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)
28 ** bug fix: threads stacks belonging to dead threads are freed by the
29 next exiting thread, no need to gc to collect thread stacks anymore
30 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
31 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
32 ** bug fix: dynamic variable and thread start related gc lossage
33 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
35 * fixed some bugs revealed by Paul Dietz' test suite:
36 ** SUBTYPEP is slightly more accurate on heinously complicated
37 CONS types where some of the members have uncertain (in the
38 NIL, NIL sense) type relationships to each other.
40 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
41 no longer purified unless :PURIFY T is explicitly specified.
42 ** Non-purified cores are significantly smaller than before
44 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
45 * new feature: timers based on Zach Beane's excellent timer package
46 * added support for the following external formats: koi8-u,
47 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
48 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
49 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
50 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
51 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
52 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
53 * incompatible change: a threaded SBCL will no longer revert to
54 non-threaded mode on non-NPTL systems, but refuse to start entirely.
55 * bug fix: interrupts are disabled until startup is complete; no
56 more sigsegvs when receiving a signal to soon
57 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
59 * bug fix: add a workaround for the memory randomization features in
60 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
61 workaround will only be in effect on systems with the proc filesystem
63 * bug fix: printing objects of type HASH-TABLE signals a
64 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
66 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
67 have been running for over 50 days. (reported by Gilbert Baumann)
68 * bug fix: the logic for getting names of functions gets less
69 confused when confronded with alternate-metaclass
70 funcallable-instances. (reported by Cyrus Harmon)
71 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
72 consistent, even on internal alternate-metaclass objects.
73 * bug fix: flush closure information collected by physical
74 environment analysis prepass before the main pass. (bug reported
76 * bug fix: compiler pack phase does not modify a hashtable, which is
77 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
78 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
79 may be expanded. (reported by Alexander <malishev> on c.l.l)
80 * bug fix: no more style-warnings when compiling code using LAST
81 under high-SPEED low-SPACE settings. (reported by David Wallin
83 * fixed some incompatibilities between SBCL's MOP and the MOP
85 ** the METAOBJECT class is now implemented;
86 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
87 STANDARD-OBJECT, as required;
88 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
89 now compatible (as required by VALIDATE-SUPERCLASS); there
90 remains a consistency requirement over the presence or absence
91 of the FUNCTION class in the superclass list of finalized
92 classes; see the manual for more details;
93 ** the :AROUND method for COMPUTE-SLOTS on
94 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
95 requested slot ordering.
97 ** bug fix: parent thread now can be gc'ed even with a live
99 ** bug fix: binding a special with PROGV to no value is not
100 the same as that symbol not having been bound (thanks to
102 ** bug fix: binding specials is thread safe (thanks to
104 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
105 and SB-SPROF do not die with 'no handler for signal XX in
106 interrupt_handle_now(..)' anymore
107 ** bug fix: WITH-TIMEOUT works with multiple running threads
108 * fixed some bugs revealed by Paul Dietz' test suite:
109 ** ENSURE-GENERIC-FUNCTION should take a method class object for
110 the :method-class keyword argument.
112 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
113 * new port: the Solaris operating system on x86 processors is now
114 mostly supported, though some rough edges in the environment
115 remain. (thanks to Daisuke Homma)
116 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
117 space for its heap. (thanks to Thiemo Seufer)
118 * optimization: SBCL on MIPS platforms now supports dynamic-extent
119 closures. (thanks to Thiemo Seufer)
120 * minor incompatible change: eof selects abort in the debugger.
121 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
122 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
124 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
125 deprecated, and will go away in a future revision of SBCL.
126 * minor incompatible change: GC-ON and GC-OFF are no longer
127 implemented with a counter, it does not matter how many times gc
128 is switched on or off
129 * bug fix: discriminating functions for generic function classes
130 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
131 make invalid assumptions about method precedence order. (reported
133 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
134 platforms. (thanks to Thiemo Seufer)
135 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
136 * bug fix: correct pathname printing with printer escaping is on.
137 (thanks to Kevin Reid)
138 * bug fix: complex VOP definitions in "user-space" no longer trigger
139 package locks. (reported by Zach Beane)
140 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
141 for funcallable-instance functions, and is no different from
143 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
144 characters in the current readtable. (reported by Nicholas Neuss)
145 * bug fix: SUBTYPEP on various CONS types returns more a more
146 accurate acknowledgment of its certainty.
147 * optimizations: REMOVE-DUPLICATES now runs in linear time on
148 lists in some cases. This partially fixes bug 384.
149 * flush all standard streams before prompting in the REPL and the
151 * bug fix: signal handling and triggering gc do not conflict
152 directly anymore, in particular a high frequency sb-sprof does
153 not prevent gc from running
154 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
155 approximation for timezone and DST information between the
156 universal time epoch and the smallest negative 32-bit time_t.
157 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
158 year 1899 so long as the encoded time is non-negative.
159 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
160 the correct answer for non-constant rotate amounts when performing
161 an inline 32-bit rotation.
163 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
164 there is only one thread in the session
165 ** bug fix: memory leak for streams created in one thread and
166 written to in another
167 ** bug fix: lockup when compiled with gcc4
168 ** bug fix: race that allows the gc to be triggered when gc is
170 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
171 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
173 ** bug fix: finished threads are gc'ed properly
174 * fixed some bugs revealed by Paul Dietz' test suite:
175 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
176 has arguments with a different set of applicable methods from
177 the orignal arguments.
178 ** The type error thrown by MAP now has a correct expected-type
180 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
181 name a compiled function.
182 ** Three MISC tests where a large bignum was improperly coerced to
183 a float that couldn't represent that bignum during type
184 derivation were fixed.
185 ** SUBTYPEP can now handle types involving the negation of a
186 list-form FUNCTION type.
187 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
188 as (COMPLEX (AND RATIO (NOT FIXNUM))).
189 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
191 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
192 * New feature: Experimental support for bivalent streams: streams
193 opened with :element-type :default now allow character and binary
194 (unsigned-byte 8) I/O
195 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
196 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
198 * bug fix: no more highly sporadic "couldn't check whether ~S is
199 readable" when reading a stream and an interrupt hits in the middle
200 of a select system call
201 * compiler better recognizes complex arrays (reported by Eduardo
203 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
204 for variables on SBCL built with linkage-tables. (reported by Luis
206 * various error reporting improvements.
207 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
208 (Thanks to Thiemo Seufer)
209 * bug fix: nested reader invokations work correctly
210 * bug fix: it is possible to have simultaneous references to foreign
211 code and foreign data with the same name.
213 ** added x86-64 support
214 ** incompatible change: the threading api now works with thread
215 objects instead of thread ids
216 ** bug fix: threads are protected from signals and interruption when
217 starting up or going down
218 ** bug fix: a race where an exiting thread could lose its stack to gc
219 ** bug fix: don't halt on infinite error in threads if possible
220 ** fixed numerous gc deadlocks introduced in the pthread merge
221 ** bug fix: fixed thread safety issues in read and print
222 ** bug fix: debugger doesn't hang on session lock if interrupted at
223 an inappropriate moment
224 ** bug fix: run-program is now thread safe(r)
225 ** bug fix: inner with-recursive-lock no longer releases the mutex
226 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
227 * fixed some bugs revealed by Paul Dietz' test suite:
228 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
229 EXPECTED-TYPE slots filled.
230 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
232 ** MAKE-CONDITION accepts classes as type-designators.
233 ** COMPILE may never return NIL.
234 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
235 range before calling Unix time functions
237 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
238 * numerous signal handling fixes to increase stability
239 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
240 * minor incompatible change: we now correctly canonize default
241 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
242 as per the MOP, rather than the historical (INITARG INITFUNCTION
243 INITFORM). (reported by Bruno Haible)
244 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
245 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
246 for more information.
247 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
248 pathname is a directory pathname.
249 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
250 * dynamic space size on PPC has been increased to 768Mb. (thanks to
252 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
253 :METACLASS argument in addition to a class name. (reported by
254 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
255 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
256 (reported by Bruno Haible; patch from Gerd Moellmann)
257 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
259 * bug fix: more cleanups to the floating point exception handling on
260 x86-64 (thanks to James Knight)
261 * bug fix: the compiler does not try to constant fold calls of
262 COERCE and BIT-* functions when they return freshly constructed
263 objects. (reported by Nikodemus Siivola and Paul Dietz)
264 * optimization: improved the MIPS versions of generic (in the
265 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
266 * optimization: direct conversion of (unsigned-byte 32) to floats on
267 the PowerPC platform.
268 * optimization: structure instances with raw slots now use less
269 memory, and probably show better memory locality. (thanks to
271 * optimization: DYNAMIC-EXTENT declarations for lists and closures
272 are treated as requests for stack allocation on the x86-64,
273 Alpha-32, PPC and SPARC platforms.
274 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
275 the wrong file descriptor; implementation of SOCKET-OPEN-P.
276 (thanks to Tony Martinez)
278 ** gcing a dead thread can no longer lead to lockups
279 ** threads block signals until they are set up properly
280 ** errno is no longer shared by threads
281 ** interrupt-thread restores the eflags register on x86
282 ** fixed some lockups due to gc/thread interaction
283 * fixed some bugs revealed by Paul Dietz' test suite:
284 ** invalid dotted lists no longer raise a read error when
286 ** signal an error if a symbol that names a declaration is used
287 as the name of a type, or vice versa
288 ** allow using the (declare (typespec var*)) abbreviation for
289 (declare (type typespec var*)) with all type specifiers
290 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
291 ** always return NIL from PPRINT-POP when OBJECT is NIL
292 ** don't signal errors when pretty-printing malformed LABELS,
293 FLET or MACROLET forms
294 ** declarations in a DOLIST body are also in scope for the
296 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
298 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
301 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
302 * fixed cross-compiler leakages that prevented building a 32-bit
303 target with a 64-bit host compiler.
304 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
305 opened with :IF-EXISTS :OVERWRITE.
306 * fixed bug 281: error for an invalid qualifier in a short-form method
307 combination method is not signalled until the faulty method is called.
308 * bug fix: iteration variable type inferrer failed to deal with open
309 intervals. (reported by Alan Shields)
310 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
311 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
312 or more runtime options were provided to the sbcl binary.
313 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
314 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
315 * optimization: there's now a fast-path for fixnum arguments in the
316 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
317 * optimization: the code generated on x86-64 is more compact thanks
318 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
319 * bug fix: floating-point exception handling now partly works on
320 x86-64. (Thanks to James Knight)
321 * improvement to the MIPS backend from Thiemo Seufer: C-style
322 64-bit long long arguments and return values to alien functions
324 * fixed some bugs revealed by Paul Dietz' test suite:
325 ** the type-error signalled from WARN has a filled-in DATUM slot.
326 ** the type-error required when a stream is not associated with
327 a file has the stream as its datum.
328 ** type-errors on single-floats on x86-64 no longer have
329 :INVALID-OBJECT as the datum
330 ** the type-errors signalled for invalid function names now have
331 a correct expected type
332 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
333 for structures defined with a :TYPE. Documentation strings for
334 typed structures are no longer immediately discarded
335 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
336 broadcast streams and synonym streams. FILE-LENGTH now also works
337 on broadcast streams.
339 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
340 * incompatible change: the --noprogrammer option, deprecated since
341 version 0.7.5, has been removed. Please use the equivalent
342 --disable-debugger option instead.
343 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
345 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
346 inoperational for a while now) has been completely removed.
347 * null lexical environments are now printed as #<NULL-LEXENV>,
348 significantly reducing the amount of clutter in typical backtraces.
349 * documentation on weak pointers, finalization, and after GC hooks
350 has been added to the manual.
351 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
352 as other specialized array types, is much faster. SUBSEQ and
353 COPY-SEQ on such arrays have also been sped up.
354 * optimization: EQL is now more efficient when at least other argument
355 is known to be of type (OR FIXNUM (NOT NUMBER)).
356 * fixed bug: the runtime is now less vulnerable to changes in the
357 size of the SBCL object on OS X, and virtual memory is reserved for
358 all spaces that need to be at a fixed address.
359 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
360 * fixed bug: finalizers and after GC hooks that cause consing are now
362 * fixed bug: compiler error messages and summaries are now printed to
363 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
364 * fixed inference of the upper bound of an iteration variable.
365 (reported by Rajat Datta).
366 * fixed bug 211e: calling local functions with duplicated constant
367 keyword argument no longer causes a bogus style warning about an
369 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
370 ftype declaration to be disregarded. (reported by Dave Roberts)
371 * fixed bug 373: caused by erronous compilation of references to alien
372 variables in the runtime on ppc/darwin.
373 * fixed bug 376: CONJUGATE type deriver.
374 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
375 (reported by Baughn on #lisp)
376 * a cross-compiler bug on non-x86 platforms has been identified and
377 fixed. (thanks to Bruno Haible)
378 * improvements to the MIPS runtime code for increased stability.
379 (thanks to Thiemo Seufer)
380 * increased the maximimum compact environment size to allow
381 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
382 * partial workaround for bug 135: don't name the function we're
383 calling for hairy cases of EVAL, so as not to accumulate one environment
384 entry per call to EVAL. (reported by Kevin Reid)
385 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
386 * optimization: type testing for non-vector arrays should be faster.
387 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
389 * bug fix: setting 31st element of a bit vector to zero did not work
391 * bug fix: redefining a class definition which failed due to a
392 previous accessor / function clash now works (but see BUGS entry
393 #380 for more problems in this area). (thanks to Zach Beane)
394 * the long form of DEFINE-METHOD-COMBINATION disables method group
395 checking when given a single method group with pattern *. (thanks
397 * on x86 compiler supports stack allocation of results of simple
398 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
399 * support for building and running on Mac OS X 10.4 "Tiger" has been added
400 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
401 * fixed some bugs related to Unicode integration:
402 ** the restarts for recovering from input and output encoding
403 errors only appear when there is in fact such an error to
405 * fixed some bugs revealed by Paul Dietz' test suite:
406 ** MISC.549 and similar: late transformation of unsafe type
407 assertions into derived types caused unexpected code
409 ** SCALE-FLOAT type deriver is less wrong.
410 ** type derivers for EXP, LOG and similar functions compute result
411 types for complex arguments better.
412 ** (MISC.563) CONJUGATE type deriver works for very restricted
414 ** out-of-line type testers for character strings are available.
415 ** EQUAL compiler transform understands specialness of objects
417 ** accessing double-floats stored on the stack now works on x86-64.
418 ** debugger internals could sometimes create invalid lispobjs,
419 resulting in GC crashes.
420 ** MISC.548: type check weakening can convert required type into
422 ** initialization forms for bindings are not in scope of free special
425 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
426 * incompatible change: thread support for non-NPTL systems has
427 been removed - locking is buggy and unreliable. A threaded
428 SBCL build will now warn at startup and refuse to create
429 new threads, unless futex support is detected in the kernel
430 * incompatible change: the top level REPL now has only an ABORT
431 restart associated with it, not TOPLEVEL and ABORT as it used to.
432 TOP and TOPLEVEL are now available as debugger commands for
433 returning to the top level.
434 * incompatible change: forms evaluated in the REPL now use the
435 global optimization policy.
436 * incompatible change: user- and system-initialization files are
437 no longer processed with LOAD, but by READ and EVAL; hence the
438 global optimization policy, startup package, readtable, etc,
440 * The .fasl file format number has been incremented because of
441 various incompatible changes.
442 * internal entry point details and argument counts no longer appear
443 in backtraces unless explicitly requested by setting
444 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
445 * built-in and standard functions no longer have names like "top
446 level local call to FOO".
447 * fixed bug 32: functions defined in non-null lexical environments
448 now have more legible printed representation
449 * fixed bug 33: functions defined in non-null lexical environemnts
450 are now more amenable to inspection by INSPECT.
451 * workaround for bug 354: XEPs no longer appear in backtraces unless
452 explicitly requested.
453 * fixed bug: receiving the signal which results from attempting to
454 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
455 SIGBUS on other *BSDs) on architectures where the C stack is also
456 the Lisp stack (x86 and x86-64) from foreign code no longer leads
457 to debugger confusion or wild instability. (reported by Cheuksan
459 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
460 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
461 (reported by Lutz Euler)
462 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
463 compiled code. (reported by Paul Dietz)
464 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
465 specialized parameter to a method to something that is not TYPEP
466 the specializer is now possible.
