vector destination is open coded. (reported by
1859 Brian Downing on c.l.l)
1860 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1861 documentation string.
1862 * fixed some bugs revealed by Paul Dietz' test suite:
1863 ** COPY-ALIST now signals an error if its argument is a dotted
1865 ** Condition slots are now accessed more correctly in the presence
1866 of multiple initargs for a given slot.
1867 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1868 exclude restarts of the same name associated with a different
1870 ** DEFCLASS of forward-referenced classes with another
1871 forward-referenced class in the superclasses list no longer
1873 ** Condition slots are now initialized once each, not multiple
1874 times. (thanks to Gerd Moellmann)
1875 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1877 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1878 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1879 arguments contain duplicated elements.
1880 ** RESTART-CASE understands local macros.
1881 ** RESTART-CASE associates exactly its own restarts with a condition.
1882 ** ENDP in safe mode checks its argument to be of type LIST.
1883 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1884 in question is unbound.
1885 ** Optimized MAKE-INSTANCE functions no longer cause internal
1886 assertion failures in the presence of duplicate initargs.
1887 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1888 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1889 propagated even in the ctor optimized implementation of
1891 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1893 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1894 * a better implementation of SXHASH on (simple) bit vectors,
1895 measured both in execution speed and in distribution of results
1896 over the positive fixnums, has been installed. Likewise, a better
1897 implementation of EQUAL for simple bit vectors is now available.
1898 * fixed CEILING optimization for a divisor of form 2^k.
1899 * fixed bug 240 (emitting extra style warnings "using the lexical
1900 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1901 by Antonio Martinez)
1902 * fixed SXHASH, giving different results for NIL depending on type
1903 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1904 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1905 variable. (found by Rolf Wester)
1906 * SBCL does not ignore type declarations for special
1907 variables. (reported by rif on c.l.l 2003-03-05)
1908 * some bug fixes in contrib/sb-aclrepl/
1909 * fixed some bugs revealed by Paul Dietz' test suite:
1910 ** a bug in the CONS type specifier, whereby the CAR and CDR
1911 types got intertwined, has been fixed;
1912 ** the type system is now able to reason about the interaction
1913 between INTEGER and RATIO types more completely;
1914 ** APPEND, [N]REVERSE and NRECONC check that those their
1915 arguments, which must be proper lists, are really so;
1916 ** An array specialized to be unable to hold elements has been
1917 implemented, as required -- yes, really -- by ANSI;
1918 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1919 on malformed property lists;
1921 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1922 * incompatible packaging change: in line with Unix convention,
1923 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1924 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1925 * REQUIRE and PROVIDE are now optionally capable of doing something
1926 useful. See the documentation string for REQUIRE.
1927 * infrastructure for a managed SBCL contrib system: contributed
1928 modules in this release include:
1929 ** the ASDF system definition facility;
1930 ** an interface to the BSD Sockets API;
1931 ** an ACL-like convenience interface to the repl;
1932 (thanks to Kevin Rosenberg)
1933 ** an implementation of ROTATE-BYTE, with efficient implementation
1935 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1936 gives a closer approximation to the right answer than previously.
1937 (thanks to Raymond Toy)
1938 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1939 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1940 optional environment arguments, as required by ANSI.
1941 * fixed bugs in other functions taking environment objects, allowing
1942 calls with an explicit NIL environment argument to be compiled
1944 * fixed bug 228: primary return values from
1945 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1946 COMPILE or FUNCTION.
1947 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1948 :TYPE will no longer signal errors on innocuous objects.
1949 * fixed bug 231b: SETQ is better at respecting type declarations in
1950 the lexical environment.
1951 * fixed a bug in DEFCLASS: classes named by symbols with no or
1952 unprintable packages can now be defined.
1953 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1954 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1955 * fixed compiler failure related to checking types of functions.
1956 (reported by Robert E. Brown)
1957 * the compiler is now much more consistent in its error-checking
1958 treatment of bounding index arguments to sequence functions: in
1959 (SAFETY 3) code, errors will be signalled in almost all cases if
1960 invalid sequence bounding indices are passed to functions defined
1961 by ANSI to operate on sequences.
1962 * fixed a bug in the build procedure: documentation of SBCL-specific
1963 packages is now preserved and available in the final Lisp image.
1964 * lifted FDEFINITION lookup out of loops in the implementation of
1965 many list operations. (thanks to Robert E. Brown)
1966 * fixed a bug in the reader: the #n# reader macro now works for
1967 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1968 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1969 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1970 E. Brown and Tony Martinez respectively)
1971 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1972 SLOT-BOUNDP: in particular, it is now able to optimize them much
1973 better, and is now not vulnerable to having packages renamed.
1974 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1975 always called when a slot is not present in an instance. (thanks
1977 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1978 longer signal an error on unreading a character following EOF on
1979 the previous constituent stream. (thanks to Tony Martinez)
1980 * fixed some bugs revealed by Paul Dietz' test suite:
1981 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1982 not just nonnegative fixnums;
1983 ** the logical bit-array operators such as BIT-AND now accept an
1984 explicit NIL for their "opt-arg" argument (to indicate a
1985 freshly-consed result bit-array);
1986 ** ELT now signals an error on an invalid sequence index in safe
1988 ** the type system is now cleverer about negations of numeric
1989 types, and consequently understands the BIGNUM and RATIO types
1991 ** the type system is now cleverer about the interaction between
1992 INTEGER and RATIO types: while bugs still remain, many more
1993 cases are accurately computed;
1994 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1995 if it is in the last clause;
1996 ** CONSTANTLY now correctly returns a side-effect-free function in
1998 ** DECLARE is no longer treated as a special-operator; in
1999 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2000 * incremented fasl file version number due to the change in the
2001 DEFSTRUCT-SLOT-DESCRIPTION structure.
2003 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2004 * minor incompatible change: code processed by the "interpreter" or
2005 EVAL now has a compilation optimization policy of (DEBUG 2)
2006 (changed from (DEBUG 1)) to improve debuggability of interactive
2007 development, and to allow the use of the debug RETURN command in
2009 * an experimental implementation of the RETURN command for the
2010 debugger has been included. (thanks to Frederik Kuivinen)
2011 * fixed bug 62: constraints were not propagated into a loop.
2012 * fixed bug in embedded calls of SORT (reported and investigated by
2014 * fixed some bugs revealed by Paul Dietz' test suite:
2015 ** printing and reading of arrays with some dimensions having
2016 length 0 (thanks to Gerd Moellmann);
2017 ** BOA constructor with &AUX argument without a default value does
2018 not cause a type error;
2019 ** CONSTANTP now returns true for all self-evaluating objects.
2021 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2022 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2023 accessors that are related by inheritance, as specified in the
2024 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2025 Valtteri Vuorikoski)
2026 * The compiler is now able to inline functions that were defined in
2027 a complex lexical environment (e.g. inside a MACROLET).
2028 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2030 * Support for the upcoming FreeBSD-5.0 release has been included.
2031 (thanks to Dag-Erling Smorgrav)
2032 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2033 effect when it is not in a toplevel context.
2034 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2035 stand a better chance of being correct. (thanks to Gerd
2037 * fixed bug in COERCE, which now signals an error on coercing a
2038 rational to a bounded real type which excludes the expected
2040 * The compiler is now able to derive types more accurately from the
2041 COERCE and COMPILE functions.
2042 * fixed bug 223: functional binding is considered to be constant
2043 only for symbols in the CL package.
2044 * fixed bug 231: SETQ did not check the type of a variable being set
2045 (reported by Robert E. Brown)
2046 * A new optimization for MAKE-INSTANCE has been included, fixing
2047 various bugs (including relating to :ALLOCATION :CLASS slots and
2048 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2050 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2051 ** As required by ANSI, LOOP now disallows anonymous collection
2052 clauses such as COLLECT I in conjunction with aggregate boolean
2053 clauses such as THEREIS (= I 1);
2054 ** LOOP now signals an error when any variable is reused in the
2055 same loop (including the potentially useful construct analogous
2056 to WITH A = 1 WITH A = (1+ A);
2057 ** IT is only a special loop symbol within the first clause of a
2058 conditional loop clause;
2059 ** LOOP with a typed iteration variable over a hashtable now
2060 signals a type error iff it should.