467 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
468 face of package deletion.
469 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
470 pathname. (reported by David Tolpin on c.l.l)
471 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
472 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
473 STRING-STREAM and FILE-STREAM, respectively.
474 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
475 than previously; in addition, its test suite is now run on build.
476 (thanks to Luke Gorrie)
477 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
478 * optimization: added a immediate representation for single-floats
480 * fixed some bugs revealed by Paul Dietz' test suite:
481 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
482 ** attempting to create a package with a colliding nickname causes
483 correctable errors to be signalled.
484 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
485 ** several x86-64 backend bugs related to sign-extension of immediate
488 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
489 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
490 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
492 * fixed loading of multiply forward-referenced layouts.
493 (thanks to Cheuksan Wang)
494 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
495 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
496 (see COMPILE-FILE documentation for details of :PRINT option)
497 * fixed bugs 19 and 317: fixed-format floating point printing is
498 more accurate. This also fixes a bug reported by Adam Warner
499 related to the ~@F format directive.
500 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
502 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
503 dispatch table if the from-char is a dispatch macro character.
504 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
505 by the wordsize no longer ignore the last word. (reported by Lutz
507 * fixed bug in type checking of dynamic-extent variables. (reported
509 * optimization: sequence traversal functions use their freedom to
510 coerce function designators to functions.
511 * optimization: code with many calls to CLOS methods specialized on
512 CLOS classes has had redundant type checks removed.
513 * optimization: added declarations to speed up operations that access
514 the internal character database (for example STRING-UPCASE)
515 * optimization: comparison operations between floats and sufficiently small
516 fixnums no longer create extra rationals
517 * fixed some bugs related to Unicode integration:
518 ** portions of multibyte characters at the end of buffers for
519 character-based file input are correctly transferred to the
520 start of the buffer at the next read.
521 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
522 passing it through to OPEN.
523 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
524 argument given to internal calls to OPEN.
525 ** the built-in comment readers (introduced by character sequences
526 ";" and "#|") are more forgiving to encoding errors; they will
527 STYLE-WARN and then attempt to resync the stream at a character
528 boundary. (thanks to Teemu Kalvas)
529 * fixed some bugs in the x86-64 port:
530 ** Negative short int return values from c-calls are sign-extended
532 ** The stack is aligned to 16-bytes for c-calls, as required by
533 the ABI. (thanks to Cheuksan Wang)
534 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
535 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
536 for enums. (thanks to Vincent Arkesteijn)
537 ** Multiple small optimizations and bugfixes for floating point
539 * fixed some bugs revealed by Paul Dietz' test suite:
540 ** Space, Tab, Linefeed, Return and Page have the invalid
541 secondary constituent character trait.
542 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
544 ** WITH-INPUT-FROM-STRING should only update the index place on
546 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
548 ** Bit-array operations (BIT-AND and similar) worked incorrectly
549 with one-dimensional arrays with fill pointers.
550 ** TYPE-OF failed on a complex with an integer realpart and a
552 ** compiler failure during type inference for the code of form
553 (IF (EQL X (THE ...)) ...) (MISC.535).
555 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
556 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
557 platform. The port supports 61-bit fixnums, large memory spaces
558 and reloading shared object files.
559 * enhancement: saving cores with foreign code loaded is now
560 supported on ppc/Darwin in addition to the previously supported
562 * enhancement: the statistical profiler now walks deeper into the
563 call stack for more meaningful call-graphs and accrued time
564 reports (x86/x86-64 only). It also now reports time spent in
566 * enhancement: it is now possible to trace most individual methods
567 of a generic function in addition to tracing the generic function
569 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
571 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
572 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
573 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
574 produces an error. (thanks to Vincent Arkesteijn)
575 * bug fix: NAMESTRING on pathnames with :WILD components in their
576 directories works correctly. (thanks to Artem V. Andreev)
577 * fixed bug 125: compiler preserves identity of closures. (reported
579 * bug fixed: functions with &REST arguments sometimes failed with
580 "Undefined function" when compiled with (DEBUG 3). (reported by
582 * bug fix: overflow during compiling of setting element of a bit
583 vector with constant index and value. (reported by Timmy Douglas)
584 * build fix: fixed the dependence on *LOAD-PATHNAME* and
585 *COMPILE-FILE-PATHNAME* being absolute pathnames.
586 * on x86 compiler partially supports stack allocation of dynamic-extent
588 * GO and RETURN-FROM do not check the extent of their exit points
589 when compiled with SAFETY 0.
590 * fixed some bugs related to Unicode integration:
591 ** encoding and decoding errors are now much more robustly
592 handled; it should now be possible to recover even from invalid
593 input or output to the terminal. (thanks to Teemu Kalvas)
594 ** provided a first cut at implementing STRING-TO-OCTETS and
595 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
596 ** altered the SB-MD5 contributed module slightly, changing the
597 interface just enough for it to be supportable for builds where
598 lisp characters are not eight bits.
599 * fixed some bugs revealed by Paul Dietz' test suite:
600 ** the FORMATTER-generated functions for ~V[ conditionals require
601 the correct number of arguments.
602 ** READ-FROM-STRING returns the mandated second value when applied
603 to displaced strings.
604 ** the #\Rubout and #\Backspace characters are treated as invalid
605 constituent characters by the tokenizer.
607 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
608 * new feature: reloading changed shared object files with
609 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
610 * new feature: references to foreign variables and functions
611 can now be compiled and loaded before the corresponding shared
612 object file is loaded, as long as the foreign definitions are
613 available at runtime.
614 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
615 supported. (thanks to Dan Debertin)
616 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
617 just keywords. (thanks to Vincent Arkesteijn)
618 * fixed bug #331: structure-class instances corresponding to
619 DEFSTRUCT forms are now created eagerly.
620 * fixed bug #345: backtraces from calls to undefined functions work
621 on x86 as well. Related bug #61 is now also partially fixed on x86
622 (backtraces from throws to unknown catch tags.)
623 * bug fix: lambda-list parsing is now stricter vrt. order and number
624 of lambda-list keywords.
625 * bug fix: as specified by AMOP, an error is signalled if a
626 class-option appears multiple times in a DEFCLASS form. (reported
628 * bug fix: RANDOM can be compiled when the compiler derives the type
629 of its numeric argument as a disjoint set of small integers.
630 (reported by Paul Dietz)
631 * bug fix: starting a core saved with shared objects loaded when
632 those objects are not available no longer causes threaded SBCL to
633 hang. (reported by Sean Ross)
634 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
636 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
637 argument, which must always be NIL. (reported by Kalle Niemitalo)
638 * bug fix: printing 1.0d+23 no longer results in an error.
639 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
640 * bug fix: structure slot setters preserve evaluation order. (thanks
642 * fixed some bugs related to Unicode integration:
643 ** RUN-PROGRAM can allow its child to take input from a Lisp
644 stream. (reported by Stefan Scholl)
645 ** ASDF-INSTALL successfully downloads .tar.gz files in all
646 locales. (reported by Ken Causey)
647 * fixed some bugs revealed by Paul Dietz' test suite:
648 ** INCF, DECF and REMF evaluate their place form as specified in
650 ** FORMATTER expands ~{ ~} iteration directives with V or #
651 parameters correctly.
652 ** FORMATTER deals with the ~@[ ~] conditional directive where the
653 consequent uses no arguments correctly.
654 ** the system has a partial understanding of the (COMPLEX RATIO)
656 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
659 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
660 * new feature: a build-time option (controlled by the :SB-UNICODE
661 keyword feature, enabled by default) for building the system with
662 support for the entire 21-bit character space defined by the
664 * new feature: the system now has rudimentary external-format
665 support; the primary user-visible change at this time is that
666 characters with the high bit set (such as lower-case-e-acute) will
667 print correctly to a terminal in a UTF-8 environment.
668 * minor incompatible change: BASE-CHAR no longer names a class;
669 however, CHARACTER continues to do so, as required by ANSI.
670 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
671 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
672 has been moved to the SB-EXT package (temporarily re-exported from
674 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
675 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
677 * fixed bug #351: better error-handlind and reporting for malformed
679 * fixed bug #350: bignum-printing is now more memory-efficient,
680 allowing printing of very large bignums, eg. (expt 2 10000000).
681 (reported by Bruno Haible)
682 * fixed bug #302: better primitive-type selection for intersection
684 * fixed bug #308: non-graphic characters now all have names, as
685 required. (reported by Bruno Haible)
686 * bug fix: redefining a class with different superclasses now correctly
687 removes it from the direct-subclasses of its previous superclasses.
688 (reported by David Morse)
689 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
690 new value now works. (reported by Bruno Haible)
691 * bug fix: correct canonicalization of multiple non-standard slot
692 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
693 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
694 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
695 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
696 now exists, an signals an error.
697 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
698 during macroexpansion for non-iteration variables as well. (reported
699 by Bruno Haible for CMUCL)
700 * bug fix: Cyclic structures and unprintable objects in compiler
701 messages no longer cause errors. (reported by Bruno Haible)
702 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
703 and READ-FROM-STRING all now return a primary value of NIL if
704 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
705 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
706 (reported by Bruno Haible for CMUCL)
707 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
708 do not designate a condition. (reported by Bruno Haible for
710 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
711 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
712 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
713 permitting resolution in favour of any of the conflicting symbols.
714 (reported by Bruno Haible for CMUCL)
715 * bug fix: EQUAL compiler optimizations is less aggressive on
716 strings which can potentially compare true despite having distinct
717 specialized array element types.
718 * bug fix: unit enumerations can be defined without dividing by
719 zero. (thanks to Vincent Arkesteijn)
720 * FORMAT compile-time argument count checking has been enhanced.
721 (report from Bruno Haible for CMUCL)
722 * a partial workaround for the bug 262: the compiler does not try to
723 inline-expand a local function doing RETURN-FROM from a deleted
724 BLOCK. (thanks to Peter Denno for the bug report and to David
725 Wragg for the simple test case)
726 * fixed some bugs revealed by Paul Dietz' test suite:
727 ** RENAME-PACKAGE allows all package designators as new package
729 ** constraint propagation and lambda variable substitution are
730 more cautious in dealing with partially deleted code.
731 ** compiler sometimes lost reoptimization passes.
732 ** CERROR, given a condition as condition designator, passes its
733 remaining arguments to the continue format control without
735 ** Case-altering FORMAT directives work correctly on non-ASCII
737 ** The REAL type specifier handles bounds outside the floating
738 point ranges without signalling FLOATING-POINT-OVERFLOW.
739 ** Functions with IR1-transformations can create intercomponent
740 references to global functions.
741 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
743 ** FORMAT ~R treats a nil value for its first parameter correctly.
745 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
746 * enhancement: saving cores with foreign code loaded is now
747 supported on x86/NetBSD and sparc/Linux in addition to the previously
749 * bug fix: on some platforms repeated installations caused multiple
750 copies of HTML documentation to be installed -- should not happen
751 any more. (reported by Stefan Scholl)
752 * bug fix: parsing self-recursive alien record types multiple times
753 no longer causes infinite recursion. (reported by Thomas F. Burdick,
754 original patch by Helmut Eller for CMUCL)
755 * bug fix: stack-exhaustion detection works now on NetBSD as well.
756 (thanks to Richard Kreuter)
757 * bug fix: defining classes whose accessors are methods on existing
758 generic functions in other (locked) packages no longer signals
759 bogus package lock violations. (reported by François-René Rideau)
760 * bug fix: special variables as DEFMETHOD parameters no longer have
761 associated bogus type declarations. (reported by David Wragg and
763 * bug fix: read-write consistency on streams of element-type
764 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
765 * bug fix: redefiniton of the only method of a generic function with
766 no DEFGENERIC no longer emits a full WARNING. In addition,
767 redefinition of generic functions with no DEFGENERIC to an
768 incompatible lambda list now signals an error. (thanks to Zach
770 * bug fix: DEFGENERIC now works even when there's a function of the
771 same name in an enclosing lexical environment. (thanks to Zach
773 * fixed compiler failure, caused by instrumenting code during
774 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
775 * optimization: added loop analysis and improved register allocation
776 * fixed some bugs revealed by Paul Dietz' test suite:
777 ** POSITION on displaced vectors with non-zero displacement
778 returns the right answer.
779 ** (SIMPLE-STRING) is a valid type specifier for sequence
781 ** *PRINT-LEVEL* handling for slotless structures is pedantically
783 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
784 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
786 ** The justification version of the FORMAT ~< directive treats
787 non-zero minpad parameter correctly.
789 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
790 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
791 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
792 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
793 the supported interface.
794 * new feature: Single-stepping of code compiled with DEBUG 2 or
795 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
796 * new feature: saving cores with foreign code loaded is now
797 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
798 Timothy Moore's work for CMUCL)
799 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
800 parameters to * instead of NIL if no initform is supplied.
801 (reported by Johan Bockgaard)
802 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
803 correctly when FUNCALL appears as the car of the form. Note:
804 despite this FUNCALL forms are not currently subject to
805 compiler-macro expansion. (port of Raymond Toy's fix for the
806 same from CMUCL, reported by Johan Bockgaard)
807 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
808 (thanks for Teemu Kalvas)
809 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
810 case instead of terminating the iteration (thanks for Julian
811 Squires, Sean Champ and Raymond Toy)
812 * bug fix: incorrect expansion of defgeneric that caused a style
813 warning. (thanks for Zach Beane)
814 * bug fix: slot accessor effective method computation works properly
815 for classes with multiple non-standard applicable methods on
816 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
817 * on x86 compiler supports stack allocation of results of LIST and
818 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
819 CMUCL implementation by Gerd Moellmann)
820 * fixed some bugs revealed by Paul Dietz' test suite:
821 ** FORMAT strings with both the ~<~:;~> form of the justification
822 directive and pretty-printing directives cause an error.
824 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
825 * incompatible change: the internal functions
826 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
827 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
828 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
829 instead of the old functions.
830 * new feature: on platforms where "dladdr" is available foreign
831 function names now appear in backtraces. (based on Helmut Eller's
833 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
834 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
836 * documentation: ASDF manual is now distributed with SBCL.
837 * bug fix: SBCL can now load its contributed modules with REQUIRE
838 even if the system-provided entries have been removed from
839 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
841 * bug fix: backtraces involving undefined functions or assembly
842 routines are more informative. (thanks to Brian Downing)
843 * bug fix: mutually referent alien structures now work correctly.
844 (reported by Rick Taube)
845 * bug fix: structures defined by WITH-ALIEN can be referred to
846 within other definitions in the same WITH-ALIEN.
847 * bug fix: division operators (MOD, TRUNCATE and the like) with
848 constant zero divisors and integer dividends no longer generate
850 * bug fix: provide default methods for INPUT-STREAM-P and
851 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
852 * bug fix: improve the detection and resolution of MOP metacycles
853 (where effective method computation depends on the generic function
854 having its effective method computed). (reported by Bruno Haible)
855 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
856 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
857 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
858 * bug fix: pathnames with relative directory components can be
859 represented relative to default pathnames.
860 * optimization: in taking the GCD of bignums, reduce the two bignums
861 to approximately the same size (using Euclid's algorithm) before
862 applying the more sophisticated binary GCD. (thanks to Juho
864 * optimization: COUNT on bitvectors now operates word-at-a-time.
865 * optimization: ASH with a positive, but not necessarily constant,
866 (leftwards) shift, when in a modular context, is compiled to a
868 * fixed some bugs revealed by Paul Dietz' test suite:
869 ** FORMAT variable parameters ("~V<char>") are defaulted properly
870 if the corresponding argument is NIL.
871 ** FORMAT directives accepting a minpad parameter treat negative
873 ** literal commas as character parameters to FORMAT directives are
875 ** literal spaces directly after ~<Newline> directives within a
876 format-logical-block (~:< ~@:>) do not induce :FILL-style
877 conditional newlines.
878 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
879 as specified: it no longer includes conditional newlines).
880 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
882 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
883 :PER-LINE-PREFIX argument does not evaluate to a string.
885 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
886 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
887 the manual for details; for now, package locks can be disabled by
888 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
889 if no major problems are found then it is likely that they will be
890 compiled in unconditionally.