2061 * fixed some other bugs revealed by Paul Dietz' test suite:
2062 ** FILE-STREAM now names the class previously known as FD-STREAM;
2063 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2064 argument) no longer signals an error;
2065 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2066 of the slot symbol, rather than using the current package
2067 ((:CONC-NAME "") continues to intern the slot's name in the
2069 * incremented fasl file version number, because of the incompatible
2070 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2071 of the new implementation of DEFINE-COMPILER-MACRO.
2073 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2074 * Support for building SBCL for MIPS platforms running in
2075 little-endian mode has now been checked in, and basic
2076 functionality on said platforms verified.
2077 * minor incompatible change: PCL now records the pathname of a file
2078 in which methods and the like are defined, rather than its
2080 * minor incompatible change: TRUENAME now considers the truename of
2081 a file naming a directory to be the pathname with :DIRECTORY
2082 component indicating that directory.
2083 * minor incompatible change: a NAMED clause in the extended form of
2084 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2085 reason for the previous behaviour is unclear.
2086 * more systematization and improvement of CLOS and MOP conformance
2087 in PCL (thanks to Gerd Moellman and Pierre Mai):
2088 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2090 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2091 primary methods with no specializers;
2092 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2094 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2095 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2096 FORWARD-REFERENCED-CLASSes; error reporting on
2097 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2099 ** SXHASH on CLOS instances now uses a slot internal to the
2100 instance to return different numbers on distinct instances,
2101 while preserving the same return value through invocations of
2103 ** DEFMETHOD signals errors when methods with longer incongruent
2104 lambda lists are added to generic functions;
2105 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2106 CLASS, as specified in AMOP;
2107 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2108 on the order returned by the primary method for classes of
2109 class STANDARD-CLASS;
2110 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2111 * fixed some bugs shown by Paul Dietz' test suite:
2112 ** DOLIST puts its body in TAGBODY;
2113 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2115 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2116 value producing form;
2117 ** if more variables are given to PROGV than values, extra
2118 variables are bound and made to have no value;
2119 ** NSUBSTITUTE on list arguments gets the right answer with
2121 ** ELT signals an error of type TYPE-ERROR when the index argument
2122 is not a valid sequence index;
2123 ** LOOP signals (at macroexpansion time) an error of type
2124 PROGRAM-ERROR when duplicate variable names are found;
2125 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2126 ** FUNCALL of special-operators now cause an error of type
2128 ** PSETQ now works as required in the presence of side-effecting
2129 symbol-macro places;
2130 ** NCONC accepts any object as its last argument;
2131 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2133 ** loop-for-as-package does not require a package to be explicitely
2135 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2137 * fixed bug 166: compiler preserves "there is a way to go"
2138 invariant when deleting code.
2139 * fixed bug 172: macro lambda lists with required arguments after
2140 &REST arguments now cause an error to be signalled. (thanks to
2142 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2143 bug. (thanks to Matthew Danish)
2144 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2146 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2147 arguments are lexically rebound. (thanks to Gerd Moellmann and
2149 * fixed bug 194: error messages are now more informative when there
2150 is no primary method applicable in a call to a generic
2151 function. (thanks to Gerd Moellmann)
2152 * fixed bug in command line argument checking (thanks to Julian
2154 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2155 (thanks to Matthew Danish)
2156 * incremented fasl file version number, because of the
2157 SXHASH-related changes in the layout of CLOS data structures
2159 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2160 * minor incompatible change: The runtime (the Unix executable named
2161 "sbcl") is now much pickier about the .core files it will load.
2162 Essentially it now requires .core files to descend from the same
2163 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2164 as the runtime does. (The intent is to prevent the crashes which
2165 can occur, and which can even be reported as mysterious failures,
2166 when people patch the sources or change the build parameters
2167 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2168 sbcl and .core files.)
2169 * fixed bug: VALUES-LIST is no longer optimized away.
2170 * fixed bug 142: The FFI conversion of C string values to Lisp
2171 string values no longer conses excessively. (thanks to Nathan
2172 Froyd porting Raymond Toy's fix to CMU CL)
2173 * began to systematize and improve MOP conformance in PCL (thanks to
2174 Nathan Froyd, Gerd Moellman and Pierre Mai):
2175 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2177 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2178 ** FINALIZE-INHERITANCE is now called on class finalization;
2179 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2180 argument precedence order.
2181 * fixed bug 202: The compiler no longer fails on functions whose
2182 derived types contradict their declared type.
2183 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2184 so it can be non-toplevel.
2185 * The fasl file version number has changed (because of the new
2186 implementation of DEFMACRO).
2187 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2188 safe code, that any length requirement by their type-specifier
2189 argument is valid. The exceptions to this are described in bug
2191 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2192 functions now check the types of their inputs as required by ANSI.
2193 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2194 introduced symbol is DECLAREd to be SPECIAL.
2195 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2196 * fixed bug: :COUNT argument to sequence functions may be negative
2197 * fixed bug: body of DO-SYMBOLS may contain declarations
2198 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2199 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2200 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2201 symbol macro only once
2202 * fixed printing of call frame when argument list is unavailable
2203 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2204 * compiler no longer signals WARNING on unknown keyword
2207 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2208 * A beta-quality port to the mips architecture running Linux,
2209 based on the old CMUCL backend, has been made. It has been tested
2210 on a big-endian kernel, and works sufficiently well to be able to
2211 rebuild itself; it has not been tested in little-endian mode.
2212 * fixed an inconsistency between gencgc.c and purify.c which made
2213 dumping/loading .core files unreliable
2214 * fixed bug 120a: The compiler now deals correctly with IFs where
2215 the consequent is the same as the alternative, instead of
2216 misderiving the return type. (thanks to Alexey Dejneka)
2217 * fixed bug 113: Logical pathnames are now dumpable (the logical
2218 host is resolved at load-time, throwing an error if it is not
2220 * fixed bug 174: FORMAT's error message is slightly clearer when a
2221 non-printing character is used in a format directive.
2222 * fixed several bugs in compiler checking of type declarations, i.e.
2223 violations of the Python "declarations are assertions" principle
2224 (thanks to Alexey Dejneka)
2225 * fixed several bugs in PCL's error checking (thanks to Gerd
2227 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2229 * fixed bug in compilation of functions as first class values
2230 (thanks to Antonio Martinez-Shotton)
2231 * The compiler's handling TYPE-ERRORs which it can prove will
2232 inevitably happen at runtime has been cleaned up and corrected
2233 in several ways. (thanks to Alexey Dejneka)
2234 * improved argument type checking for various basic arithmetic
2235 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2236 lavished on them in the past that they can be compiled in many
2237 ways in different special cases
2238 * fixed bug 181: compiler checks validity of user supplied type
2240 * cleaned up code flushing in optimization: Function calls which
2241 should signal errors for safety purposes (e.g. which ANSI says
2242 should signal errors when their arguments are of incorrect type)
2243 are no longer optimized away.
2244 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2245 * incremented fasl file version number, because changes in the
2246 implementation of sequence functions like COERCE caused
2247 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2248 in old inline expansions) to become undefined. (Actually these
2249 changes were later undone, so we might very well be binary
2250 compatible with 0.7.7 after all, but leaving the version number
2251 incremented seemed like the simplest and most conservative
2254 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2255 * An alpha-quality port to the parisc architecture running Linux,
2256 based on the old CMUCL backend, has been made. This, even more so
2257 than the other backends, should be considered still a work in
2258 progress; known problems include that the Linux kernel in 64-bit
2259 mode does not propagate the correct sigcontext structure to
2260 userspace, and consequently SBCL on a parisc64 kernel will not
2262 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2263 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2264 suppressed.) Also now that the compiler is looking at declarations
2265 in the environment, it checks optimization declarations as well,
2266 and suppresses inlining when (> DEBUG SPEED).
2267 * More fixes have been made to treatment of floating point exception
2268 treatment and other Unix signals. In particular, floating point
2269 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2270 * The detection and handling of control stack exhaustion (infinite
2271 or very deeply nested recursion) has changed. Stack exhaustion
2272 detection is now done by write-protecting pages at the OS level
2273 and applies at all optimization settings; when found, a
2274 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2275 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2276 be caught using IGNORE-ERRORS.
2277 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2278 correctly when transforming recursive LABELS functions to LETs.
2279 (thanks to Alexey Dejneka)
2280 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2281 that are names of constants or global variables.
2282 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2283 alien routines with docstrings.
2284 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2285 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2287 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2288 (thanks to Eric Marsden)
2289 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2290 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2291 * Bug fix: The SPARC backend can now compile functions involving
2292 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2293 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2294 to a functions expecting a PATHNAME-DESIGNATOR.
2295 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2296 lambda lists. (thanks to Alexey Dejneka)
2297 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2298 functions so that EXPORTing the name of the class doesn't cause
2299 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2300 to Antonio Martinez for reporting this)
2301 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2302 constant-folded without failing an assertion. (thanks to Einar
2303 Floystad Dorum for reporting this)
2304 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2305 types are now dealt with more correctly by the compiler.
2306 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2307 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2308 behaviour analogous to RENAME-FILE. This puts its behaviour more
2309 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2311 * The fasl file version number has changed again. (because of the
2312 bug fix involving the names of PCL MAKE-INSTANCE functions)
2314 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2315 * bug fix: Floating point exceptions are treated much more
2316 consistently on the x86/Linux and PPC/Linux platforms.
2317 * Array initialization with :INITIAL-ELEMENT is now much faster for
2318 cases when the compiler cannot open code the array creation, but
2319 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2320 array accesses have also seen a speed increase.
2321 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2322 specified. (This was at the root of some bad interactions between
2323 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2325 * bug fix: Internal error arguments for undefined functions are now
2326 computed correctly on the PPC/Linux platform.
2327 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2328 Raymond Toy's patch for CMU CL)
2329 * Support for the Solaris 9 operating environment has been included
2330 (thanks to Daniel Merritt)
2331 * A very ugly but hopefully complete draft of the missing FFI chapter
2332 of the manual has been created by reformatting the corresponding
2333 CMU CL manual chapter into (currently very ugly and incoherent)
2334 DocBook and bringing it up to date for SBCL behavior. Thus, the
2335 manual is now essentially complete, at least by my extreme
2336 once-and-only-once standards, whereby it's acceptable to refer to
2337 the doc strings of SB-EXT functions as the primary documentation.
2338 * The fasl file version number has changed again, due to cleanup of
2339 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2340 is no longer a static symbol.)
2342 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2343 * SBCL now builds with OpenMCL (version 0.12) as the
2344 cross-compilation host; also, more progress has been made toward
2345 bootstrapping under CLISP.
2346 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2348 * bug 158 fixed: The compiler can now deal with integer loop
2349 increments different from 1; fixing this turned out also to fix
2351 * bug 169 fixed: no more bogus warnings about using lexical bindings
2352 despite the presence of perfectly good SPECIAL declarations (thanks
2353 to David Lichteblau)
2354 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2355 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2356 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2358 * bug 180 fixed: Method combination specifications no longer ignore
2359 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2360 * bug fix: Structure type predicate functions now check their argument
2361 count as they should.
2362 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2363 correctly. (thanks to Pierre Mai)
2364 * minor incompatible change: The --noprogrammer option is deprecated
2365 in favor of the new --disable-debugger option, which is very similar.
2366 (The major difference is that it takes effect at a slightly different
2367 time at startup, causing handling of errors in --sysinit and
2368 --userinit files will be affected differently.) The
2369 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2370 been added to allow this functionality to be controlled from ordinary
2371 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2372 maintainers, who might want to run non-interactive scripts to
2373 build SBCL cores which will later be used interactively.)
2374 * minor incompatible change: The LOAD function no longer, when given
2375 a wild pathname to load, loads all files matching that pathname.
2376 Instead, an error of type FILE-ERROR is signalled.
2378 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2379 * bug 147 fixed: The compiler preserves its block link/count
2380 invariants more correctly now so that it doesn't crash. (thanks
2382 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2384 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2385 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2386 no-value-for-this-character as NIL (as specified by ANSI).
2387 * HOST-NAMESTRING on physical pathnames now returns a string that is
2388 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2389 (thanks to Christophe Rhodes)
2390 * The Alpha port handles icache flushing more correctly. (thanks to
2392 * More progress has been made toward bootstrapping under CLISP. (thanks
2393 to Christophe Rhodes)
2394 * The fasl file format has changed again, because dynamic loading
2395 on OpenBSD (which has non-ELF object files) motivated some cleanups
2396 in the way that foreign symbols are transformed and passed around.
2397 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2398 is no longer treated as whitespace by the reader, but instead as
2399 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2400 |A
\7fB|, instead of A as it used to.
2402 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2403 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2404 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2405 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2406 this as of 0.7.1.45, but was left out of the previous news section
2407 (thanks to Dan Barlow)
2408 * SBCL now runs on the Solaris operating system on SPARC architectures
2409 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2410 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2411 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2412 * SPARC backend cleanups, allowing builds of cores optimized for V8
2413 and V9 SPARCS, and also emission of code targeted to a particular
2414 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2416 * SBCL is closer to bootstrapping under CLISP, thanks to various
2417 fixes by Christophe Rhodes.
2418 * The fasl file format has changed again, to allow the compiler's
2419 INFO database to support symbol macros.
2420 * The user manual (in doc/) is formatted into HTML more nicely.
2421 (thanks to coreythomas)
2422 * The system is smarter about SUBTYPEP relationships, especially
2423 those involving NOT types (including types such as ATOM which are
2424 represented internally using NOT types). Thus SUBTYPEP is less
2425 likely to return (VALUES NIL NIL) in general, and in particular
2426 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2428 * The fasl file format has changed again, because the internal
2429 representation of types now includes a new slot to support the new
2430 SUBTYPEP-of-NOT-types logic.
2431 * (not a change in the main branch of SBCL, but a related prototype
2432 which can hopefully be merged into the main branch of SBCL in the
2433 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2434 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2435 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2436 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2437 dispatch macro character combinations. (thanks to Alexey Dejneka)
2438 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2439 that superficially look like logical namestrings correctly.
2440 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2442 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2443 default supplied-p) for &optional and &key arguments. (thanks to
2446 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2447 * incompatible change: The compiler is now less aggressive about
2448 tail call optimization, doing it only when (> SPACE DEBUG) or
2449 (> SPEED DEBUG). (This is an incompatible change because there are
2450 programs which relied on the old CMU-CL-style behavior to optimize
2451 away their unbounded recursion which will now die of stack overflow.)
2452 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2453 for non-GENCGC systems has been increased to 20M (since that
2454 seems much closer to the likely performance optimum for modern
2455 systems than the old 4M value was)
2456 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2457 and *DEBUG-PRINT-LEVEL*
2458 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2459 of CMU CL's support for SPARC, and various endianness and other
2460 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2461 * new syntactic sugar for the Unix command line: --load foo.bar is now
2462 an alternate notation for --eval '(load "foo.bar")'.
2464 ** The system now detects stack overflow and handles it gracefully,
2465 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2466 optimization settings. (This is a good thing in general, and
2467 its introduction in this version should be particularly timely
2468 for anyone whose code fails because of suppression of tail
2470 ** The system now hunts for the C variable "environ" in a more
2471 devious way, to avoid segfaults when the C library version differs
2472 between compile time and run time. (thanks to Christophe Rhodes)
2473 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2474 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2475 ** The compiler no longer issues bogus style warnings for undefined
2476 classes in the same source file as the DEFCLASSes which defined
2477 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2479 ** fixes in CONDITION class precedence list for undefined function
2480 errors (thanks to Alexei Dejneka)
2481 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2482 correctly. (thanks to Dan Barlow)
2483 ** portability fixes aiming at bootstrapping under CLISP (thanks
2484 to Dave McDonald and Christophe Rhodes)
2485 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2486 ** fixes in type translation and and type inference (thanks to
2488 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2489 ** various fixes in the new ports (thanks to Dan Barlow)
2490 * several changes related to debugging:
2491 ** suppression of tail recursion, as noted above
2492 ** stack overflow detection, as noted above
2493 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2494 is now the default. (For some time encapsulation has been more
2495 reliable than the breakpoint-based :ENCAPSULATE NIL
2496 implementation, at least on X86 systems; and I just noticed that
2497 encapsulation also seems closer to the spirit of the ANSI
2500 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2502 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2503 up properly again. (There was a packaging bug in 0.7.0 which
2504 left their definitions in SB-SYS::LOAD-FOREIGN and
2505 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2506 vital for most things which interface to C-level interfaces,
2507 like extensions working with sockets or databases or
2508 Perl-compatible regexes or whatever, and the need to fix
2509 this bug is the main reason that 0.7.1 was released so
2511 ** DEFGENERIC is now choosier about the methods it redefines, so that
2512 reLOADing a previously-LOADed file containing DEFGENERICs does
2513 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2514 cycle now works as it should. (thanks to Alexey Dejneka)
2515 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2516 fixed by Christophe Rhodes. (He actually submitted this patch
2517 months ago, and I delayed until after 0.7.0.)