891 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
892 unsupported operators on all platforms. To load a shared library
893 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
894 object file, link it into a shared library outside of SBCL and
895 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
897 * bug workaround: ROOM T and the GC were not working together
898 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
899 reporting facility aren't preserved. That reporting has been
900 disabled, so now until and unless someone figures out how to make it
901 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
902 * minor incompatible change: as threatened around sbcl-0.8.0, the
903 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
904 The symbols are also exported from SB-PCL for backwards
905 compatibility, but more so than before SB-PCL should be treated as
906 an implementation-internal package.
907 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
909 * the SB-SPROF contrib now works on (most) non-x86 architectures.
910 It is known as of this release not to work on the Alpha, however.
911 * fixed bug #167: errors signalled due to illegal syntax in method
912 bodies are now more legible.
913 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
914 designators and can hence be used with TYPEP.
915 * fixed bug #333: CHECK-TYPE now ensures that the type error
916 signalled, if any, has the right object to be accessed by
917 TYPE-ERROR-DATUM. (reported by Tony Martinez)
918 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
919 5.1.2.3 for multiple-value place subforms. (reported by Kalle
921 * fixed bug #334: programmatic addition of slots using specialized
922 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
923 and :ALLOCATION :CLASS slots.
924 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
925 float exponents. (rereported by Peter Seibel)
926 * fixed a bug: #\Space (and other whitespace characters) are no
927 longer considered to be macro characters in standard syntax by
929 * fixed bug: initialization of condition class metaobjects no longer
930 causes an instance of the condition to be created. (reported by
932 * fixed bug: it is now possible to have slots such that
933 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
934 description is neither :INSTANCE nor :CLASS.
935 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
936 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
937 SLOT-VALUE-USING-CLASS), and no longer causes errors with
938 non-standard SLOT-DEFINITION-ALLOCATION values.
939 * fixed bugs: various race conditions handling exiting threads.
940 CL-PPCRE's thread torture test now passes.
941 * fixed arguably-a-bug: GC time proportional to number of threads in
942 system even when most of them are idle
943 * optimization: improved performance of BIT and SBIT on bit-vectors.
944 * .fasl file incompatibility: The fasl file version number has
945 been incremented because of changes associated with package locks.
947 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
948 * minor incompatible change: the system no longer provides
949 optimization hints (conditions which are TYPEP
950 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
951 mode; these hints are still emitted when higher SPEED optimization
953 * new contrib module: a sampling profiler (profiling by statistical
954 sampling, rather than by instrumenting functions) is available as
955 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
956 * the behaviour of the standard function ED is now customizeable by
957 third parties through a hook variable: see ED's documentation
958 string for information on the protocol.
959 * the compiler no longer emits efficiency notes for (FUNCALL X)
960 when the type of X is uncertain under default optimization
962 * fixed bug 276: mutating a binding of a specialized parameter to a
963 method to something that is not TYPEP the specializer is now
965 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
966 exhibit the required behaviour on the x86 platform. (thanks to
967 Peter van Eynde, Eric Marsden and Bruno Haible)
968 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
969 even for difficult arguments. (reported by Peter Graves)
970 * fixed bug 141a: the backquote printer now descends quoted
972 * fixed another bug in backquote printing: no more destructive
973 modification of the form's list structure. (reported by Brian
975 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
976 move between its address being taken and the call to
977 interrupt_thread, fixing a crashing race condition.
978 * the SB-POSIX contrib implementation has been adjusted so that it
979 no longer exhibits ridiculously poor performance when constructing
980 instances corresponding to C structs.
982 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
983 * minor incompatible change: the sb-grovel contrib now treats C
984 structures as alien (in the sense of SB-ALIEN) objects rather than
985 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
986 has implications for memory management of client code
987 (sb-grovel-returned objects must now be manually managed) and for
988 type safety (alien objects now have full types).
989 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
990 used to control emission of compiler diagnostics, rather than the
991 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
992 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
993 quality should be considered deprecated.
994 * install.sh now installs the user manual as well
995 * (not quite a new documentable feature, but worth considering in
996 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
997 beginnings of a semantically meaningful condition hierarchy is
998 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1000 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1001 designator as the defaults argument.
1002 * fixed bug: Displaced arrays whose displaced-to array has become
1003 too small now cause ARRAY-DIMENSION to signal an error, providing
1004 for safer bounds-checking. (reported by Bruno Haible)
1005 * fixed bug: DEFCLASS slot definitions with identical :READER and
1006 :WRITER names now signal a reasonable error. (reported by Thomas
1008 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1010 * fixed bug: Invalid slot specification errors now print correctly.
1011 (thanks to Zach Beane)
1012 * fixed bug 320: Shared to local slot value transfers in class
1013 redefinitions now happen corrently from superclasses as
1014 well. (reported by Bruno Haible)
1015 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1017 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1018 improper lists correctly. (reported by Bruno Haible)
1019 * fixed bug 313: source-transform for <fun-name> was erroneously
1020 applied to a call of a value of a variable with name <fun-name>.
1021 (reported by Antonio Menezes Leitao)
1022 * fixed bug 307: The obsolete instance protocol ensures that
1023 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1024 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1025 remove dead unknown-values globs from the middle of the stack.
1026 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1027 easier to resume long computations after using *BREAK-ON-SIGNALS*
1028 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1029 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1030 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1031 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1032 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1033 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1035 * fixed a (fairly theoretical) bug in string printing: if
1036 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1037 does not have array-element-type equal to the most general string
1039 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1040 function name. (thanks to Nikodemus Siivola)
1041 * fixed bug: compile-time format string checker failed on
1042 non-closed ~{. (reported by Thomas F Burdick)
1043 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1044 don't warn on legal loop code involving more than one aggregate
1045 boolean. (thanks to Nikodemus Siivola)
1046 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1047 in keyword position to MAKE-PACKAGE was not being reported as
1049 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1050 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1052 * fixed bugs 280 and 312: the checking for multiple definitions in a
1053 file is less likely to become confused by uses of inline
1055 * fixed bug: the #S reader macro performs the keyword coercion
1056 specified for slot names. (reported by Kalle Niemitalo)
1057 * fixed bug: lambda lists may contain symbols whose names start with
1058 & but are not lambda-list-keywords; their occurrence triggers a
1060 * fixed bug 321: define-method-combination argument lambda lists do
1061 not cause invalid code to be generated when &optional or &aux
1062 variables have default values. (reported by Bruno Haible)
1063 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1064 have CLOS classes; this appears also to have fixed reported
1065 instability in stack exhaustion detection.
1066 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1067 MUFFLE-WARNING when no associated restart is present is now
1069 * optimization: rearranged the expansion of various defining macros
1070 so that each expands into only one top-level form in a
1071 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1073 * optimization: used a previously little-used slot in symbols to
1074 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1076 * fixed some bugs revealed by Paul Dietz' test suite:
1077 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1079 ** Fixed an optimization bug related to inheritance of initforms
1080 from local to shared slots.
1081 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1082 ** CAST optimizer forgot to flush argument derived type.
1083 ** print/read consistency on floats is now orders of magnitude
1084 more likely. (thanks also to Bruno Haible for a similar report
1086 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1087 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1088 for objects of type REAL. Make it so.
1089 ** SXHASH is specified (infelicitously) to respect similarity,
1090 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1091 it so. (thanks to Markus Ziegler)
1092 ** on the Alpha, the compiler succeeds in compiling functions
1093 returning a known number of arguments greater than 63.
1094 ** fixed handling of invalid NIL arguments in keyword position
1096 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1098 ** provided a readably-printable representation for RANDOM-STATE
1100 ** ensured that pathnames, where they have a namestring, always
1101 print using #P"..." syntax.
1103 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1104 * Support for the forthcoming 2.0 version of the NetBSD kernel
1105 running on x86 hardware has been added. (thanks to Perry
1106 E. Metzger most immediately, and others for their past work)
1107 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1108 current release version; SBCL's OpenBSD support had been broken
1109 since about the time of OpenBSD's switch to ELF binary format.)
1110 * [placeholder for DX summary]
1111 ** user code with &REST lists declared dynamic-extent, under high
1112 speed or space and low safety and debug optimization policy.
1113 * The manual has been converted to Texinfo format and the debugger
1114 chapter from the cmucl manual has been added.
1115 * A facility has been added to extract documentation strings from
1116 sbcl and store them as Texinfo-formatted snippets for inclusion in
1117 the manual (via Texinfo's @include directive)
1118 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1119 the test case to Dave Roberts)
1120 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1121 pointers. (reported by Sean Ross)
1122 * bug fix: PROFILE output is printed nicely even for large numerical
1123 values. (thanks to Zach Beane)
1124 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1125 greater than 32 handle EOF correctly.
1126 * bug fix: on X86 an immediate argument of the IMUL instruction is
1127 correctly printed in disassembly. (thanks to Lutz Euler)
1128 * bug fix: class slots in redefined classes preserve their old
1129 values. (thanks to Bruno Haible and Nikodemus Siivola)
1130 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1131 succeeds. (reported by Marco Baringer)
1132 * fixed some bugs revealed by Paul Dietz' test suite:
1133 ** READ-SEQUENCE now works on ECHO-STREAMs.
1134 ** RATIONALIZE works more according to its specification. (thanks
1137 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1138 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1139 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1140 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1141 should matter to you only if you rebind the printer control
1142 variables and then find you want different bindings in the
1143 debugger than in the ordinary execution of your program.)
1144 * The runtime build system has been tweaked to support building
1145 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1146 assembler and linker. (thanks to Nikodemus Siivola)
1147 * Unbound, undefined, undeclared variables now trigger full
1148 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1149 more often programmer error than deliberate exploitation of undefined
1151 * optimization: the hash algorithm for strings has changed to one
1152 that is less vulnerable to spurious collisions. (thanks to Juho
1154 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1155 needless bounds checking. (thanks to Juho Snellman)
1156 * optimization: implemented multiplication as a modular
1157 (UNSIGNED-BYTE 32) operation on the PPC backend.
1158 * fixed some bugs revealed by Paul Dietz' test suite:
1159 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1161 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1162 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1163 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1165 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1167 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1169 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1170 their output stream on EOF from read.
1171 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1172 have been read to end-of-file.
1173 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1175 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1176 description of determination of which consecutive characters
1178 ** Printing the "Space" character with escaping on now yields "#\\ ",
1179 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1180 ** Reading floating-point numbers with *READ-BASE* set to a number
1181 less than 10 works correctly.
1182 ** Reading floating-point numbers with *READ-BASE* set to a number
1183 more than 10 works correctly.
1184 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1185 the readtable currently in effect.
1187 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1188 * minor incompatible change: parsing of namestrings on a physical
1189 (Unix) host has changed; numbers after the final #\. in a
1190 namestring are no longer interpreted as a version field. This is
1191 intented to be largely invisible to the user, except that the
1192 meaning of the namestring "*.*.*" has changed: it now refers to a
1193 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1194 should usually be replaced by
1195 (make-pathname :name :wild :type :wild :version :wild)
1196 with the added benefit that this is more likely to be portable.
1197 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1198 option to OPEN now signals an error if the file being opened
1199 exists; this may have an impact on existing code.
1200 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1201 signals received, on the PowerPC platforms (both Linux and
1202 Darwin). (thanks to Pierre Mai for pointing out the location of
1204 * several fixes on OS X: The system now builds and runs cleanly on
1205 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1206 (thanks to Brian Mastenbrook)
1207 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1208 second-resolution: integer multiples of 1/3600 between -24 and 24.
1209 (thanks to Vincent Arkesteijn)
1210 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1211 type when called with 1 argument; PEEK-CHAR checked type of
1212 PEEK-TYPE only after having read first character from a
1213 stream. (reported by Peter Graves)
1214 * bug fix: the garbage collector now has much better locality
1215 behaviour, and in particular no longer treats all memory as being
1216 exhausted when many small objects point to each other in a deeply
1218 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1219 recognized as being TYPEP their class.
1220 * bug fix: the PUSHNEW documentation string has been corrected.
1221 (thanks to Vincent Arkesteijn)
1222 * bug fix: defaulting of the value for the last of an atypically
1223 large number of multiple values being bound was not being
1224 performed correctly on the Alpha or PPC platforms
1225 * optimization: implemented multiplication as a modular
1226 (UNSIGNED-BYTE 32) operation on the x86 backend.
1227 * optimization: SEARCH on simple-base-strings can now be open-coded.
1228 (see also contrib/compiler-extras.lisp for inspiration for
1229 teaching the compiler about the Boyer-Moore algorithm).
1230 * value, returned by MAX (and MIN) called with several EQUALP, but
1231 not EQL, arguments now does not depend on compiler settings.
1232 * fixed some bugs revealed by Paul Dietz' test suite:
1233 ** in stack analysis liveness information is propagated from
1234 non-local entry points.
1235 ** pathwise CAST removing failed when the CAST node did not start
1237 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1238 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1240 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1241 host is already defined.
1242 ** RENAME-FILE works on streams instead of signalling an internal
1244 ** PEEK-CHAR uses the current readtable when determining whether
1245 or not a character is whitespace.
1246 ** MERGE-PATHNAMES handles the case when the pathname does not
1247 specify a name while the default-pathname specifies a version.
1248 ** Pathnames now stand a better chance of respecting print/read
1250 ** Attempting to use standardized file system operators with a
1251 pathname with invalid :DIRECTORY components signals a
1253 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1255 ** DIRECTORY on logical pathnames is more correct.
1256 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1257 signal a TYPE-ERROR if their argument is not a stream.
1258 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1259 designator argument does not designate a stream.
1260 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1261 examining the synonym.
1262 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1264 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1265 element-type larger than ([UN]SIGNED-BYTE 32).
1267 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1268 * When built with the :SB-FUTEX feature, threaded builds now take
1269 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1270 for faster/more reliable mutex and condition variable support.
1271 * Incompatible change (but one you probably shouldn't have been using
1272 anyway): the interface and code for arbitrating between multiple
1273 threads in the same user session has been redesigned.
1274 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1275 error when called without an explicit environment argument.
1276 (thanks to Nikodemus Siivola)
1277 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1278 confused about their position. (thanks to Adam Warner and Gerd
1280 * bug fix: Pretty printing backquoted forms with unquotations in the
1281 argument list position of various code constructs such as LAMBDA
1282 now works correctly. (reported by Paul Dietz)
1283 * bug fix: Pretty printing unquotations no longer loses all
1284 stream position information.
1285 * optimization: performance of string output streams is now less
1286 poor for multiple small sequence writes.
1287 * optimization: performance of CSUBTYPEP in the presence of complex
1288 expressions involving CONS and NOT many times has been improved.
1289 (reported by Paul Dietz)
1290 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1292 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1293 streams for the REPL, for the debugger, and so on. (thanks to
1295 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1296 * fixed some bugs revealed by Paul Dietz' test suite:
1297 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1298 is ignored if it is too small, rather than propagating through
1299 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1300 ** extremely complex negations of CONS types were not being
1301 sufficiently canonicalized, leading to inconsistencies in
1303 ** VALUES tranformer lost derived type.
1305 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1306 * fixed a bootstrapping bug: the build process no longer assumes
1307 that the various BOOLE-related constants have the same value in
1308 host and target lisps. (noted by Paul Dietz' test suite on an
1309 SBCL binary built from CLISP)
1310 * The system can now be dynamically linked on the MIPS platform,
1311 which enables dynamic loading of foreign code from Lisp. (thanks
1312 to Ralf Baechle for discussions on the MIPS ABI)
1313 * The system now records debugging information for its own source
1314 files in a filesystem-position-independent manner, relative to
1315 the "SYS" logical host.
1316 * fixed a compiler bug: MV-LET convertion did not check references
1317 to the "max args" entry point. (reported by Brian Downing)
1318 * tweaked disassembly notes to be less confident about proclaiming
1319 some instruction as an LRA. (thanks to Brian Downing)
1320 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1321 now each have their own history, command character, and other
1322 characteristics. (thanks to David Lichteblau)
1323 * fixed some bugs revealed by Paul Dietz' test suite:
1324 ** compiler failure in compiling LOGAND expressions including a
1326 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1327 shift greater than 32.