2518 ** Bug 111 (internal compiler confusion about runtime checks on
2519 FUNCTION types) was fixed by Alexey Dejneka.
2520 * Some internal cleanups (getting rid of variables which aren't
2521 needed now that the byte interpreter is gone) caused the fasl
2522 file format number to change again.
2524 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2525 * major incompatible change: The default fasl file extension, i.e. the
2526 default extension for files produced by COMPILE-FILE, has changed
2527 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2529 ** There are many changes in the implementation of the compiler.
2530 SBCL is now essentially a compiler-only implementation of ANSI
2531 Common Lisp. EVAL still "interprets" a few special cases, but
2532 almost all the interesting cases are handled by creating
2533 a LAMBDA expression, calling COMPILE on it, then calling
2534 FUNCALL on the result.
2535 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2536 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2537 Since the code is newer, there might still be some new bugs
2538 (though not as many as before Martin Atzmueller's fixes:-). But
2539 the new code is substantially simpler and clearer, and hopefully
2540 any remaining bugs will be simpler, less fundamental, and more
2541 fixable then the bugs in the old code.
2542 ** The revised compiler is still a little unsteady on its feet.
2544 *** The debugging information it produces (particularly the names
2545 of FUNCTION objects) is sometimes much less useful than what
2546 the old compiler produced.
2547 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2548 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2549 within a MACROLET) has been temporarily weakened.
2550 ** There are new compiler optimizations for various functions:
2551 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2552 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2553 *** the math functions TRUNCATE, FLOOR, and CEILING
2554 *** the function-of-all-trades COERCE
2555 Mostly these should be transparent, but there's one
2556 potentially-annoying problem (bug 117): when the compiler
2557 inline-expands a function and does type analysis on the result,
2558 it can create control paths which have type mismatches, and
2559 when it can't prove that those control paths aren't taken,
2560 it will issue WARNINGs about the type mismatches. This is
2561 a particular problem in practice for the new sequence functions.
2562 It's not clear how this should be fixed, and for now, a
2563 workaround is given in the entry for 117 in the BUGS file.
2564 ** (Because of the interaction between the two previous items --
2565 occasional inlining problems and new inline expansions -- some
2566 of the new sequence function optimizations won't really kick in
2567 completely until debugging information, and then inlining, are
2568 straightened out in some future version.)
2569 * minor incompatible changes:
2570 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2571 DIRECTORY no longer implicitly promotes NIL slots of its
2572 pathname argument to :WILD. In particular, when you ask for the
2573 contents of a directory (which you used to be able to do without
2574 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2575 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2576 ** changes in behavior that ANSI explicitly defines to be
2577 implementation dependent:
2578 *** The new compiler-only implementation still conforms with ANSI,
2579 but acts a little different than before. Besides the obvious
2580 changes in performance tradeoffs (that the cost per form passed
2581 to EVAL has gone up, and the cost per form executed by EVAL
2582 has gone down), the behavior of the system changes a little
2583 because there are no longer any interpreted function objects.
2584 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2585 e.g. doing COMPILE on the output of interactive DEFUN is
2587 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2589 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2590 has changed from (:CL) to NIL.
2591 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2592 e.g. #\Newline and #\Tab, don't have names specified in the
2593 ANSI Common Lisp standard, is now based on their ASCII symbolic
2594 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2595 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2596 are no longer used for output.
2597 ** changes in internal implementation constants:
2598 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2599 4 million. (If your application spends a lot of time GCing and
2600 you have a lot of RAM, you might want to experiment with
2601 increasing it even more.)
2602 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2603 However, almost all old code should still continue to work without
2604 immediate update, as SB-C-CALL is now a (deprecated) nickname
2606 ** Old operator names in the style DEF-FOO are now deprecated in
2607 favor of new corresponding names DEFINE-FOO, for consistency with
2608 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2609 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2610 internal symbols, but a few supported extensions like
2611 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2612 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2613 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2614 etc. as you get deeper into recursive calls to the debugger
2615 command loop, instead of the old "5]", "5]]", "5]]]"
2616 sequence. (I was motivated to do this when squabbles between
2617 ILISP and SBCL left me very deeply nested in the debugger. In the
2618 short term, this change will probably provoke more ILISP/SBCL
2619 squabbles, but hopefully it will be an improvement in the long run.)
2620 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2621 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2622 function, since declaiming return types involving aliens is
2623 (1) annoyingly messy to do by hand and (2) vital to efficient
2624 compilation of code which calls such functions.
2625 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2626 longer reexported by the SB-EXT package. They're solely useful
2627 for alien code, so it seems more logical that you should get
2628 them from the SB-ALIEN package, not in SB-EXT.
2629 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2630 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2631 features. Instead, the code that they used to control is always
2632 built into the system.
2633 * many other bug fixes
2634 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2635 advantage of the new EVAL-WHEN stuff and to clean them up in
2636 general, and they are now more ANSI-compliant in a number of
2637 ways. Martin Atzmueller is responsible for a lot of this.
2638 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2640 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2641 *** correct ERROR type for various file operations
2642 *** some fixes for Lisp streams
2643 *** DEFMETHOD syntax checking
2644 *** changing old weird representation of debug information as
2645 strings (which, among their other deficiencies, don't transform
2646 correctly when you rename packages, and don't change their
2647 print representation when you change things like *PACKAGE*
2648 and *PRINT-LENGTH*) to symbols and lists of symbols
2649 He also made several improvements and fixed several bugs in DESCRIBE.
2650 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2652 *** misbehavior of WRITE-STRING/WRITE-LINE
2653 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2654 and several other LOOP problems as well
2655 *** DIRECTORY when similar filenames are present
2656 *** DEFGENERIC with :METHOD options
2657 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2658 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2659 *** argument ordering in FIND with :TEST option
2660 *** mishandled package designator argument in APROPOS-LIST
2661 *** various problems in the backquote readmacro
2662 *** a bug in APROPOS
2663 *** probably some others that I'm not describing very well here,
2664 since the CVS log documents them by reference to sbcl-devel
2665 messages, and the SourceForge archives aren't working well.:-(
2666 ** Dan Barlow improved the Alpha port (and is making progress on the
2667 PPC port, for those of you who think different).
2668 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2669 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2670 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2671 continues to work on the SPARC port (for those of you in a position
2672 to look down upon our little PC-compatible boxes from a great height).
2673 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2674 and uses that copy, rather than the current dynamic value, when
2675 it's trying to decide whether to truncate output. Thus e.g.
2676 (let ((*print-lines* 50))
2677 (pprint-logical-block (stream nil)
2679 (let ((*print-lines* 8))
2680 (print (aref possiblybigthings i) stream)))))
2681 should now truncate the logical block only at 50 lines, instead of
2682 often truncating it at 8 lines, as it did before.
2683 * The doc/cmucl/ directory, containing old CMU CL documentation
2684 from the time of the fork, is no longer part of the base system.
2685 SourceForge has shut down its anonymous FTP service, and with it
2686 my original plan for distributing the old CMU CL documentation
2687 there. For now, if you need these files you can download an old
2688 SBCL source release and extract them from it.
2689 * The fasl file version number changed again, for dozens of reasons,
2690 some of which are apparent above.
2692 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2693 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2694 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2695 patch, so that DISASSEMBLE output is much nicer.
2696 * The code in the SB-PROFILE package now seems reasonably stable.
2697 I still haven't decided what the final interface should look like
2698 (I'd like PROFILE to interact cleanly with TRACE, since both
2699 facilities use function encapsulation) but if you have a need
2700 for profiling now, you can probably use it successfully with
2701 the current CMU-CL-style interface.
2702 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2703 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2704 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2705 link now returns the dangling link itself, and for similar
2706 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2707 link itself. (In these cases the old code signalled an error and
2708 looped endlessly, respectively.) Thus, DIRECTORY now works even
2709 in the presence of dangling and cyclic symbolic links.
2710 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2711 is now a supported extension again, since the consensus on
2712 sbcl-devel was that it can be useful for ordinary development
2713 work, not just for debugging SBCL itself.