1328 ** FLUSH-DEST did not mark blocks for type check regeneration.
1329 ** HANDLER-CASE failed to accept declarations in handler clauses
1330 in some circumstances.
1332 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1333 * New code in contrib/sb-introspect (still probably not entirely
1334 stable yet) provides some support for smart Lisp development
1335 environments like SLIME.
1336 * The conditions signalled for errors occurring when loading .fasl
1337 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1338 in a way which should help ASDF recover gracefully.
1339 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1340 stuff has been cleaned up. If you code contrib/ stuff, this might
1341 affect you, and you can look at contrib/README, contrib/STANDARDS,
1342 and/or the 0.8.4.27 diff to check.
1343 * In full calls the compiler now does not generate checks for declared
1344 argument types for all arguments.
1345 * various threading fixes
1346 ** and some experimental patches which didn't make it into
1347 the main tree for this release, but which are shipped in
1348 contrib/experimental-thread.patch as a possible fix for some
1349 failures (deadlock, spinning...) in GC-intensive multithreaded
1351 * fixed PPC build problem (source code incompatibility of different
1352 library versions): added offsetof() hackery which attempts to divine
1353 where glibc maintainers put uc_mcontext today
1354 * fixed bug 282: compiler does not trust type assertions while passing
1355 arguments to a full call.
1356 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1357 &OPTIONAL VALUES type parameter.
1358 * fix bug 214: algorithm for noting rejected templates is now more
1359 similar to that of template seletion. (also reported by rydis on
1361 * fixed bug 141b: printing backquoted information readably and prettily
1362 inserts a space where necessary.
1363 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1364 (CHARACTER 10) are now reported as errors, rather than propagated
1365 as unknown types. (reported by piso on #lisp)
1366 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1367 with values NIL and :ERROR. (thanks to Milan Zamazal)
1368 * fixed bug 191c: CLOS now does proper keyword argument checking as
1369 described in CLHS 7.6.5 and 7.6.5.1.
1370 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1371 longer raise an error; further, using a list as a for-as-arithmetic
1372 counter now raises a meaningful error.
1373 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1374 now understood by sequence creation functions such as MAKE-SEQUENCE
1376 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1377 a STRING-INPUT-STREAM.
1378 * compiler enhancement: SIGNUM is now better able to derive the type
1380 * type declarations inside WITH-SLOTS are checked. (reported by
1382 * fixed some bugs revealed by Paul Dietz' test suite:
1383 ** incorrect optimization of TRUNCATE for a positive first
1384 argument and negative second.
1385 ** compiler failure in let-convertion during flushing dead code.
1386 ** compiler failure while deriving type of TRUNCATE on an
1387 interval, containing 0.
1388 ** ASH of a negative bignum by a negative bignum count now returns
1390 ** intersection of CONS types now canonicalizes properly, fixing
1391 inconsistencies in SUBTYPEP.
1393 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1394 * incompatible change: The --disable-debugger command line
1395 option now clobbers the debugger at a more fundamental
1396 level, by redefining #'INVOKE-DEBUGGER instead of by
1397 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1398 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1399 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1400 enter the debugger REPL and then suffer recursive errors
1401 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1402 while under the new system, BREAK in --disable-debugger mode
1403 terminates the system just as an unhandled error would.
1404 * fixed compiler performance when processing loops with a step >1;
1405 * bug fix: DOCUMENTATION now retrieves generic function
1406 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1407 support has been systematized, and now supports the methods
1408 specified by ANSI, along with a default method and a method for
1409 slot documentation. (reported by Nathan Froyd)
1410 * bug fix: effective methods associated with a generic function are
1411 no longer cached over a change of that generic function's method
1412 combination. (reported by Andreas Fuchs)
1413 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1414 names a non-existent directory. (thanks to Andreas Fuchs)
1415 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1416 the PPC platform, lead to stack corruption; this has been fixed.
1417 (reported by Rainer Joswig)
1418 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1419 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1420 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1421 arithmetic in general, is now much more reliable on the PPC
1423 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1424 platform now returns the right answer.
1425 * optimization: restored some effective method precomputation in
1426 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1427 precomputation is now tunable.
1428 * optimization: compiler-internal data structure use has been
1429 reviewed, and changes have been made that should improve the
1430 performance of the compiler by about 20%.
1431 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1432 simple-base-strings and simple-bit-vectors is improved.
1433 * optimization: the optimization of 32-bit logical and arithmetic
1434 functions introduced in version 0.8.3 on the x86 has been
1435 implemented on the mips, ppc and sparc platforms; an
1436 implementation of the same facility, but for 64-bit arithmetic,
1437 has been added for the alpha.
1438 * microoptimization: the compiler is better able to make use of the
1439 x86 LEA instruction for multiplication by constants.
1440 * bug fix: in some situations compiler did not report usage of
1441 generic arithmetic in (SPEED 3) policy.
1442 * bug 145b fix: compiler used wrong type specifier while converting
1443 MEMBER-types to numeric.
1444 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1446 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1447 index argument now works properly on the Alpha platform.
1448 * bug fix: floating point exception treatment on the Alpha platform
1450 * bug fix: FILE-POSITION works much better on string input and
1451 output streams. (thanks to Nikodemus Siivola)
1452 * bug fix: many threading/garbage collection symptoms sorted.
1453 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1454 might be pseudo-atomic.
1455 * internal change: Stopping for GC is now done with signals not
1456 ptrace. GC is now done in whichever thread wanted it, instead of
1458 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1460 * bug fix: result form in DO is not contained in the implicit
1462 * incompatible change: ICR structure is changed; the value part of
1463 CONTINUATION is now called LVAR; corresponding functions are
1464 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1465 * added type deriver for ISQRT (thanks to Robert E. Brown).
1466 * bug fix: better support for loading from the command line when an
1467 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1469 * fixed some bugs revealed by Paul Dietz' test suite:
1470 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1471 ** ROUND and FROUND now give the right answer when given very
1472 small float arguments.
1473 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1475 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1476 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1477 ** LCM should return a non-negative integer.
1478 ** PARSE-INTEGER returned the index of a terminator instead of the
1479 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1480 ** PARSE-INTEGER returned an incorrect index being applied to a
1482 ** LCM with two arguments of 0 returns 0 rather than signalling
1484 ** unsigned addition of a 32-bit constant with the high bit set no
1485 longer causes an internal compiler error on the x86.
1486 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1487 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1488 with negative last argument.
1489 ** byte specifiers with zero size and position no longer cause
1490 an error during type derivation.
1491 ** bignum multiplication on the Alpha platform now returns the
1493 * porting: The system now builds on SuSE AMD64, although it still
1494 generates a 32-bit binary.
1495 * .fasl file incompatibility: The fasl file version number has
1496 been incremented (because of the changes to internal compiler
1497 data structures referred to above).
1499 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1500 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1501 more accurately, on the Darwin kernel running on PowerPC hardware.
1502 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1503 * Compiler code deletion notes now signal a condition of type
1504 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1505 an associated MUFFLE-WARNING restart.
1506 * The compiler now performs limited argument count validation of
1507 constant format strings in FORMAT, and where appropriate in ERROR,
1508 CERROR and WARN. (thanks to Gerd Moellmann)
1509 * New ASDF-INSTALL contrib can be used for automatic download and
1510 installation of third-party Lisp code from CCLAN or other sites
1512 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1513 which forces another thread to execute a function supplied by the
1515 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1516 now accept and act upon their :ELEMENT-TYPE keyword argument.
1517 (reported by Martin Atzmueller, Edi Weitz)
1518 * bug fix: FILE-POSITION now accepts position designators up to
1519 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1520 is the greater. (thanks to Patrik Nordebo)
1521 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1522 Kalle Olavi Niemitalo)
1523 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1524 printable. (reported by Eric Marsden)
1525 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1526 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1527 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1528 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1529 (thanks to Stig Sandoe)
1530 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1531 UNREAD-CHAR now correctly includes the unread character in the
1532 target sequence. (thanks to Gerd Moellmann)
1533 * bug fix in threaded builds: the system can now be suspended and
1534 resumed by shell job control with minimal disruption.
1535 * bug fixes in times and timezones >2038AD
1536 * better handling of "where is GNU make?" problem in build scripts
1537 (thanks to Nikodemus Siivola)
1538 * new optimization: inside a named function any reference to a
1539 function with the same name is considered to be a self-reference;
1540 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1541 optimization quality.
1542 * new optimization on x86: logical functions and + now have
1543 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1544 used when the result is truncated to 32 bits.
1545 * VALUES declaration is partially enabled.
1546 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1547 * fixed some bugs revealed by Paul Dietz' test suite:
1548 ** The system now obeys the constraint imposed by
1549 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1550 types form a lattice under type intersection.
1551 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1552 ** ASSOC now ignores NIL elements in an alist.
1553 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1554 and (1+ MOST-POSITIVE-FIXNUM) answers.
1555 ** The addition of a method with invalid qualifiers to a generic
1556 function does not cause an error to be signalled immediately;
1557 a warning is signalled, and the error is generated only on
1558 calling the generic function.
1559 * changed .fasl file version number, in part to add type codes for
1560 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1561 obscure ANSI requirements
1563 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1564 * fixed bug 148: failure to inline-expand a local function left
1565 garbage, confusing the compiler.
1566 * fixed bugs 3cd: structure slot readers perform type check if the
1567 slot can have an invalid value (i.e. it is either not initialized
1568 or can be written with a less specific slot writer).
1569 * bug fix: the compiler now traps array references to elements off
1570 the end of an array; previously, the bounds checking in some
1571 circumstances could go off-by-one.
1572 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1574 * type declarations for array element types now obey the description
1575 on the CLHS page "Declaration TYPE", as per discussions on
1576 sbcl-help around 2003-05-08. This means that a declaration
1577 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1578 declaration, all references to BAR will be asserted or assumed
1579 (with THE, so dependent on compiler policy) to involve objects of
1580 type FOO. Note that no such declaration is implied in
1581 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1582 * declared types of functions from the "Conditions"
1583 chapter. (reported by Paul Dietz)
1584 * bug fix: CERROR accepts a function as its first argument.
1585 * bug fix: NTH an NTHCDR accept a bignum as index
1586 arguments. (reported by Adam Warner)
1587 * optimization: character compare routines now optimize comparing
1588 against a constant character. (reported by Gilbert Baumann)
1589 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1590 argument now works properly on the MIPS platform.
1591 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1592 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1593 declared to have a functional type, does not cause an error
1595 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1596 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1597 * fixed bug: if last continuation of a deleted block has a
1598 destination, this destination should be deleted too. (reported by
1600 * fixed a bug in the bootstrap process: the host compiler's values
1601 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1602 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1603 test case from Patrik Nordebo)
1604 * improved the ability of the disassembler on the PPC platform to
1605 provide helpful disassembly notes.
1606 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1607 the class in more cases than previously.
1608 * bug fix: FILE-POSITION now understands :START and :END for
1609 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1610 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1611 8) data. (thanks to David Lichteblau)
1612 * bug fix: it is possible to add a method to a generic function
1613 without lambda list.
1614 * bug fix: reader failed to signal END-OF-FILE inside an
1615 object representation. (reported by Nikodemus Siivola)
1616 * fixed some bugs revealed by Paul Dietz' test suite:
1617 ** LAST and [N]BUTLAST should accept a bignum.
1618 ** condition slot accessors are methods.
1619 ** (VECTOR NIL) is a subtype of STRING.
1621 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1622 * minor incompatible change: some nonsensical specialized lambda
1623 lists (used in DEFMETHOD) which were previously ignored now signal
1625 * minor incompatible change: the system is now aware of the types of
1626 variables in the COMMON-LISP package, and will signal errors for
1627 most violations of these type constraints (where previously they
1628 were silently accepted).
1629 * minor incompatible change: COMPILE-FILE now uses the freedom
1630 afforded (ANSI 3.2.2.3) to use derived function types for
1631 functions defined in the same file. This also permits the system
1632 to warn on static type mismatches and function
1633 redefinition. (Currently it does not work with high DEBUG level.)
1634 * minor incompatible change: VALUES declaration is disabled.
1635 * When issuing notes, the compiler now signals a condition of type
1636 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1637 restart for use in user handlers. It is expected that the
1638 COMPILER-NOTE condition will eventually become a condition
1639 supertype to a hierarchy of note types, which will then be
1640 handleable in a similar fashion. However, at the moment, no such
1641 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1642 but it's an implementation detail, not a classification for the
1644 * Changes in type checking closed the following bugs:
1645 ** type checking of unused values (192b, 194d, 203);
1646 ** template selection based on unsafe type assertions (192c, 236);
1647 ** type checking in branches (194bc).
1648 * A short form of VALUES type specifier has ANSI meaning (it has
1649 increased the number of situations when SBCL cannot perform type
1651 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1652 keywords or constants is permissible.
1653 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1654 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1655 classes. (thanks to Antonio Martinez)
1656 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1657 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1658 argument to a pretty-print stream. Instead, it leaves any such
1659 operations to DESCRIBE-OBJECT methods.
1660 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1662 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1663 lambda lists. (thanks to David Lichteblau)
1664 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1665 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1666 (thanks to Gerd Moellmann)
1667 * fixed bug 246: increased compilation speed of long
1668 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1670 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1672 * DEFCONSTANT now throws a condition of type
1673 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1674 constant to a non-EQL value; CONTINUE and ABORT restarts
1675 respectively change and preserve the value.
1676 * fixed bug 63: The code walker, part of the implementation of CLOS,
1677 is now better at handling symbol macros.
1678 * bug fix: There is no longer an internal implementation type named
1679 CL:LENGTH. (reported by Raymond Toy)
1680 * bug fix: In macro-like defining macros/special operators the
1681 implicit block does not enclose the lambda list.
1682 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1683 enclosed in parentheses) are not suitable as type specifiers, and
1684 their use properly signals an error now.
1685 * bug fix: An argument count mismatch for a type specifier in code
1686 being compiled no longer causes an unhandled error at compile
1687 time, but signals a compile-time warning.
1688 * fixed simple vector readable printing
1689 * bug fix: DESCRIBE takes more care over whether the class
1690 precedence list slot of a class is bound before accessing it.
1691 (reported by Markus Krummenacker)
1692 * bug fix: FORMATTER can successfully compile pretty-printer format
1693 strings which use variants of the ~* directive inside.
1694 * bug fix: SEARCH now applies its TEST predicate to the elements of
1695 the arguments in the correct order. (thanks to Wolfhard Buss)
1696 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1697 in type inference. (thanks to Robert E. Brown)
1698 * bug fix: Reading in symbols with an explicit package name of ""
1699 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1700 * fixed some bugs revealed by Paul Dietz' test suite:
1701 ** NIL is now allowed as a structure slot name.
1702 ** Arbitrary numbers, not just REALs, are allowed in certain
1703 circumstances in LOOP for-as-arithmetic clauses.
1704 ** Multiple class redefinitions before slot access no longer
1705 causes a type error.
1706 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1707 association between the name and a class.
1708 ** Generic functions with non-standard method-combination and over
1709 five methods all of which return constants no longer return NIL
1710 after the first few invocations. (thanks to Gerd Moellmann)
1711 ** CALL-NEXT-METHOD with no arguments now passes the original
1712 values of the arguments, even in the presence of assignment.
1713 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1715 ** DISASSEMBLE works with closures and funcallable instances.
1716 ** ADD-METHOD now returns the generic function, not the new method.
1717 ** FIND-METHOD signals an error if the lengths of the specializers
1718 is incompatible with the generic function, even if the ERRORP
1720 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1721 which its argument is a member.
1722 ** DEFCLASS only redefines the class named by its class-name
1723 argument if that name is the proper name of the class;
1724 otherwise, it creates a new class.
1725 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1726 of the UNBOUND-SLOT condition to the name of the slot.
1727 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1728 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1729 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1730 SLOT-MAKUNBOUND in the specified fashion.
1732 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1733 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1734 cross-compilation host. As a consequence, we can now bootstrap our
1735 way up to SBCL starting with a bare gcc toolchain and human-readable
1736 source code (first the source to CLISP, then the source to SBCL).