2714 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2715 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2716 that currently-defined functions might be redefined later with
2717 different return types.
2718 * Hash tables can be printed readably, as inspired by CMU CL code
2719 of Eric Marsden and SBCL code of Martin Atzmueller.
2720 * better error handling in CLOS method combination, thanks to
2721 Martin Atzmueller porting Pierre Mai's CMU CL patches
2722 * more overflow fixes for >16Mbyte I/O buffers
2723 * A bug in READ has been fixed, so that now a single Ctrl-D
2724 character suffices to cause end-of-file on character streams.
2725 In particular, now you only need one Ctrl-D at the command
2726 line (not two) to exit SBCL.
2727 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2729 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2730 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2731 does the right thing.
2732 * The implementation of some type tests, especially for CONDITION
2733 types, is now tidier and maybe faster, due to CMU CL code
2734 originally by Douglas Crosher, ported by Martin Atzmueller.
2735 * Some math functions have been fixed, and there are new
2736 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2737 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2738 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2739 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2740 likely to be built into sbcl-0.7.0.)
2741 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2742 CL, ported to SBCL by Martin Atzmueller)
2743 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2744 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2745 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2746 * There's a new slam.sh hack to shorten the edit/compile/debug
2747 cycle for low-level changes to SBCL itself, and a new
2748 :SB-AFTER-XC-CORE target feature to control the generation of
2749 the after-xc.core file needed by slam.sh.
2750 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2751 COMPILE-FILE is no longer supported, so that now every function
2752 gets an entry point, so that block compilation looks a little
2753 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2754 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2755 returns the number of bytes consed since the system started,
2756 rather than the number consed since the first time the function
2757 was called. (The new definition parallels ANSI functions like
2758 CL:GET-INTERNAL-RUN-TIME.)
2759 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2760 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2761 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2762 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2763 * Fasl file version numbers are now independent of the target CPU,
2764 since historically most system changes which required version
2765 number changes have affected all CPUs equally. Similarly,
2766 the byte fasl file version is now equal to the ordinary
2769 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2770 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2771 is no longer recognized. I apologize for this, because it was
2772 listed in SB-EXT as a supported extension, but I found that
2773 its existing behavior was poorly specified, as well as incorrectly
2774 specified, and it looked like too much of a mess to straighten it
2775 out. I have enough on my hands trying to get ANSI stuff to work..
2776 * many patches ported from CMU CL by Martin Atzmueller, with
2777 half a dozen bug fixes in pretty-printing and the debugger, and
2778 half a dozen others elsewhere
2779 * fixed bug 13: Floating point infinities are now supported again.
2780 They might still be a little bit flaky, but thanks to bug reports
2781 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2782 as flaky as they were.
2783 * The --noprogrammer command line option is now supported. (Its
2784 behavior is slightly different in detail from what the old man
2785 page claimed it would do, but it's still appropriate under the
2786 same circumstances that the man page talks about.)
2787 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2788 are now supported, and enabled by default. Thus, the compiler can
2789 handle many floating point and complex operations much less
2790 inefficiently. (Thus e.g. you can implement a complex FFT
2792 * The compiler now detects type mismatches between DECLAIM FTYPE
2793 and DEFUN better, and implements CHECK-TYPE more correctly, and
2794 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2795 thanks to patches from Martin Atzmueller.
2796 * various fixes to make the cross-compiler more portable to
2797 ANSI-conforming-but-different cross-compilation hosts (notably
2798 Lispworks for Windows, following bug reports from Arthur Lemmens)
2799 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2800 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2801 fixed thanks to Pierre Mai's CMU CL patches.
2802 * fixes to keep the system from overflowing internal counters when
2803 it tries to use i/o buffers larger than 16M bytes
2804 * fixed bug 45a: Various internal functions required to support
2805 complex special functions have been merged from CMU CL sources.
2806 (When I was first setting up SBCL, I misunderstood a compile-time
2807 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2808 * improved support for type intersection and union, fixing bug 12
2809 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2810 more obscure bugs as well
2811 * some steps toward byte-compiling non-performance-critical
2812 parts of the system, courtesy of patches from Martin Atzmueller
2813 * Christophe Rhodes has made some debian packages of sbcl at
2814 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2815 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2816 stable, but are nonetheless usable. When he's ready, I'd be happy
2817 to add them to the SourceForge "File Releases" section. (And if
2818 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2819 * new fasl file format version number (because of changes in
2820 internal representation of (OR ..) types to accommodate the new
2821 support for (AND ..) types, among other things)
2823 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2824 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2826 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2828 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2829 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2830 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2831 * DESCRIBE now works on structure objects again.
2832 * Most function call argument type mismatches are now handled as
2833 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2834 know whether the function will be redefined before the call is
2835 executed. (The compiler could flag local calls with full WARNINGs,
2836 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2837 it doesn't keep track of enough information to know whether calls
2838 are local in this sense.)
2839 * Compiler output is now more verbose, with messages truncated
2840 later than before. (There should be some supported way for users
2841 to override the default verbosity, but I haven't decided how to
2842 provide it yet, so this behavior is still controlled by the internal
2843 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2844 src/compiler/ir1util.lisp.)
2845 * Fasl file format version numbers have increased again, because
2846 support for the Gray streams extension changes the layout of the
2847 system's STREAM objects.
2848 * The Gray subclassable streams extension now works, thanks to a
2849 patch from Martin Atzmueller.
2850 * The full LOAD-FOREIGN extension (not just the primitive
2851 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2852 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2853 but like most other programs, it defaults to copying the Unix
2854 environment from the original process instead of starting the
2855 new process in an empty environment.
2856 * Extensions which manipulate the Unix environment now support
2857 an :ENVIRONMENT keyword option which doesn't smash case or
2858 do other bad things. The CMU-CL-style :ENV option is retained
2859 for porting convenience.
2860 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2861 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2863 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2865 * A patch from Martin Atzmueller seems to have solved the SIGINT
2866 problem, and as far as we know, signal-handling now works cleanly.
2867 (If you find any new bugs, please report them!)
2868 * The system no longer defaults Lisp source file names to types
2869 ".l", ".cl", or ".lsp", but only to ".lisp".
2870 * The compiler no longer uses special default file extensions for
2871 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2872 seems to expect a single default extension for all compiled code,
2873 and there's no compelling reason to try to stretch the standard
2874 to allow two different extensions.) Instead, byte-compiled files
2875 default to the same extension as native-compiled files.
2876 * Fasl file format version numbers have increased again, because
2877 a rearrangement of internal implementation packages made some
2878 dumped symbols in old fasl files unreadable in new cores.
2879 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2880 many fewer weird special cases.
2881 * Bug #17 (differing COMPILE-FILE behavior between logical and
2882 physical pathnames) has been fixed, and some related misbehavior too,
2883 thanks to a patch from Martin Atzmueller.
2884 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2885 by Tim Moore, ported to SBCL by Martin Atzmueller.
2886 * Martin Atzmueller fixed several filesystem-related problems,
2887 including bug #36, in part by porting CMU CL patches, which were
2888 written in part by Paul Werkowski.
2889 * More compiler warnings in src/runtime/ are gone, thanks to
2890 more patches from Martin Atzmueller.
2891 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2894 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2896 * DESCRIBE now works on CONDITION objects.
2897 * The debugger now handles errors which arise when trying to print
2898 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2900 * The build system now uses an additional file, customize-target-features.lisp,
2901 to allow local modifications to the target *FEATURES* list. (The point of
2902 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2903 debugging features enabled, without having to worry about propagating it
2904 into everyone's system when I do a "cvs update".) When no
2905 customize-target-features.lisp file exists, the target *FEATURES* list
2906 should be constructed the same way as before.
2907 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2908 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2909 warn about behavior which is undefined under the ANSI spec. Note: This
2910 is specified by ANSI, but it's not very popular with programmers.
2911 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2912 macro in the SBCL sources for an example of a workaround which you
2913 might use to make portable ANSI-standard code which does what you want.
2914 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2915 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2916 when it's used as a non-toplevel form. (This is required in order
2917 to implement the DEFCONSTANT-EQX macro.)
2918 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2919 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2920 funny things, due to bugs in the implementation of EVAL-WHEN
2921 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2922 non-toplevel DEFCONSTANTs are uncommon.)
2923 * The core file version number and fasl file version number have been
2924 incremented, because the old noncompliant DEFCONSTANT behavior involved
2925 calling functions which no longer exist, and because I also took the
2926 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2927 * fixed bug 1 (error handling before read-eval-print loop starts), and
2928 redid debugger restarts and related debugger commands somewhat while
2930 ** The QUIT debugger command is gone, since it did something
2931 rather different than the SB-EXT:QUIT command, and since it never
2932 worked properly outside the main toplevel read/eval/print loop.
2933 Invoking the new TOPLEVEL restart provides the same functionality.
2934 ** The GO debugger command is also gone, since you can just invoke
2935 the CONTINUE restart directly instead.
2936 ** The TOP debugger command is also gone, since it's redundant with the
2937 FRAME 0 command, and since it interfered with abbreviations for the
2939 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2941 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2942 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2943 some porting work by Martin Atzmueller)
2944 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2945 with a boatload of patches which he ported to SBCL. Now that those
2947 ** The system tries to make sure that its low-priority messages
2948 are prefixed by semicolons, to help people who like to use
2949 syntax highlighting in their ILISP buffer. (This patch
2950 was originally due to Raymond Toy.)
2951 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2952 patches originally written by Raymond Toy.
2953 ** The compiler understands coercion between single-value and
2954 multiple-VALUES type expressions better, getting rid of some very
2955 weird behavior, thanks to patches originally by Robert MacLachlan
2956 and Douglas Crosher.
2957 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2958 lambda lists, thanks to a patch originally by Pierre Mai.
2959 ** The system no longer bogusly warns about "abbreviated type
2961 ** The compiler gets less confused by inlining and RETURN-FROM,
2962 thanks to some patches originally by Tim Moore.
2963 ** The system no longer hangs when dumping circular lists to fasl
2964 files, thanks to a patch originally from Douglas Crosher.
2965 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2966 undefined function error.
2967 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2968 in structure slot initforms) for now, documented workaround instead:-|
2969 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2970 * fixed bug 5: added stubs for various Gray stream functions called
2971 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2972 installed, at least appropriate type errors are generated
2973 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2974 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2975 consistently in DEFMETHOD forms.
2976 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2977 it doesn't seem to affect SBCL after all
2978 * The C runtime system now builds with better optimization and many
2979 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2981 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2983 * The system is now under CVS at SourceForge (instead of the
2984 CVS repository on my home machine).
2985 * The new signal handling code has been tweaked to treat register
2986 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2987 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2988 downstream, e.g. in debug-int.lisp, has implicit dependencies
2989 on the unsignedness of integer representation of machine words,
2990 and that caused the system to bomb out with infinite regress
2991 when trying to recover from type errors involving signed values,
2992 e.g. (BUTLAST '(1 2 3) -1).)
2993 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2994 (This was one of the bugs Peter Van Eynde reported back in July.)
2995 * The system now uses code inspired by Colin Walters' O(N)
2996 implementation of MAP (from the cmucl-imp@cons.org mailing
2997 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2998 inline the MAP operation, and there is more than one
2999 sequence argument to the MAP call (so that it can't just
3000 do ETYPECASE once and for all based on the type of the
3001 single sequence argument). (The old non-inline implementation
3002 of the general M-argument sequence-of-length-N case required
3003 O(M*N*N) time when any of the sequence arguments were LISTs.)
3004 * The QUIT :UNIX-CODE keyword argument has been renamed to
3005 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3006 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3008 * Raymond Wiker's patches to port dynamic loading from Linux to
3009 FreeBSD have been added.
3010 * The BUGS file is now more nearly up to date, thanks in large part
3011 to Martin Atzmueller's review of it.
3012 * The debugger now flushes standard output streams before it begins
3013 its output ("debugger invoked" and so forth).
3014 * The core version number and fasl file version number have both
3015 been incremented, because of incompatible changes in the layout
3017 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3018 used to not be called for a saved Lisp image.)
3019 * Martin Atzmueller's version of a patch to fix a compiler crash,
3020 as posted on sbcl-devel 13 September 2000, has been installed.
3021 * Instead of installing Martin Atzmueller's patch for the
3022 compiler transform for SUBSEQ, I deleted the compiler transform,
3023 and transforms for some similar consing operations.
3024 * A bug in signal handling which kept TRACE from working on OpenBSD
3026 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3027 away by constant folding
3028 * The system now defines its address space constants in one place
3029 (in the Lisp sources), and propagates them automatically elsewhere
3030 (through GENESIS and the sbcl.h file). Therefore, patching the
3031 address map is less unnecessarily tedious and error-prone. The
3032 Lisp names of address space constants have also been systematized.
3033 * CVS tags like dollar-Header-dollar have been removed from
3034 the sources, because they have never saved me trouble and
3035 they've been source of trouble working with patches and other
3036 diff-related operations.
3037 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3038 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3040 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3042 * The system has been ported to OpenBSD.
3043 * The system now compiles with a simple "sh make.sh" on the systems
3044 that it's supported on. I.e., now you no longer need to tweak
3045 text in the target-features.lisp-expr and symlinks in src/runtime/
3046 by hand, the make.sh takes care of it for you.
3047 * The system is no longer so grossly inefficient when compiling code
3048 involving vectors implemented as general (not simple) vectors (VECTOR T),
3049 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3050 which dares to use the various sequence functions on non-simple
3051 vectors, takes less of a performance hit.
3052 * There is now a primitive type predicate VECTOR-T-P
3053 to test for the (VECTOR T) type, so that e.g.
3054 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3055 can now be compiled with some semblance of efficiency. (The old code
3056 turned the type declaration into a full call to %TYPEP at runtime!)
3057 * AREF on (VECTOR T) is still not fast, since it's still compiled
3058 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3059 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3060 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3061 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3062 without making full calls to SUBTYPEP at runtime.
3063 (Some analogous efficiency issues for non-simple vectors specialized to
3064 element types other than T, or for non-simple multidimensional arrays,
3065 have not been addressed. They could almost certainly be handled the
3066 same way if anyone is motivated to do so.)
3067 * The changes in array handling break binary compatibility, so
3068 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3069 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3070 instead of (VALUES T T).
3071 * By following the instructions that Dan Barlow posted to sbcl-devel
3072 on 2 July 2000, I was able to enable primitive dynamic object
3073 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3074 functionality is not implemented (since it calls ld to resolve
3075 library references automatically, requiring RUN-PROGRAM for its
3076 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3077 try to resolve library references) is now supported.
3078 * The system now flushes the standard output streams when it terminates,
3079 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3080 them at several other probably-convenient times, e.g. in each pass of
3081 the toplevel read-eval-print loop, and after evaluating a form given
3082 as an "--eval" command-line option. (These changes were motivated by a
3083 discussion of stream flushing issues on cmucl-imp in August 2000.)
3084 * The source transform for TYPEP of array types no longer assumes
3085 that an array whose element type is a not-yet-defined type
3086 is implemented as an array of T, but instead punts, so that the
3087 type will be interpreted at runtime.
3088 * There is now some support for cross-compiling in make.sh: each of
3089 the phases of make.sh has its own script. (This should be transparent
3090 to people doing ordinary, non-cross-compile builds.)
3091 * Since my laptop doesn't have hundreds of megabytes of memory like
3092 my desktop machine, I became more motivated to do some items on
3093 my to-do list in order to reduce the size of the system a little:
3094 ** Arrange for various needed-only-at-cold-init things to be
3095 uninterned after cold init. To support this, those things have
3096 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3097 symbols with such names are now uninterned after cold init).
3098 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3099 fasl files for cold load.
3100 ** Remove the old compiler structure pooling code (which used to
3101 be conditional on the target feature :SB-ALLOC) completely.
3102 ** Redo the representation of some data in cold init to be more compact.
3103 (I also looked into supporting byte compiled code at bootstrap time,
3104 which would probably reduce the size of the system a lot, but that
3105 looked too complicated, so I punted for now.)
3106 * The maximum signal nesting depth in the src/runtime/ support code has
3107 been reduced from 4096 to 256. (I don't know any reason for the very
3108 large old value. If the new smaller value turns out to break something,
3109 I'll probably just bump it back up.)
3110 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3112 * Many, many bugs reported by Peter Van Eynde have been added to
3113 the BUGS list; some have even been fixed.