1737 * A contributed module containing a partial implementation of the
1738 simple-streams interface has been included. (thanks to Rudi
1740 * A contributed module implementing the RFC1321 Message Digest
1741 Algorithm, known as MD5, has been included.
1742 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1743 no longer has any effect, as the code controlled by this feature
1744 has been deleted. (As far as we know, no-one has ever built using
1745 this feature, and its semantics were confused in any case).
1746 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1747 work on conditions (as required by the ANSI specification),
1748 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1749 expected behaviour on conditions. Users should note, however,
1750 that such behaviour is not required by the ANSI specification,
1751 and so use of this behaviour may render their code unportable.
1752 * fixed some bugs revealed by Paul Dietz' test suite:
1753 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1755 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1756 ** SXHASH on condition objects no longer returns NIL.
1757 ** :ALLOCATION :CLASS slots are better treated; their values are
1758 updated on class redefinition, and initforms inherited from
1759 superclasses are applied.
1760 ** REMOVE-METHOD returns its generic function argument even when
1761 no method was removed.
1762 ** SHARED-INITIALIZE now initializes the values of the requested
1763 slots, including those with :ALLOCATION :CLASS.
1764 ** ALLOCATE-INSTANCE now works on structure classes defined via
1765 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1767 ** SLOT-EXISTS-P now works on conditions, as well as structures
1769 ** MAKE-LOAD-FORM now has the required methods on
1770 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1771 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1772 keyword, and now implements the SLOT-NAMES argument.
1773 ** methods with &OPTIONAL arguments no longer allow too many
1774 arguments to be passed in the call without error.
1775 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1776 option is consistent with the required arguments of the generic
1777 function lambda list.
1778 * bug fix: REQUIRE accepts a string designator. (Thanks to
1780 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1781 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1782 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1783 * bug fix: The compiler now checks for duplicated variables in macro
1785 * bug fix: SETQ on globals returns the correct value.
1786 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1787 not a COMPILER-ERROR (followed by some other strange error on
1788 choosing the CONTINUE restart).
1789 * bug fix: make.sh and friends are now more consistent in the way that
1790 they look for GNU "make".
1792 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1793 * experimental native threads support (on x86 Linux >=2.4 only).
1794 This is not compiled in by default: you need to add :SB-THREAD to
1795 the target features. See the "Beyond ANSI" chapter of the manual
1797 * fix for longstanding nonANSIism: The old distinction between
1798 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1799 The return value from CL:FIND-CLASS is now a CLOS class, and
1800 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1801 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1802 * An interface to the MetaObject Protocol, as described in Kiczales,
1803 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1804 MIT Press, 1991, is available from the SB-MOP package.
1805 * incompatible change: the SB-PCL package should now be considered
1806 a private implementation detail, and no longer a semi-private MOP
1808 * minor incompatible change: due to rearrangement for threads, the
1809 control stack and binding stack are now allocated at arbitrary
1810 addresses instead of being hardcoded per-port. Users affected by
1811 this probably have to be doing advanced things with shared
1812 libraries, and will know who they are.
1813 * minor incompatible change: Previously, all --eval forms used were
1814 processed with READ before any of them were processed with EVAL.
1815 Now each --eval form is processed with both READ and EVAL before
1816 the next --eval form is processed. (Thus package operations like
1817 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1818 work as the user might reasonably expect.)
1819 * minor incompatible change: *STANDARD-INPUT* is now only an
1820 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1822 * minor incompatible change: Y-OR-N-P is now character-oriented,
1823 not line oriented. Also, YES-OR-NO-P now works without errors.
1824 (thanks to Antonio Martinez)
1825 * sb-aclrepl module improvements: an integrated inspector, added
1826 repl features, and a bug fix to :trace command.
1827 * Known functions, which cannot be open coded by the backend, are
1828 considered to be able to check types of their arguments. (fixing
1829 a bug report by Nathan J. Froyd)
1830 * fixed a bug in computing method discriminating functions: It is
1831 now possible to define methods specialized on classes which have
1832 forward-referenced superclasses. (thanks to Gerd Moellmann)
1833 * fixed evaluation order in optional entries (reported by Gilbert
1835 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1836 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1837 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1838 required-by-AMOP NAME argument, as well as CLASS and
1839 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1840 * fixed bug 20: DEFMETHOD can define methods using names that are
1841 not the proper names of classes to designate class specializers.
1842 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1843 Unix file descriptors, instead of blowing up. (thanks to Antonio
1845 * Garbage collection refactoring: user-visible change is that a
1846 call to the GC function during WITHOUT-GCING will not do garbage
1847 collection until the end of the WITHOUT-GCING. If you were doing
1848 this you were probably losing anyway.
1849 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1850 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1851 TYPEP the latter but not the former.
1852 * The compiler issues a full WARNING for calls to undefined functions
1853 with names from the CL package.
1854 * MAP-INTO for a vector destination is open coded. (reported by
1855 Brian Downing on c.l.l)
1856 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1857 documentation string.
1858 * fixed some bugs revealed by Paul Dietz' test suite:
1859 ** COPY-ALIST now signals an error if its argument is a dotted
1861 ** Condition slots are now accessed more correctly in the presence
1862 of multiple initargs for a given slot.
1863 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1864 exclude restarts of the same name associated with a different
1866 ** DEFCLASS of forward-referenced classes with another
1867 forward-referenced class in the superclasses list no longer
1869 ** Condition slots are now initialized once each, not multiple
1870 times. (thanks to Gerd Moellmann)
1871 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1873 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1874 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1875 arguments contain duplicated elements.
1876 ** RESTART-CASE understands local macros.
1877 ** RESTART-CASE associates exactly its own restarts with a condition.
1878 ** ENDP in safe mode checks its argument to be of type LIST.
1879 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1880 in question is unbound.
1881 ** Optimized MAKE-INSTANCE functions no longer cause internal
1882 assertion failures in the presence of duplicate initargs.
1883 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1884 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1885 propagated even in the ctor optimized implementation of
1887 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1889 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1890 * a better implementation of SXHASH on (simple) bit vectors,
1891 measured both in execution speed and in distribution of results
1892 over the positive fixnums, has been installed. Likewise, a better
1893 implementation of EQUAL for simple bit vectors is now available.
1894 * fixed CEILING optimization for a divisor of form 2^k.
1895 * fixed bug 240 (emitting extra style warnings "using the lexical
1896 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1897 by Antonio Martinez)
1898 * fixed SXHASH, giving different results for NIL depending on type
1899 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1900 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1901 variable. (found by Rolf Wester)
1902 * SBCL does not ignore type declarations for special
1903 variables. (reported by rif on c.l.l 2003-03-05)
1904 * some bug fixes in contrib/sb-aclrepl/
1905 * fixed some bugs revealed by Paul Dietz' test suite:
1906 ** a bug in the CONS type specifier, whereby the CAR and CDR
1907 types got intertwined, has been fixed;
1908 ** the type system is now able to reason about the interaction
1909 between INTEGER and RATIO types more completely;
1910 ** APPEND, [N]REVERSE and NRECONC check that those their
1911 arguments, which must be proper lists, are really so;
1912 ** An array specialized to be unable to hold elements has been
1913 implemented, as required -- yes, really -- by ANSI;
1914 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1915 on malformed property lists;
1917 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1918 * incompatible packaging change: in line with Unix convention,
1919 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1920 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1921 * REQUIRE and PROVIDE are now optionally capable of doing something
1922 useful. See the documentation string for REQUIRE.
1923 * infrastructure for a managed SBCL contrib system: contributed
1924 modules in this release include:
1925 ** the ASDF system definition facility;
1926 ** an interface to the BSD Sockets API;
1927 ** an ACL-like convenience interface to the repl;
1928 (thanks to Kevin Rosenberg)
1929 ** an implementation of ROTATE-BYTE, with efficient implementation
1931 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1932 gives a closer approximation to the right answer than previously.
1933 (thanks to Raymond Toy)
1934 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1935 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1936 optional environment arguments, as required by ANSI.
1937 * fixed bugs in other functions taking environment objects, allowing
1938 calls with an explicit NIL environment argument to be compiled
1940 * fixed bug 228: primary return values from
1941 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1942 COMPILE or FUNCTION.
1943 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1944 :TYPE will no longer signal errors on innocuous objects.
1945 * fixed bug 231b: SETQ is better at respecting type declarations in
1946 the lexical environment.
1947 * fixed a bug in DEFCLASS: classes named by symbols with no or
1948 unprintable packages can now be defined.
1949 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1950 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1951 * fixed compiler failure related to checking types of functions.
1952 (reported by Robert E. Brown)
1953 * the compiler is now much more consistent in its error-checking
1954 treatment of bounding index arguments to sequence functions: in
1955 (SAFETY 3) code, errors will be signalled in almost all cases if
1956 invalid sequence bounding indices are passed to functions defined
1957 by ANSI to operate on sequences.
1958 * fixed a bug in the build procedure: documentation of SBCL-specific
1959 packages is now preserved and available in the final Lisp image.
1960 * lifted FDEFINITION lookup out of loops in the implementation of
1961 many list operations. (thanks to Robert E. Brown)
1962 * fixed a bug in the reader: the #n# reader macro now works for
1963 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1964 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1965 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1966 E. Brown and Tony Martinez respectively)
1967 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1968 SLOT-BOUNDP: in particular, it is now able to optimize them much
1969 better, and is now not vulnerable to having packages renamed.
1970 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1971 always called when a slot is not present in an instance. (thanks
1973 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1974 longer signal an error on unreading a character following EOF on
1975 the previous constituent stream. (thanks to Tony Martinez)
1976 * fixed some bugs revealed by Paul Dietz' test suite:
1977 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1978 not just nonnegative fixnums;
1979 ** the logical bit-array operators such as BIT-AND now accept an
1980 explicit NIL for their "opt-arg" argument (to indicate a
1981 freshly-consed result bit-array);
1982 ** ELT now signals an error on an invalid sequence index in safe
1984 ** the type system is now cleverer about negations of numeric
1985 types, and consequently understands the BIGNUM and RATIO types
1987 ** the type system is now cleverer about the interaction between
1988 INTEGER and RATIO types: while bugs still remain, many more
1989 cases are accurately computed;
1990 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1991 if it is in the last clause;
1992 ** CONSTANTLY now correctly returns a side-effect-free function in
1994 ** DECLARE is no longer treated as a special-operator; in
1995 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1996 * incremented fasl file version number due to the change in the
1997 DEFSTRUCT-SLOT-DESCRIPTION structure.
1999 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2000 * minor incompatible change: code processed by the "interpreter" or
2001 EVAL now has a compilation optimization policy of (DEBUG 2)
2002 (changed from (DEBUG 1)) to improve debuggability of interactive
2003 development, and to allow the use of the debug RETURN command in
2005 * an experimental implementation of the RETURN command for the
2006 debugger has been included. (thanks to Frederik Kuivinen)
2007 * fixed bug 62: constraints were not propagated into a loop.
2008 * fixed bug in embedded calls of SORT (reported and investigated by
2010 * fixed some bugs revealed by Paul Dietz' test suite:
2011 ** printing and reading of arrays with some dimensions having
2012 length 0 (thanks to Gerd Moellmann);
2013 ** BOA constructor with &AUX argument without a default value does
2014 not cause a type error;
2015 ** CONSTANTP now returns true for all self-evaluating objects.
2017 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2018 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2019 accessors that are related by inheritance, as specified in the
2020 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2021 Valtteri Vuorikoski)
2022 * The compiler is now able to inline functions that were defined in
2023 a complex lexical environment (e.g. inside a MACROLET).
2024 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2026 * Support for the upcoming FreeBSD-5.0 release has been included.
2027 (thanks to Dag-Erling Smorgrav)
2028 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2029 effect when it is not in a toplevel context.
2030 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2031 stand a better chance of being correct. (thanks to Gerd
2033 * fixed bug in COERCE, which now signals an error on coercing a
2034 rational to a bounded real type which excludes the expected
2036 * The compiler is now able to derive types more accurately from the
2037 COERCE and COMPILE functions.
2038 * fixed bug 223: functional binding is considered to be constant
2039 only for symbols in the CL package.
2040 * fixed bug 231: SETQ did not check the type of a variable being set
2041 (reported by Robert E. Brown)
2042 * A new optimization for MAKE-INSTANCE has been included, fixing
2043 various bugs (including relating to :ALLOCATION :CLASS slots and
2044 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2046 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2047 ** As required by ANSI, LOOP now disallows anonymous collection
2048 clauses such as COLLECT I in conjunction with aggregate boolean
2049 clauses such as THEREIS (= I 1);
2050 ** LOOP now signals an error when any variable is reused in the
2051 same loop (including the potentially useful construct analogous
2052 to WITH A = 1 WITH A = (1+ A);
2053 ** IT is only a special loop symbol within the first clause of a
2054 conditional loop clause;
2055 ** LOOP with a typed iteration variable over a hashtable now
2056 signals a type error iff it should.
2057 * fixed some other bugs revealed by Paul Dietz' test suite:
2058 ** FILE-STREAM now names the class previously known as FD-STREAM;
2059 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2060 argument) no longer signals an error;
2061 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2062 of the slot symbol, rather than using the current package
2063 ((:CONC-NAME "") continues to intern the slot's name in the
2065 * incremented fasl file version number, because of the incompatible
2066 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2067 of the new implementation of DEFINE-COMPILER-MACRO.
2069 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2070 * Support for building SBCL for MIPS platforms running in
2071 little-endian mode has now been checked in, and basic
2072 functionality on said platforms verified.
2073 * minor incompatible change: PCL now records the pathname of a file
2074 in which methods and the like are defined, rather than its
2076 * minor incompatible change: TRUENAME now considers the truename of
2077 a file naming a directory to be the pathname with :DIRECTORY
2078 component indicating that directory.
2079 * minor incompatible change: a NAMED clause in the extended form of
2080 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2081 reason for the previous behaviour is unclear.
2082 * more systematization and improvement of CLOS and MOP conformance
2083 in PCL (thanks to Gerd Moellman and Pierre Mai):
2084 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2086 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2087 primary methods with no specializers;
2088 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2090 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2091 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2092 FORWARD-REFERENCED-CLASSes; error reporting on
2093 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2095 ** SXHASH on CLOS instances now uses a slot internal to the
2096 instance to return different numbers on distinct instances,
2097 while preserving the same return value through invocations of
2099 ** DEFMETHOD signals errors when methods with longer incongruent
2100 lambda lists are added to generic functions;
2101 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2102 CLASS, as specified in AMOP;
2103 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2104 on the order returned by the primary method for classes of
2105 class STANDARD-CLASS;
2106 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2107 * fixed some bugs shown by Paul Dietz' test suite:
2108 ** DOLIST puts its body in TAGBODY;
2109 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2111 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2112 value producing form;
2113 ** if more variables are given to PROGV than values, extra
2114 variables are bound and made to have no value;
2115 ** NSUBSTITUTE on list arguments gets the right answer with
2117 ** ELT signals an error of type TYPE-ERROR when the index argument
2118 is not a valid sequence index;
2119 ** LOOP signals (at macroexpansion time) an error of type
2120 PROGRAM-ERROR when duplicate variable names are found;
2121 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2122 ** FUNCALL of special-operators now cause an error of type
2124 ** PSETQ now works as required in the presence of side-effecting
2125 symbol-macro places;
2126 ** NCONC accepts any object as its last argument;
2127 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2129 ** loop-for-as-package does not require a package to be explicitely
2131 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2133 * fixed bug 166: compiler preserves "there is a way to go"
2134 invariant when deleting code.
2135 * fixed bug 172: macro lambda lists with required arguments after
2136 &REST arguments now cause an error to be signalled. (thanks to
2138 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2139 bug. (thanks to Matthew Danish)
2140 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2142 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2143 arguments are lexically rebound. (thanks to Gerd Moellmann and
2145 * fixed bug 194: error messages are now more informative when there
2146 is no primary method applicable in a call to a generic
2147 function. (thanks to Gerd Moellmann)
2148 * fixed bug in command line argument checking (thanks to Julian
2150 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2151 (thanks to Matthew Danish)
2152 * incremented fasl file version number, because of the
2153 SXHASH-related changes in the layout of CLOS data structures
2155 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2156 * minor incompatible change: The runtime (the Unix executable named
2157 "sbcl") is now much pickier about the .core files it will load.