3114 * While enabling dynamic object file loading, I tried to make the
3115 code easier to understand, renaming various functions and variables
3116 with less ambiguous names, and changing some function calling
3117 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3118 * While trying to figure out how to do the OpenBSD port, I tried to
3119 clean up some of the code in src/runtime/. In particular, I dropped
3120 support for non-POSIX signal handling, added various comments,
3121 tweaked the code to reduce the number of compilation warnings, and
3122 renamed some files to increase consistency.
3123 * To support the new automatic configuration functionality in make.sh,
3124 the source file target-features.lisp-expr has been replaced with the
3125 source file base-target-features.lisp-expr and the machine-generated
3126 file local-target-features.lisp-expr.
3127 * fixed a stupid quoting error in make.sh so that using CMU CL
3128 "lisp -batch" as cross-compilation host works again
3130 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3132 * DESCRIBE no longer tries to call itself recursively to describe
3133 bound/fbound values, so that it no longer fails on symbols which are
3134 bound to themselves (like keywords, T, and NIL).
3135 * DESCRIBE now works on generic functions.
3136 * The printer now prints less-screwed-up representations of closures
3137 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3138 * A private symbol is used instead of the :EMPTY keyword previously
3139 used to mark empty slots in hash tables. Thus
3140 (DEFVAR *HT* (MAKE-HASH-TABLE))
3141 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3142 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3143 now does what ANSI says that it should. (You can still get
3144 similar noncompliant behavior if bang on the hash table
3145 implementation with all the symbols you get back from
3146 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3147 This breaks binary compatibility, since tests for equality to
3148 :EMPTY are wired into things like the macroexpansion of
3149 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3151 * There's now a minimal placeholder implementation for CL:STEP,
3152 as required by ANSI.
3153 * An obscure bug in the interaction of the normal compiler, the byte
3154 compiler, inlining, and structure predicates has been patched
3155 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3156 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3158 * Missing ordinary arguments in a macro call are now detected even
3159 when the macro lambda list contains &KEY or &REST.
3160 * The debugger no longer complains about encountering the top of the
3161 stack when you type "FRAME 0" to explicitly instruct it to go to
3162 the top of the stack. And it now prints the frame you request even
3163 if it's the current frame (instead of saying "You are here.").
3164 * As specified by ANSI, the system now always prints keywords
3165 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3167 * The default initial SIZE of HASH-TABLEs is now smaller.
3168 * Type information from CLOS class dispatch is now propagated
3169 into DEFMETHOD bodies, so that e.g.
3170 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3172 is now basically equivalent to
3173 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3174 (DECLARE (TYPE SINGLE-FLOAT X))
3176 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3177 operation, without having to do run-time type dispatch.
3178 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3179 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3180 * Since I don't seem to be making big file reorganizations very often
3181 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3182 is overflowing my ability to conveniently back them up), I've finally
3183 checked the system into CVS. (The CVS repository is on my home system,
3184 not at SourceForge -- putting it on SourceForge might come later.)
3185 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3186 high-level GC-NOTIFY-FOO functions send their output. (There's
3187 still very little control of where low-level verbose GC functions
3188 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3189 less than it used to -- the GC-NOTIFY-FOO functions are now under
3190 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3191 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3192 in only one place in the source code, and propagates it automatically
3193 everywhere that it's needed. Thus e.g. when I bump the version from
3194 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3195 * The C source files now include boilerplate legalese and documentation
3196 at the head of each file (just as the Lisp source files already did).
3197 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3198 to his page will be replaced with a link to his new CLiki service.
3200 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3202 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3203 * The build process now looks for GNU make under the default name "gmake",
3204 instead of "make" as it used to. If GNU make is not available as "gmake"
3205 on your system, you can change this default behavior by setting the
3206 GNUMAKE environment variable.
3207 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3208 can build without error under CMU CL.
3210 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3212 * There is now a partial SBCL user manual (with some new text and some
3213 text cribbed from the CMU CL manual).
3214 * The beginnings of a profiler have been added (starting with the
3215 CMU CL profiler and simplifying and cleaning up). Eventually the
3216 main interface should be through the TRACE macro, but for now,
3217 it's still accessed through vaguely CMU-CL-style functions and macros
3218 exported from the package SB-PROFILE.
3219 * Some problems left over from porting CMU CL to the new
3220 cross-compilation bootstrap process have been cleaned up:
3221 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3222 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3223 part when getting rid of the compiler *BACKEND* stuff.)
3224 ** The value of *NULL-TYPE* was screwed up, because it was
3225 being initialized before the type system knew the final
3226 definition of the 'NULL type. This screwed up several key
3227 optimizations in the compiler, causing inefficiency in all sorts
3228 of places. (I found it because I wanted to understand why
3229 GET-INTERNAL-RUN-TIME was consing.)
3230 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3231 PROCLAIM FTYPE information. Unfortunately this broke binary
3232 compatibility again, since now the forms output by DEFGENERIC
3233 to refer to functions which didn't exist in 0.6.3.
3234 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3235 can use the new (as of 0.6.3) transform for SOME into MAP into
3237 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3238 values to (INTEGER 0 1000000), so that the time code will no longer
3239 occasionally get blown up by Linux returning 1000000 microseconds
3240 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3241 its output conform to the ANSI spec. (Alas, this makes its output
3242 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3243 * A full call to MAP NIL with a single sequence argument no longer conses.
3244 * fixes to problems pointed out by Martin Atzmueller:
3245 * The manual page no longer talks about multiprocessing as though
3246 it were currently supported.
3247 * The ILISP support patches have been removed from the distribution,
3248 because as of version 5.10.1, ILISP now supports SBCL without us
3249 having to maintain patches.
3250 * added a modified version of Raymond Toy's recent CMU CL patch for
3251 EQUALP comparison of HASH-TABLE
3253 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3255 * The system still can't cross-compile itself with
3256 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3257 entails), but at least it can compile more of itself that way
3258 than it used to be able to, and various buglets which were uncovered
3259 by trying to cross-compile itself that way have now been fixed.
3260 * This release breaks binary compatibility again. This time
3261 at least I've incremented the FASL file format version to 2, so that the
3262 problem can be detected reliably instead of just causing weird errors.
3263 * various new style warnings:
3264 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3265 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3266 ** using the lexical binding of a variable named in the *FOO* style
3267 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3268 as specified by ANSI.
3269 * *RANDOM-STATE* is no longer automatically initialized from
3270 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3271 default behavior of the system is to repeat its behavior every time
3272 it's run. If you'd like to change this behavior, you can always
3273 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3274 old convention there was no comparably easy way to get the system to
3275 repeat its behavior every time it was run.
3276 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3277 FTYPE declarations has been removed, in favor of their ANSI
3278 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3279 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3280 the types of their sequence arguments can be determined at compile time.
3281 This is done through a new open code expansion for MAP which eliminates
3282 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3283 argument types can be determined at compile time.
3284 * The optimizer now transforms COERCE into an identity operation when it
3285 can prove that the coerced object is already of the correct type. (This
3286 can be a win for machine generated code, including the output of other
3287 optimization transforms, such as the MAP transform above.)
3288 * Credit information has been moved from source file headers into CREDITS.
3289 * Source file headers have been made more standard.
3290 * The CASE macro now compiles without complaining even when it has
3293 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3295 * (Note that the way that the PCL macroexpansions were rewritten
3296 to accommodate the change in DEFGENERIC below breaks binary
3297 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3298 not run under sbcl-0.6.2. Once we get out of alpha releases,
3299 i.e. hit release 1.0.0, we'll probably try to maintain binary
3300 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3301 and sbcl-1.4.4. Until then, however, it might be fairly common
3302 for maintenance releases to break binary compatibility.)
3303 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3305 * The reporting of unhandled signals has been changed to print some
3306 explanatory text as well as the report form. (Previously only
3307 the report form was printed.)
3308 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3309 it defines, so that the compiler no longer issues undefined function
3310 warnings for compiled-but-not-yet-loaded generic functions.
3311 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3312 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3313 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3314 section of the ANSI spec are supported.
3315 * The "" nickname for the "KEYWORD" package has been removed.
3316 The reader still handles symbol tokens which begin with a package marker
3317 as keywords, but it doesn't expose its mechanism for doing so in the
3318 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3319 * The system now issues STYLE-WARNINGs for contradictory TYPE
3320 proclamations. (Warnings for contradictory FTYPE proclamations would
3321 be nice too, but those can't be done usefully unless the type system
3322 is made smarter about FUNCTION types.)
3323 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3324 systematized, so that "*target-*.lisp is supposed to exist only on the
3325 target and imply that there's a related file which exists on the
3326 host, and *host-*.lisp is supposed to exist only on the host and imply
3327 that there's a related file which exists on the target. This involves a
3328 lot of renaming. Hopefully the acute confusion caused by the renaming
3329 will be justified by the reduction in chronic confusion..
3330 ** runtime-type.lisp -> early-target-type.lisp
3331 ** target-type.lisp -> late-target-type.lisp
3332 ** early-host-format.lisp -> early-format.lisp
3333 ** late-host-format.lisp -> late-format.lisp
3334 ** host-error.lisp -> misc-error.lisp
3335 ** early-error.lisp -> early-target-error.lisp
3336 ** late-error.lisp -> late-target-error.lisp
3337 ** host-defboot.lisp -> early-defboot.lisp
3338 ** code/misc.lisp -> code/target-misc.lisp
3339 ** code/host-misc.lisp -> code/misc.lisp
3340 ** code/numbers.lisp -> code/target-numbers.lisp
3341 ** code/early-numbers.lisp -> numbers.lisp
3342 ** early-host-type.lisp -> early-type.lisp
3343 ** late-host-type.lisp -> late-type.lisp
3344 ** host-typep.lisp -> typep.lisp
3345 ** load.lisp -> target-load.lisp
3346 ** host-load.lisp -> load.lisp
3347 ** host-disassem.lisp -> disassem.lisp
3348 ** host-insts.lisp -> insts.lisp
3349 ** byte-comp.lisp -> target-byte-comp.lisp
3350 ** host-byte-comp.lisp -> byte-comp.lisp
3351 ** host-signal.lisp -> signal.lisp
3352 ** host-defstruct.lisp -> defstruct.lisp
3353 ** late-target-type.lisp -> deftypes-for-target.lisp
3354 Furthermore, several other previously target-only files foo.lisp (e.g.
3355 hash-table.lisp and random.lisp) have been split into a target-and-host
3356 foo.lisp file and a target-only target-foo.lisp file, with their key type
3357 definitions in the target-and-host part, so that the cross-compiler will
3358 know more about target types.
3359 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3360 gone away. In their place are various *BACKEND-FOO* variables
3361 corresponding to the slots of the old structure.
3362 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3363 into the target SBCL has been fixed.
3364 * The chill.lisp system for loading cold code into a running SBCL
3366 * Support for the CMU CL "scavenger hook" extension has been removed.
3367 (It was undocumented and unused in the CMU CL sources that SBCL was
3368 derived from, and stale in sbcl-0.6.1.)
3369 * Various errors in the cross-compiler type system were detected
3370 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3371 (enabling various consistency checks). Many of them were fixed,
3372 but some hard problems remain, so the compiler is back to
3373 running without *TYPE-SYSTEM-INITIALIZED* for now.
3374 * As part of the cross-compiler type system cleanup, I implemented
3375 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3376 * I have started adding UNCROSS calls throughout the type system
3377 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3378 on cross-compiler input files will be able to go away, and various
3380 * CONSTANTP now returns true for quoted forms (as explicitly required
3383 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3385 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3386 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3387 (MAKE-LIST -1). In the longer term, it ought to become true
3388 that declarations are assertions even at SAFETY 1. For now, it's not
3389 quite true even at SAFETY 3, but it's at least more nearly true..
3390 (Note that this change seems to increases the size of the system by
3391 O(5%) and to decrease the speed of the compiler by 20% or more.)
3392 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3393 for the problem of printing dozens of lines of distracting information
3394 about low-level system machinery as part of the top stack frame
3395 on entry to the debugger when an undefined function was called.
3396 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3397 is not reset to COMMON-LISP-USER.
3398 * Compilation of stuff related to dyncount.lisp has been made conditional
3399 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3400 smaller. The various dyncount-related symbols have been moved into
3401 a new "SB-DYNCOUNT" package.
3402 * tty-inspect.lisp has been renamed to inspect.lisp.
3403 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3404 feature has gone away. (When we eventually port to other flavors of
3405 libc and/or Unix, we'll try to make the differences between flavors
3406 invisible at the user level.)
3407 * Various other *FEATURES* tags, and/or their associated conditionals,
3408 have been removed if obsolescent, or given better documentation, or
3409 sometimes given more-mnemonic names.
3411 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3413 * tidied up "make.sh" script
3414 * tidied up system directory structure
3415 * better "clean.sh" behavior
3416 * added doc/FOR-CMUCL-DEVELOPERS
3417 * many many small tweaks to output format, e.g. removing possibly-confusing
3418 trailing #\. character in DESCRIBE-INSTANCE
3419 * (EQUALP #\A 'A) no longer signals an error.
3420 * new hashing code, including EQUALP hashing
3421 * tidied up Lisp initialization and toplevel
3422 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3423 * command line argument processing
3424 * added POSIX-GETENV function to deal with Unix-ish environment variables
3425 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3426 terminating SBCL on EOF
3427 * non-verbose GC by default
3428 * There is no more "sbcl" shell script; the sbcl file is now the C
3429 runtime executable (just like CMU CL).
3430 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3432 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3433 symbol and package information as Lisp native symbol and package objects
3434 instead of strings naming symbols and strings naming packages. This way,
3435 whenever packages are renamed (as in warm init), debug information is
3436 transformed along with everything else.
3437 * tweaked the optimization policy declarations which control the building
3438 of SBCL itself. Now, among other things, the system no longer saves
3439 source location debugging information. (This helps two problems at once
3440 by reducing SBCL size and by keeping SBCL from trying to look for its
3441 sources -- which may not exist -- when reporting errors.)
3442 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3443 debugging and testing purposes
3444 * cleaned up printing, making the printer call PRINT-OBJECT for
3445 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3446 methods, giving nearly-ANSI behavior
3447 * converted almost all special variables to use *FOO* naming convention
3448 * deleted PARSE-TIME functionality, since it can be done portably
3449 * moved some files out of cold init into warm init
3450 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3452 * regularized formatting of source files
3453 * added an install.sh script
3454 * fixed ridiculous memory usage of cross-compiler by making
3455 compiler/alloc.lisp not try to do pooling unless it can hook
3456 itself into the GC of the cross-compilation host. Now the system
3457 builds nicely on my old laptop.
3458 * added :SB-ALLOC in target-features.lisp-expr
3459 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3460 was not implemented)
3461 * re-did condition handling and note reporting in the compiler. Notes
3462 are no longer handled by signalling conditions. Style warnings
3463 and warnings are handled more correctly and reported in such a way
3464 that it's easy to find one or the other in your output (so that you
3465 can e.g. figure out which of many problems caused COMPILE-FILE to
3467 * changed the severity of several compiler warnings from full WARNING
3468 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3469 compiler note reporting so that it doesn't use the condition system
3470 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3471 COMPILE-FILE command)
3472 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3473 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3474 no longer supported, since their implementation was deeply intertwingled
3475 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3476 * removed bogus "support" for compiler macros named (SETF FOO), and
3477 removed the compiler macro for SETF INFO (but only after making a fool
3478 of myself on the cmucl-imp mailing list by posting a bogus patch for
3479 DEFINE-COMPILER-MACRO..)
3480 * Compiled files containing forms which have side effects on the Lisp
3481 reader (such as DEFPACKAGE forms) are now handled more correctly.
3482 (Compiler queuing of top level lambdas has been suppressed by setting
3483 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3484 * deleted various currently-unused source files, e.g. gengc.lisp. They
3485 may be added back at some point e.g. when porting to other architectures,
3486 but until they are it's distracting to distribute them and to try to
3488 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3489 there were so many of them they're just distractions, and UNCROSS is
3490 known to be able to handle the current sources
3491 * moved PROFILE functionality into TRACE, so that it will be clear
3492 how the wrapping and unwrapping of functions when you profile them
3493 interacts with the wrapping and unwrapping of functions when you
3494 trace them. (Actually, the functionality isn't there yet, but at least
3495 the interface specification is there. Hopefully, the functionality will
3496 arrive with some future maintenance release.)
3497 * removed host-oops.lisp
3498 * changed signature of QUIT function to allow UNIX-CODE argument
3499 * fixed READ-SEQUENCE bug
3500 * tweaked verbose GC output so that it looks more like the progress
3501 output that ANSI specifies for functions like LOAD
3502 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3503 * added <http://sbcl.sourceforge.com> to the banner information printed by