2158 Essentially it now requires .core files to descend from the same
2159 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2160 as the runtime does. (The intent is to prevent the crashes which
2161 can occur, and which can even be reported as mysterious failures,
2162 when people patch the sources or change the build parameters
2163 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2164 sbcl and .core files.)
2165 * fixed bug: VALUES-LIST is no longer optimized away.
2166 * fixed bug 142: The FFI conversion of C string values to Lisp
2167 string values no longer conses excessively. (thanks to Nathan
2168 Froyd porting Raymond Toy's fix to CMU CL)
2169 * began to systematize and improve MOP conformance in PCL (thanks to
2170 Nathan Froyd, Gerd Moellman and Pierre Mai):
2171 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2173 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2174 ** FINALIZE-INHERITANCE is now called on class finalization;
2175 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2176 argument precedence order.
2177 * fixed bug 202: The compiler no longer fails on functions whose
2178 derived types contradict their declared type.
2179 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2180 so it can be non-toplevel.
2181 * The fasl file version number has changed (because of the new
2182 implementation of DEFMACRO).
2183 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2184 safe code, that any length requirement by their type-specifier
2185 argument is valid. The exceptions to this are described in bug
2187 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2188 functions now check the types of their inputs as required by ANSI.
2189 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2190 introduced symbol is DECLAREd to be SPECIAL.
2191 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2192 * fixed bug: :COUNT argument to sequence functions may be negative
2193 * fixed bug: body of DO-SYMBOLS may contain declarations
2194 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2195 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2196 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2197 symbol macro only once
2198 * fixed printing of call frame when argument list is unavailable
2199 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2200 * compiler no longer signals WARNING on unknown keyword
2203 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2204 * A beta-quality port to the mips architecture running Linux,
2205 based on the old CMUCL backend, has been made. It has been tested
2206 on a big-endian kernel, and works sufficiently well to be able to
2207 rebuild itself; it has not been tested in little-endian mode.
2208 * fixed an inconsistency between gencgc.c and purify.c which made
2209 dumping/loading .core files unreliable
2210 * fixed bug 120a: The compiler now deals correctly with IFs where
2211 the consequent is the same as the alternative, instead of
2212 misderiving the return type. (thanks to Alexey Dejneka)
2213 * fixed bug 113: Logical pathnames are now dumpable (the logical
2214 host is resolved at load-time, throwing an error if it is not
2216 * fixed bug 174: FORMAT's error message is slightly clearer when a
2217 non-printing character is used in a format directive.
2218 * fixed several bugs in compiler checking of type declarations, i.e.
2219 violations of the Python "declarations are assertions" principle
2220 (thanks to Alexey Dejneka)
2221 * fixed several bugs in PCL's error checking (thanks to Gerd
2223 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2225 * fixed bug in compilation of functions as first class values
2226 (thanks to Antonio Martinez-Shotton)
2227 * The compiler's handling TYPE-ERRORs which it can prove will
2228 inevitably happen at runtime has been cleaned up and corrected
2229 in several ways. (thanks to Alexey Dejneka)
2230 * improved argument type checking for various basic arithmetic
2231 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2232 lavished on them in the past that they can be compiled in many
2233 ways in different special cases
2234 * fixed bug 181: compiler checks validity of user supplied type
2236 * cleaned up code flushing in optimization: Function calls which
2237 should signal errors for safety purposes (e.g. which ANSI says
2238 should signal errors when their arguments are of incorrect type)
2239 are no longer optimized away.
2240 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2241 * incremented fasl file version number, because changes in the
2242 implementation of sequence functions like COERCE caused
2243 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2244 in old inline expansions) to become undefined. (Actually these
2245 changes were later undone, so we might very well be binary
2246 compatible with 0.7.7 after all, but leaving the version number
2247 incremented seemed like the simplest and most conservative
2250 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2251 * An alpha-quality port to the parisc architecture running Linux,
2252 based on the old CMUCL backend, has been made. This, even more so
2253 than the other backends, should be considered still a work in
2254 progress; known problems include that the Linux kernel in 64-bit
2255 mode does not propagate the correct sigcontext structure to
2256 userspace, and consequently SBCL on a parisc64 kernel will not
2258 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2259 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2260 suppressed.) Also now that the compiler is looking at declarations
2261 in the environment, it checks optimization declarations as well,
2262 and suppresses inlining when (> DEBUG SPEED).
2263 * More fixes have been made to treatment of floating point exception
2264 treatment and other Unix signals. In particular, floating point
2265 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2266 * The detection and handling of control stack exhaustion (infinite
2267 or very deeply nested recursion) has changed. Stack exhaustion
2268 detection is now done by write-protecting pages at the OS level
2269 and applies at all optimization settings; when found, a
2270 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2271 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2272 be caught using IGNORE-ERRORS.
2273 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2274 correctly when transforming recursive LABELS functions to LETs.
2275 (thanks to Alexey Dejneka)
2276 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2277 that are names of constants or global variables.
2278 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2279 alien routines with docstrings.
2280 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2281 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2283 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2284 (thanks to Eric Marsden)
2285 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2286 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2287 * Bug fix: The SPARC backend can now compile functions involving
2288 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2289 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2290 to a functions expecting a PATHNAME-DESIGNATOR.
2291 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2292 lambda lists. (thanks to Alexey Dejneka)
2293 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2294 functions so that EXPORTing the name of the class doesn't cause
2295 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2296 to Antonio Martinez for reporting this)
2297 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2298 constant-folded without failing an assertion. (thanks to Einar
2299 Floystad Dorum for reporting this)
2300 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2301 types are now dealt with more correctly by the compiler.
2302 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2303 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2304 behaviour analogous to RENAME-FILE. This puts its behaviour more
2305 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2307 * The fasl file version number has changed again. (because of the
2308 bug fix involving the names of PCL MAKE-INSTANCE functions)
2310 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2311 * bug fix: Floating point exceptions are treated much more
2312 consistently on the x86/Linux and PPC/Linux platforms.
2313 * Array initialization with :INITIAL-ELEMENT is now much faster for
2314 cases when the compiler cannot open code the array creation, but
2315 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2316 array accesses have also seen a speed increase.
2317 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2318 specified. (This was at the root of some bad interactions between
2319 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2321 * bug fix: Internal error arguments for undefined functions are now
2322 computed correctly on the PPC/Linux platform.
2323 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2324 Raymond Toy's patch for CMU CL)
2325 * Support for the Solaris 9 operating environment has been included
2326 (thanks to Daniel Merritt)
2327 * A very ugly but hopefully complete draft of the missing FFI chapter
2328 of the manual has been created by reformatting the corresponding
2329 CMU CL manual chapter into (currently very ugly and incoherent)
2330 DocBook and bringing it up to date for SBCL behavior. Thus, the
2331 manual is now essentially complete, at least by my extreme
2332 once-and-only-once standards, whereby it's acceptable to refer to
2333 the doc strings of SB-EXT functions as the primary documentation.
2334 * The fasl file version number has changed again, due to cleanup of
2335 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2336 is no longer a static symbol.)
2338 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2339 * SBCL now builds with OpenMCL (version 0.12) as the
2340 cross-compilation host; also, more progress has been made toward
2341 bootstrapping under CLISP.
2342 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2344 * bug 158 fixed: The compiler can now deal with integer loop
2345 increments different from 1; fixing this turned out also to fix
2347 * bug 169 fixed: no more bogus warnings about using lexical bindings
2348 despite the presence of perfectly good SPECIAL declarations (thanks
2349 to David Lichteblau)
2350 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2351 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2352 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2354 * bug 180 fixed: Method combination specifications no longer ignore
2355 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2356 * bug fix: Structure type predicate functions now check their argument
2357 count as they should.
2358 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2359 correctly. (thanks to Pierre Mai)
2360 * minor incompatible change: The --noprogrammer option is deprecated
2361 in favor of the new --disable-debugger option, which is very similar.
2362 (The major difference is that it takes effect at a slightly different
2363 time at startup, causing handling of errors in --sysinit and
2364 --userinit files will be affected differently.) The
2365 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2366 been added to allow this functionality to be controlled from ordinary
2367 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2368 maintainers, who might want to run non-interactive scripts to
2369 build SBCL cores which will later be used interactively.)
2370 * minor incompatible change: The LOAD function no longer, when given
2371 a wild pathname to load, loads all files matching that pathname.
2372 Instead, an error of type FILE-ERROR is signalled.
2374 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2375 * bug 147 fixed: The compiler preserves its block link/count
2376 invariants more correctly now so that it doesn't crash. (thanks
2378 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2380 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2381 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2382 no-value-for-this-character as NIL (as specified by ANSI).
2383 * HOST-NAMESTRING on physical pathnames now returns a string that is
2384 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2385 (thanks to Christophe Rhodes)
2386 * The Alpha port handles icache flushing more correctly. (thanks to
2388 * More progress has been made toward bootstrapping under CLISP. (thanks
2389 to Christophe Rhodes)
2390 * The fasl file format has changed again, because dynamic loading
2391 on OpenBSD (which has non-ELF object files) motivated some cleanups
2392 in the way that foreign symbols are transformed and passed around.
2393 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2394 is no longer treated as whitespace by the reader, but instead as
2395 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2396 |A
\7fB|, instead of A as it used to.
2398 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2399 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2400 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2401 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2402 this as of 0.7.1.45, but was left out of the previous news section
2403 (thanks to Dan Barlow)
2404 * SBCL now runs on the Solaris operating system on SPARC architectures
2405 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2406 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2407 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2408 * SPARC backend cleanups, allowing builds of cores optimized for V8
2409 and V9 SPARCS, and also emission of code targeted to a particular
2410 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2412 * SBCL is closer to bootstrapping under CLISP, thanks to various
2413 fixes by Christophe Rhodes.
2414 * The fasl file format has changed again, to allow the compiler's
2415 INFO database to support symbol macros.
2416 * The user manual (in doc/) is formatted into HTML more nicely.
2417 (thanks to coreythomas)
2418 * The system is smarter about SUBTYPEP relationships, especially
2419 those involving NOT types (including types such as ATOM which are
2420 represented internally using NOT types). Thus SUBTYPEP is less
2421 likely to return (VALUES NIL NIL) in general, and in particular
2422 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2424 * The fasl file format has changed again, because the internal
2425 representation of types now includes a new slot to support the new
2426 SUBTYPEP-of-NOT-types logic.
2427 * (not a change in the main branch of SBCL, but a related prototype
2428 which can hopefully be merged into the main branch of SBCL in the
2429 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2430 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2431 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2432 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2433 dispatch macro character combinations. (thanks to Alexey Dejneka)
2434 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2435 that superficially look like logical namestrings correctly.
2436 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2438 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2439 default supplied-p) for &optional and &key arguments. (thanks to
2442 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2443 * incompatible change: The compiler is now less aggressive about
2444 tail call optimization, doing it only when (> SPACE DEBUG) or
2445 (> SPEED DEBUG). (This is an incompatible change because there are
2446 programs which relied on the old CMU-CL-style behavior to optimize
2447 away their unbounded recursion which will now die of stack overflow.)
2448 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2449 for non-GENCGC systems has been increased to 20M (since that
2450 seems much closer to the likely performance optimum for modern
2451 systems than the old 4M value was)
2452 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2453 and *DEBUG-PRINT-LEVEL*
2454 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2455 of CMU CL's support for SPARC, and various endianness and other
2456 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2457 * new syntactic sugar for the Unix command line: --load foo.bar is now
2458 an alternate notation for --eval '(load "foo.bar")'.
2460 ** The system now detects stack overflow and handles it gracefully,
2461 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2462 optimization settings. (This is a good thing in general, and
2463 its introduction in this version should be particularly timely
2464 for anyone whose code fails because of suppression of tail
2466 ** The system now hunts for the C variable "environ" in a more
2467 devious way, to avoid segfaults when the C library version differs
2468 between compile time and run time. (thanks to Christophe Rhodes)
2469 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2470 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2471 ** The compiler no longer issues bogus style warnings for undefined
2472 classes in the same source file as the DEFCLASSes which defined
2473 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2475 ** fixes in CONDITION class precedence list for undefined function
2476 errors (thanks to Alexei Dejneka)
2477 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2478 correctly. (thanks to Dan Barlow)
2479 ** portability fixes aiming at bootstrapping under CLISP (thanks
2480 to Dave McDonald and Christophe Rhodes)
2481 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2482 ** fixes in type translation and and type inference (thanks to
2484 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2485 ** various fixes in the new ports (thanks to Dan Barlow)
2486 * several changes related to debugging:
2487 ** suppression of tail recursion, as noted above
2488 ** stack overflow detection, as noted above
2489 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2490 is now the default. (For some time encapsulation has been more
2491 reliable than the breakpoint-based :ENCAPSULATE NIL
2492 implementation, at least on X86 systems; and I just noticed that
2493 encapsulation also seems closer to the spirit of the ANSI
2496 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2498 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2499 up properly again. (There was a packaging bug in 0.7.0 which
2500 left their definitions in SB-SYS::LOAD-FOREIGN and
2501 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2502 vital for most things which interface to C-level interfaces,
2503 like extensions working with sockets or databases or
2504 Perl-compatible regexes or whatever, and the need to fix
2505 this bug is the main reason that 0.7.1 was released so
2507 ** DEFGENERIC is now choosier about the methods it redefines, so that
2508 reLOADing a previously-LOADed file containing DEFGENERICs does
2509 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2510 cycle now works as it should. (thanks to Alexey Dejneka)
2511 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2512 fixed by Christophe Rhodes. (He actually submitted this patch
2513 months ago, and I delayed until after 0.7.0.)
2514 ** Bug 111 (internal compiler confusion about runtime checks on
2515 FUNCTION types) was fixed by Alexey Dejneka.
2516 * Some internal cleanups (getting rid of variables which aren't
2517 needed now that the byte interpreter is gone) caused the fasl
2518 file format number to change again.
2520 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2521 * major incompatible change: The default fasl file extension, i.e. the
2522 default extension for files produced by COMPILE-FILE, has changed
2523 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2525 ** There are many changes in the implementation of the compiler.
2526 SBCL is now essentially a compiler-only implementation of ANSI
2527 Common Lisp. EVAL still "interprets" a few special cases, but
2528 almost all the interesting cases are handled by creating
2529 a LAMBDA expression, calling COMPILE on it, then calling
2530 FUNCALL on the result.
2531 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2532 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2533 Since the code is newer, there might still be some new bugs
2534 (though not as many as before Martin Atzmueller's fixes:-). But
2535 the new code is substantially simpler and clearer, and hopefully
2536 any remaining bugs will be simpler, less fundamental, and more
2537 fixable then the bugs in the old code.
2538 ** The revised compiler is still a little unsteady on its feet.
2540 *** The debugging information it produces (particularly the names
2541 of FUNCTION objects) is sometimes much less useful than what
2542 the old compiler produced.
2543 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2544 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2545 within a MACROLET) has been temporarily weakened.
2546 ** There are new compiler optimizations for various functions:
2547 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2548 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2549 *** the math functions TRUNCATE, FLOOR, and CEILING
2550 *** the function-of-all-trades COERCE
2551 Mostly these should be transparent, but there's one
2552 potentially-annoying problem (bug 117): when the compiler
2553 inline-expands a function and does type analysis on the result,
2554 it can create control paths which have type mismatches, and
2555 when it can't prove that those control paths aren't taken,
2556 it will issue WARNINGs about the type mismatches. This is
2557 a particular problem in practice for the new sequence functions.
2558 It's not clear how this should be fixed, and for now, a
2559 workaround is given in the entry for 117 in the BUGS file.
2560 ** (Because of the interaction between the two previous items --
2561 occasional inlining problems and new inline expansions -- some
2562 of the new sequence function optimizations won't really kick in
2563 completely until debugging information, and then inlining, are
2564 straightened out in some future version.)
2565 * minor incompatible changes:
2566 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2567 DIRECTORY no longer implicitly promotes NIL slots of its
2568 pathname argument to :WILD. In particular, when you ask for the
2569 contents of a directory (which you used to be able to do without
2570 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2571 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2572 ** changes in behavior that ANSI explicitly defines to be
2573 implementation dependent:
2574 *** The new compiler-only implementation still conforms with ANSI,
2575 but acts a little different than before. Besides the obvious
2576 changes in performance tradeoffs (that the cost per form passed
2577 to EVAL has gone up, and the cost per form executed by EVAL
2578 has gone down), the behavior of the system changes a little
2579 because there are no longer any interpreted function objects.
2580 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2581 e.g. doing COMPILE on the output of interactive DEFUN is
2583 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2585 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2586 has changed from (:CL) to NIL.
2587 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2588 e.g. #\Newline and #\Tab, don't have names specified in the
2589 ANSI Common Lisp standard, is now based on their ASCII symbolic
2590 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2591 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2592 are no longer used for output.
2593 ** changes in internal implementation constants:
2594 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2595 4 million. (If your application spends a lot of time GCing and
2596 you have a lot of RAM, you might want to experiment with
2597 increasing it even more.)
2598 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2599 However, almost all old code should still continue to work without
2600 immediate update, as SB-C-CALL is now a (deprecated) nickname
2602 ** Old operator names in the style DEF-FOO are now deprecated in
2603 favor of new corresponding names DEFINE-FOO, for consistency with
2604 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2605 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2606 internal symbols, but a few supported extensions like
2607 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2608 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2609 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2610 etc. as you get deeper into recursive calls to the debugger
2611 command loop, instead of the old "5]", "5]]", "5]]]"
2612 sequence. (I was motivated to do this when squabbles between
2613 ILISP and SBCL left me very deeply nested in the debugger. In the
2614 short term, this change will probably provoke more ILISP/SBCL
2615 squabbles, but hopefully it will be an improvement in the long run.)
2616 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2617 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2618 function, since declaiming return types involving aliens is
2619 (1) annoyingly messy to do by hand and (2) vital to efficient
2620 compilation of code which calls such functions.
2621 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2622 longer reexported by the SB-EXT package. They're solely useful
2623 for alien code, so it seems more logical that you should get
2624 them from the SB-ALIEN package, not in SB-EXT.
2625 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2626 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2627 features. Instead, the code that they used to control is always
2628 built into the system.
2629 * many other bug fixes
2630 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2631 advantage of the new EVAL-WHEN stuff and to clean them up in
2632 general, and they are now more ANSI-compliant in a number of
2633 ways. Martin Atzmueller is responsible for a lot of this.
2634 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2636 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2637 *** correct ERROR type for various file operations
2638 *** some fixes for Lisp streams
2639 *** DEFMETHOD syntax checking
2640 *** changing old weird representation of debug information as
2641 strings (which, among their other deficiencies, don't transform
2642 correctly when you rename packages, and don't change their
2643 print representation when you change things like *PACKAGE*
2644 and *PRINT-LENGTH*) to symbols and lists of symbols
2645 He also made several improvements and fixed several bugs in DESCRIBE.
2646 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2648 *** misbehavior of WRITE-STRING/WRITE-LINE
2649 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2650 and several other LOOP problems as well
2651 *** DIRECTORY when similar filenames are present
2652 *** DEFGENERIC with :METHOD options
2653 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2654 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2655 *** argument ordering in FIND with :TEST option
2656 *** mishandled package designator argument in APROPOS-LIST
2657 *** various problems in the backquote readmacro
2658 *** a bug in APROPOS
2659 *** probably some others that I'm not describing very well here,
2660 since the CVS log documents them by reference to sbcl-devel
2661 messages, and the SourceForge archives aren't working well.:-(
2662 ** Dan Barlow improved the Alpha port (and is making progress on the
2663 PPC port, for those of you who think different).
2664 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2665 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2666 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2667 continues to work on the SPARC port (for those of you in a position
2668 to look down upon our little PC-compatible boxes from a great height).
2669 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2670 and uses that copy, rather than the current dynamic value, when
2671 it's trying to decide whether to truncate output. Thus e.g.
2672 (let ((*print-lines* 50))
2673 (pprint-logical-block (stream nil)
2675 (let ((*print-lines* 8))
2676 (print (aref possiblybigthings i) stream)))))
2677 should now truncate the logical block only at 50 lines, instead of
2678 often truncating it at 8 lines, as it did before.
2679 * The doc/cmucl/ directory, containing old CMU CL documentation
2680 from the time of the fork, is no longer part of the base system.
2681 SourceForge has shut down its anonymous FTP service, and with it
2682 my original plan for distributing the old CMU CL documentation
2683 there. For now, if you need these files you can download an old
2684 SBCL source release and extract them from it.
2685 * The fasl file version number changed again, for dozens of reasons,
2686 some of which are apparent above.
2688 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2689 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2690 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2691 patch, so that DISASSEMBLE output is much nicer.
2692 * The code in the SB-PROFILE package now seems reasonably stable.
2693 I still haven't decided what the final interface should look like
2694 (I'd like PROFILE to interact cleanly with TRACE, since both
2695 facilities use function encapsulation) but if you have a need
2696 for profiling now, you can probably use it successfully with
2697 the current CMU-CL-style interface.
2698 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2699 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2700 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2701 link now returns the dangling link itself, and for similar
2702 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2703 link itself. (In these cases the old code signalled an error and
2704 looped endlessly, respectively.) Thus, DIRECTORY now works even
2705 in the presence of dangling and cyclic symbolic links.
2706 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2707 is now a supported extension again, since the consensus on
2708 sbcl-devel was that it can be useful for ordinary development
2709 work, not just for debugging SBCL itself.
2710 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2711 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2712 that currently-defined functions might be redefined later with
2713 different return types.
2714 * Hash tables can be printed readably, as inspired by CMU CL code
2715 of Eric Marsden and SBCL code of Martin Atzmueller.
2716 * better error handling in CLOS method combination, thanks to
2717 Martin Atzmueller porting Pierre Mai's CMU CL patches
2718 * more overflow fixes for >16Mbyte I/O buffers
2719 * A bug in READ has been fixed, so that now a single Ctrl-D
2720 character suffices to cause end-of-file on character streams.
2721 In particular, now you only need one Ctrl-D at the command
2722 line (not two) to exit SBCL.
2723 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2725 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2726 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2727 does the right thing.
2728 * The implementation of some type tests, especially for CONDITION
2729 types, is now tidier and maybe faster, due to CMU CL code
2730 originally by Douglas Crosher, ported by Martin Atzmueller.
2731 * Some math functions have been fixed, and there are new
2732 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2733 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2734 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2735 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2736 likely to be built into sbcl-0.7.0.)
2737 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2738 CL, ported to SBCL by Martin Atzmueller)
2739 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2740 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2741 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2742 * There's a new slam.sh hack to shorten the edit/compile/debug
2743 cycle for low-level changes to SBCL itself, and a new
2744 :SB-AFTER-XC-CORE target feature to control the generation of
2745 the after-xc.core file needed by slam.sh.
2746 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2747 COMPILE-FILE is no longer supported, so that now every function
2748 gets an entry point, so that block compilation looks a little
2749 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2750 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2751 returns the number of bytes consed since the system started,
2752 rather than the number consed since the first time the function
2753 was called. (The new definition parallels ANSI functions like
2754 CL:GET-INTERNAL-RUN-TIME.)
2755 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2756 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2757 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2758 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2759 * Fasl file version numbers are now independent of the target CPU,
2760 since historically most system changes which required version
2761 number changes have affected all CPUs equally. Similarly,
2762 the byte fasl file version is now equal to the ordinary
2765 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2766 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2767 is no longer recognized. I apologize for this, because it was
2768 listed in SB-EXT as a supported extension, but I found that
2769 its existing behavior was poorly specified, as well as incorrectly
2770 specified, and it looked like too much of a mess to straighten it
2771 out. I have enough on my hands trying to get ANSI stuff to work..
2772 * many patches ported from CMU CL by Martin Atzmueller, with
2773 half a dozen bug fixes in pretty-printing and the debugger, and
2774 half a dozen others elsewhere
2775 * fixed bug 13: Floating point infinities are now supported again.
2776 They might still be a little bit flaky, but thanks to bug reports
2777 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2778 as flaky as they were.
2779 * The --noprogrammer command line option is now supported. (Its
2780 behavior is slightly different in detail from what the old man
2781 page claimed it would do, but it's still appropriate under the
2782 same circumstances that the man page talks about.)
2783 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2784 are now supported, and enabled by default. Thus, the compiler can
2785 handle many floating point and complex operations much less
2786 inefficiently. (Thus e.g. you can implement a complex FFT
2788 * The compiler now detects type mismatches between DECLAIM FTYPE
2789 and DEFUN better, and implements CHECK-TYPE more correctly, and
2790 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2791 thanks to patches from Martin Atzmueller.
2792 * various fixes to make the cross-compiler more portable to
2793 ANSI-conforming-but-different cross-compilation hosts (notably
2794 Lispworks for Windows, following bug reports from Arthur Lemmens)
2795 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2796 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2797 fixed thanks to Pierre Mai's CMU CL patches.
2798 * fixes to keep the system from overflowing internal counters when
2799 it tries to use i/o buffers larger than 16M bytes
2800 * fixed bug 45a: Various internal functions required to support
2801 complex special functions have been merged from CMU CL sources.
2802 (When I was first setting up SBCL, I misunderstood a compile-time
2803 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2804 * improved support for type intersection and union, fixing bug 12
2805 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2806 more obscure bugs as well
2807 * some steps toward byte-compiling non-performance-critical
2808 parts of the system, courtesy of patches from Martin Atzmueller
2809 * Christophe Rhodes has made some debian packages of sbcl at
2810 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2811 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2812 stable, but are nonetheless usable. When he's ready, I'd be happy
2813 to add them to the SourceForge "File Releases" section. (And if
2814 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2815 * new fasl file format version number (because of changes in
2816 internal representation of (OR ..) types to accommodate the new
2817 support for (AND ..) types, among other things)
2819 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2820 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2822 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2824 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2825 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2826 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2827 * DESCRIBE now works on structure objects again.
2828 * Most function call argument type mismatches are now handled as
2829 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2830 know whether the function will be redefined before the call is
2831 executed. (The compiler could flag local calls with full WARNINGs,
2832 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2833 it doesn't keep track of enough information to know whether calls
2834 are local in this sense.)
2835 * Compiler output is now more verbose, with messages truncated
2836 later than before. (There should be some supported way for users
2837 to override the default verbosity, but I haven't decided how to
2838 provide it yet, so this behavior is still controlled by the internal
2839 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2840 src/compiler/ir1util.lisp.)
2841 * Fasl file format version numbers have increased again, because
2842 support for the Gray streams extension changes the layout of the
2843 system's STREAM objects.
2844 * The Gray subclassable streams extension now works, thanks to a
2845 patch from Martin Atzmueller.
2846 * The full LOAD-FOREIGN extension (not just the primitive
2847 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2848 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2849 but like most other programs, it defaults to copying the Unix
2850 environment from the original process instead of starting the
2851 new process in an empty environment.
2852 * Extensions which manipulate the Unix environment now support
2853 an :ENVIRONMENT keyword option which doesn't smash case or
2854 do other bad things. The CMU-CL-style :ENV option is retained
2855 for porting convenience.
2856 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2857 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2859 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2861 * A patch from Martin Atzmueller seems to have solved the SIGINT
2862 problem, and as far as we know, signal-handling now works cleanly.
2863 (If you find any new bugs, please report them!)
2864 * The system no longer defaults Lisp source file names to types
2865 ".l", ".cl", or ".lsp", but only to ".lisp".
2866 * The compiler no longer uses special default file extensions for
2867 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2868 seems to expect a single default extension for all compiled code,
2869 and there's no compelling reason to try to stretch the standard
2870 to allow two different extensions.) Instead, byte-compiled files
2871 default to the same extension as native-compiled files.
2872 * Fasl file format version numbers have increased again, because
2873 a rearrangement of internal implementation packages made some
2874 dumped symbols in old fasl files unreadable in new cores.
2875 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2876 many fewer weird special cases.
2877 * Bug #17 (differing COMPILE-FILE behavior between logical and
2878 physical pathnames) has been fixed, and some related misbehavior too,
2879 thanks to a patch from Martin Atzmueller.
2880 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2881 by Tim Moore, ported to SBCL by Martin Atzmueller.
2882 * Martin Atzmueller fixed several filesystem-related problems,
2883 including bug #36, in part by porting CMU CL patches, which were
2884 written in part by Paul Werkowski.
2885 * More compiler warnings in src/runtime/ are gone, thanks to
2886 more patches from Martin Atzmueller.
2887 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2890 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2892 * DESCRIBE now works on CONDITION objects.
2893 * The debugger now handles errors which arise when trying to print
2894 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2896 * The build system now uses an additional file, customize-target-features.lisp,
2897 to allow local modifications to the target *FEATURES* list. (The point of
2898 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2899 debugging features enabled, without having to worry about propagating it
2900 into everyone's system when I do a "cvs update".) When no
2901 customize-target-features.lisp file exists, the target *FEATURES* list
2902 should be constructed the same way as before.
2903 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2904 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2905 warn about behavior which is undefined under the ANSI spec. Note: This
2906 is specified by ANSI, but it's not very popular with programmers.
2907 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2908 macro in the SBCL sources for an example of a workaround which you
2909 might use to make portable ANSI-standard code which does what you want.
2910 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2911 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2912 when it's used as a non-toplevel form. (This is required in order
2913 to implement the DEFCONSTANT-EQX macro.)
2914 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2915 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2916 funny things, due to bugs in the implementation of EVAL-WHEN
2917 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2918 non-toplevel DEFCONSTANTs are uncommon.)
2919 * The core file version number and fasl file version number have been
2920 incremented, because the old noncompliant DEFCONSTANT behavior involved
2921 calling functions which no longer exist, and because I also took the
2922 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2923 * fixed bug 1 (error handling before read-eval-print loop starts), and
2924 redid debugger restarts and related debugger commands somewhat while
2926 ** The QUIT debugger command is gone, since it did something
2927 rather different than the SB-EXT:QUIT command, and since it never
2928 worked properly outside the main toplevel read/eval/print loop.
2929 Invoking the new TOPLEVEL restart provides the same functionality.
2930 ** The GO debugger command is also gone, since you can just invoke
2931 the CONTINUE restart directly instead.
2932 ** The TOP debugger command is also gone, since it's redundant with the
2933 FRAME 0 command, and since it interfered with abbreviations for the
2935 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2937 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2938 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2939 some porting work by Martin Atzmueller)
2940 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2941 with a boatload of patches which he ported to SBCL. Now that those
2943 ** The system tries to make sure that its low-priority messages
2944 are prefixed by semicolons, to help people who like to use
2945 syntax highlighting in their ILISP buffer. (This patch
2946 was originally due to Raymond Toy.)
2947 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2948 patches originally written by Raymond Toy.
2949 ** The compiler understands coercion between single-value and
2950 multiple-VALUES type expressions better, getting rid of some very
2951 weird behavior, thanks to patches originally by Robert MacLachlan
2952 and Douglas Crosher.
2953 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2954 lambda lists, thanks to a patch originally by Pierre Mai.
2955 ** The system no longer bogusly warns about "abbreviated type
2957 ** The compiler gets less confused by inlining and RETURN-FROM,
2958 thanks to some patches originally by Tim Moore.
2959 ** The system no longer hangs when dumping circular lists to fasl
2960 files, thanks to a patch originally from Douglas Crosher.
2961 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2962 undefined function error.
2963 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2964 in structure slot initforms) for now, documented workaround instead:-|
2965 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2966 * fixed bug 5: added stubs for various Gray stream functions called
2967 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2968 installed, at least appropriate type errors are generated
2969 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2970 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2971 consistently in DEFMETHOD forms.
2972 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2973 it doesn't seem to affect SBCL after all
2974 * The C runtime system now builds with better optimization and many
2975 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2977 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2979 * The system is now under CVS at SourceForge (instead of the
2980 CVS repository on my home machine).
2981 * The new signal handling code has been tweaked to treat register
2982 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2983 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2984 downstream, e.g. in debug-int.lisp, has implicit dependencies
2985 on the unsignedness of integer representation of machine words,
2986 and that caused the system to bomb out with infinite regress
2987 when trying to recover from type errors involving signed values,
2988 e.g. (BUTLAST '(1 2 3) -1).)
2989 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2990 (This was one of the bugs Peter Van Eynde reported back in July.)
2991 * The system now uses code inspired by Colin Walters' O(N)
2992 implementation of MAP (from the cmucl-imp@cons.org mailing
2993 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2994 inline the MAP operation, and there is more than one
2995 sequence argument to the MAP call (so that it can't just
2996 do ETYPECASE once and for all based on the type of the
2997 single sequence argument). (The old non-inline implementation
2998 of the general M-argument sequence-of-length-N case required
2999 O(M*N*N) time when any of the sequence arguments were LISTs.)
3000 * The QUIT :UNIX-CODE keyword argument has been renamed to
3001 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3002 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3004 * Raymond Wiker's patches to port dynamic loading from Linux to
3005 FreeBSD have been added.
3006 * The BUGS file is now more nearly up to date, thanks in large part
3007 to Martin Atzmueller's review of it.
3008 * The debugger now flushes standard output streams before it begins
3009 its output ("debugger invoked" and so forth).
3010 * The core version number and fasl file version number have both
3011 been incremented, because of incompatible changes in the layout
3013 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3014 used to not be called for a saved Lisp image.)
3015 * Martin Atzmueller's version of a patch to fix a compiler crash,
3016 as posted on sbcl-devel 13 September 2000, has been installed.
3017 * Instead of installing Martin Atzmueller's patch for the
3018 compiler transform for SUBSEQ, I deleted the compiler transform,
3019 and transforms for some similar consing operations.
3020 * A bug in signal handling which kept TRACE from working on OpenBSD
3022 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3023 away by constant folding
3024 * The system now defines its address space constants in one place
3025 (in the Lisp sources), and propagates them automatically elsewhere
3026 (through GENESIS and the sbcl.h file). Therefore, patching the
3027 address map is less unnecessarily tedious and error-prone. The
3028 Lisp names of address space constants have also been systematized.
3029 * CVS tags like dollar-Header-dollar have been removed from
3030 the sources, because they have never saved me trouble and
3031 they've been source of trouble working with patches and other
3032 diff-related operations.
3033 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3034 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3036 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3038 * The system has been ported to OpenBSD.
3039 * The system now compiles with a simple "sh make.sh" on the systems
3040 that it's supported on. I.e., now you no longer need to tweak
3041 text in the target-features.lisp-expr and symlinks in src/runtime/
3042 by hand, the make.sh takes care of it for you.
3043 * The system is no longer so grossly inefficient when compiling code
3044 involving vectors implemented as general (not simple) vectors (VECTOR T),
3045 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3046 which dares to use the various sequence functions on non-simple
3047 vectors, takes less of a performance hit.
3048 * There is now a primitive type predicate VECTOR-T-P
3049 to test for the (VECTOR T) type, so that e.g.
3050 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3051 can now be compiled with some semblance of efficiency. (The old code
3052 turned the type declaration into a full call to %TYPEP at runtime!)
3053 * AREF on (VECTOR T) is still not fast, since it's still compiled
3054 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3055 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3056 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3057 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3058 without making full calls to SUBTYPEP at runtime.
3059 (Some analogous efficiency issues for non-simple vectors specialized to
3060 element types other than T, or for non-simple multidimensional arrays,
3061 have not been addressed. They could almost certainly be handled the
3062 same way if anyone is motivated to do so.)
3063 * The changes in array handling break binary compatibility, so
3064 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3065 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3066 instead of (VALUES T T).
3067 * By following the instructions that Dan Barlow posted to sbcl-devel
3068 on 2 July 2000, I was able to enable primitive dynamic object
3069 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3070 functionality is not implemented (since it calls ld to resolve
3071 library references automatically, requiring RUN-PROGRAM for its
3072 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3073 try to resolve library references) is now supported.
3074 * The system now flushes the standard output streams when it terminates,
3075 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3076 them at several other probably-convenient times, e.g. in each pass of
3077 the toplevel read-eval-print loop, and after evaluating a form given
3078 as an "--eval" command-line option. (These changes were motivated by a
3079 discussion of stream flushing issues on cmucl-imp in August 2000.)
3080 * The source transform for TYPEP of array types no longer assumes
3081 that an array whose element type is a not-yet-defined type
3082 is implemented as an array of T, but instead punts, so that the
3083 type will be interpreted at runtime.
3084 * There is now some support for cross-compiling in make.sh: each of
3085 the phases of make.sh has its own script. (This should be transparent
3086 to people doing ordinary, non-cross-compile builds.)
3087 * Since my laptop doesn't have hundreds of megabytes of memory like
3088 my desktop machine, I became more motivated to do some items on
3089 my to-do list in order to reduce the size of the system a little:
3090 ** Arrange for various needed-only-at-cold-init things to be
3091 uninterned after cold init. To support this, those things have
3092 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3093 symbols with such names are now uninterned after cold init).
3094 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3095 fasl files for cold load.
3096 ** Remove the old compiler structure pooling code (which used to
3097 be conditional on the target feature :SB-ALLOC) completely.
3098 ** Redo the representation of some data in cold init to be more compact.
3099 (I also looked into supporting byte compiled code at bootstrap time,
3100 which would probably reduce the size of the system a lot, but that
3101 looked too complicated, so I punted for now.)
3102 * The maximum signal nesting depth in the src/runtime/ support code has
3103 been reduced from 4096 to 256. (I don't know any reason for the very
3104 large old value. If the new smaller value turns out to break something,
3105 I'll probably just bump it back up.)
3106 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3108 * Many, many bugs reported by Peter Van Eynde have been added to
3109 the BUGS list; some have even been fixed.
3110 * While enabling dynamic object file loading, I tried to make the
3111 code easier to understand, renaming various functions and variables
3112 with less ambiguous names, and changing some function calling
3113 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3114 * While trying to figure out how to do the OpenBSD port, I tried to
3115 clean up some of the code in src/runtime/. In particular, I dropped
3116 support for non-POSIX signal handling, added various comments,
3117 tweaked the code to reduce the number of compilation warnings, and
3118 renamed some files to increase consistency.
3119 * To support the new automatic configuration functionality in make.sh,
3120 the source file target-features.lisp-expr has been replaced with the
3121 source file base-target-features.lisp-expr and the machine-generated
3122 file local-target-features.lisp-expr.
3123 * fixed a stupid quoting error in make.sh so that using CMU CL
3124 "lisp -batch" as cross-compilation host works again
3126 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3128 * DESCRIBE no longer tries to call itself recursively to describe
3129 bound/fbound values, so that it no longer fails on symbols which are
3130 bound to themselves (like keywords, T, and NIL).
3131 * DESCRIBE now works on generic functions.
3132 * The printer now prints less-screwed-up representations of closures
3133 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3134 * A private symbol is used instead of the :EMPTY keyword previously
3135 used to mark empty slots in hash tables. Thus
3136 (DEFVAR *HT* (MAKE-HASH-TABLE))
3137 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3138 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3139 now does what ANSI says that it should. (You can still get
3140 similar noncompliant behavior if bang on the hash table
3141 implementation with all the symbols you get back from
3142 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3143 This breaks binary compatibility, since tests for equality to
3144 :EMPTY are wired into things like the macroexpansion of
3145 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3147 * There's now a minimal placeholder implementation for CL:STEP,
3148 as required by ANSI.
3149 * An obscure bug in the interaction of the normal compiler, the byte
3150 compiler, inlining, and structure predicates has been patched
3151 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3152 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3154 * Missing ordinary arguments in a macro call are now detected even
3155 when the macro lambda list contains &KEY or &REST.
3156 * The debugger no longer complains about encountering the top of the
3157 stack when you type "FRAME 0" to explicitly instruct it to go to
3158 the top of the stack. And it now prints the frame you request even
3159 if it's the current frame (instead of saying "You are here.").
3160 * As specified by ANSI, the system now always prints keywords
3161 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3163 * The default initial SIZE of HASH-TABLEs is now smaller.
3164 * Type information from CLOS class dispatch is now propagated
3165 into DEFMETHOD bodies, so that e.g.
3166 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3168 is now basically equivalent to
3169 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3170 (DECLARE (TYPE SINGLE-FLOAT X))
3172 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3173 operation, without having to do run-time type dispatch.
3174 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3175 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3176 * Since I don't seem to be making big file reorganizations very often
3177 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3178 is overflowing my ability to conveniently back them up), I've finally
3179 checked the system into CVS. (The CVS repository is on my home system,
3180 not at SourceForge -- putting it on SourceForge might come later.)
3181 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3182 high-level GC-NOTIFY-FOO functions send their output. (There's
3183 still very little control of where low-level verbose GC functions
3184 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3185 less than it used to -- the GC-NOTIFY-FOO functions are now under
3186 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3187 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3188 in only one place in the source code, and propagates it automatically
3189 everywhere that it's needed. Thus e.g. when I bump the version from
3190 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3191 * The C source files now include boilerplate legalese and documentation
3192 at the head of each file (just as the Lisp source files already did).
3193 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3194 to his page will be replaced with a link to his new CLiki service.
3196 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3198 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3199 * The build process now looks for GNU make under the default name "gmake",
3200 instead of "make" as it used to. If GNU make is not available as "gmake"
3201 on your system, you can change this default behavior by setting the
3202 GNUMAKE environment variable.
3203 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3204 can build without error under CMU CL.
3206 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3208 * There is now a partial SBCL user manual (with some new text and some
3209 text cribbed from the CMU CL manual).
3210 * The beginnings of a profiler have been added (starting with the
3211 CMU CL profiler and simplifying and cleaning up). Eventually the
3212 main interface should be through the TRACE macro, but for now,
3213 it's still accessed through vaguely CMU-CL-style functions and macros
3214 exported from the package SB-PROFILE.
3215 * Some problems left over from porting CMU CL to the new
3216 cross-compilation bootstrap process have been cleaned up:
3217 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3218 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3219 part when getting rid of the compiler *BACKEND* stuff.)
3220 ** The value of *NULL-TYPE* was screwed up, because it was
3221 being initialized before the type system knew the final
3222 definition of the 'NULL type. This screwed up several key
3223 optimizations in the compiler, causing inefficiency in all sorts
3224 of places. (I found it because I wanted to understand why
3225 GET-INTERNAL-RUN-TIME was consing.)
3226 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3227 PROCLAIM FTYPE information. Unfortunately this broke binary
3228 compatibility again, since now the forms output by DEFGENERIC
3229 to refer to functions which didn't exist in 0.6.3.
3230 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3231 can use the new (as of 0.6.3) transform for SOME into MAP into
3233 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3234 values to (INTEGER 0 1000000), so that the time code will no longer
3235 occasionally get blown up by Linux returning 1000000 microseconds
3236 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3237 its output conform to the ANSI spec. (Alas, this makes its output
3238 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3239 * A full call to MAP NIL with a single sequence argument no longer conses.
3240 * fixes to problems pointed out by Martin Atzmueller:
3241 * The manual page no longer talks about multiprocessing as though
3242 it were currently supported.
3243 * The ILISP support patches have been removed from the distribution,
3244 because as of version 5.10.1, ILISP now supports SBCL without us
3245 having to maintain patches.
3246 * added a modified version of Raymond Toy's recent CMU CL patch for
3247 EQUALP comparison of HASH-TABLE
3249 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3251 * The system still can't cross-compile itself with
3252 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3253 entails), but at least it can compile more of itself that way
3254 than it used to be able to, and various buglets which were uncovered
3255 by trying to cross-compile itself that way have now been fixed.
3256 * This release breaks binary compatibility again. This time
3257 at least I've incremented the FASL file format version to 2, so that the
3258 problem can be detected reliably instead of just causing weird errors.
3259 * various new style warnings:
3260 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3261 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3262 ** using the lexical binding of a variable named in the *FOO* style
3263 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3264 as specified by ANSI.
3265 * *RANDOM-STATE* is no longer automatically initialized from
3266 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3267 default behavior of the system is to repeat its behavior every time
3268 it's run. If you'd like to change this behavior, you can always
3269 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3270 old convention there was no comparably easy way to get the system to
3271 repeat its behavior every time it was run.
3272 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3273 FTYPE declarations has been removed, in favor of their ANSI
3274 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3275 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3276 the types of their sequence arguments can be determined at compile time.
3277 This is done through a new open code expansion for MAP which eliminates
3278 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3279 argument types can be determined at compile time.
3280 * The optimizer now transforms COERCE into an identity operation when it
3281 can prove that the coerced object is already of the correct type. (This
3282 can be a win for machine generated code, including the output of other
3283 optimization transforms, such as the MAP transform above.)
3284 * Credit information has been moved from source file headers into CREDITS.
3285 * Source file headers have been made more standard.
3286 * The CASE macro now compiles without complaining even when it has
3289 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3291 * (Note that the way that the PCL macroexpansions were rewritten
3292 to accommodate the change in DEFGENERIC below breaks binary
3293 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3294 not run under sbcl-0.6.2. Once we get out of alpha releases,
3295 i.e. hit release 1.0.0, we'll probably try to maintain binary
3296 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3297 and sbcl-1.4.4. Until then, however, it might be fairly common
3298 for maintenance releases to break binary compatibility.)
3299 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3301 * The reporting of unhandled signals has been changed to print some
3302 explanatory text as well as the report form. (Previously only
3303 the report form was printed.)
3304 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3305 it defines, so that the compiler no longer issues undefined function
3306 warnings for compiled-but-not-yet-loaded generic functions.
3307 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3308 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3309 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3310 section of the ANSI spec are supported.
3311 * The "" nickname for the "KEYWORD" package has been removed.
3312 The reader still handles symbol tokens which begin with a package marker
3313 as keywords, but it doesn't expose its mechanism for doing so in the
3314 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3315 * The system now issues STYLE-WARNINGs for contradictory TYPE
3316 proclamations. (Warnings for contradictory FTYPE proclamations would
3317 be nice too, but those can't be done usefully unless the type system
3318 is made smarter about FUNCTION types.)
3319 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3320 systematized, so that "*target-*.lisp is supposed to exist only on the
3321 target and imply that there's a related file which exists on the
3322 host, and *host-*.lisp is supposed to exist only on the host and imply
3323 that there's a related file which exists on the target. This involves a
3324 lot of renaming. Hopefully the acute confusion caused by the renaming
3325 will be justified by the reduction in chronic confusion..
3326 ** runtime-type.lisp -> early-target-type.lisp
3327 ** target-type.lisp -> late-target-type.lisp
3328 ** early-host-format.lisp -> early-format.lisp
3329 ** late-host-format.lisp -> late-format.lisp
3330 ** host-error.lisp -> misc-error.lisp
3331 ** early-error.lisp -> early-target-error.lisp
3332 ** late-error.lisp -> late-target-error.lisp
3333 ** host-defboot.lisp -> early-defboot.lisp
3334 ** code/misc.lisp -> code/target-misc.lisp
3335 ** code/host-misc.lisp -> code/misc.lisp
3336 ** code/numbers.lisp -> code/target-numbers.lisp
3337 ** code/early-numbers.lisp -> numbers.lisp
3338 ** early-host-type.lisp -> early-type.lisp
3339 ** late-host-type.lisp -> late-type.lisp
3340 ** host-typep.lisp -> typep.lisp
3341 ** load.lisp -> target-load.lisp
3342 ** host-load.lisp -> load.lisp
3343 ** host-disassem.lisp -> disassem.lisp
3344 ** host-insts.lisp -> insts.lisp
3345 ** byte-comp.lisp -> target-byte-comp.lisp
3346 ** host-byte-comp.lisp -> byte-comp.lisp
3347 ** host-signal.lisp -> signal.lisp
3348 ** host-defstruct.lisp -> defstruct.lisp
3349 ** late-target-type.lisp -> deftypes-for-target.lisp
3350 Furthermore, several other previously target-only files foo.lisp (e.g.
3351 hash-table.lisp and random.lisp) have been split into a target-and-host
3352 foo.lisp file and a target-only target-foo.lisp file, with their key type
3353 definitions in the target-and-host part, so that the cross-compiler will
3354 know more about target types.
3355 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3356 gone away. In their place are various *BACKEND-FOO* variables
3357 corresponding to the slots of the old structure.
3358 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3359 into the target SBCL has been fixed.
3360 * The chill.lisp system for loading cold code into a running SBCL
3362 * Support for the CMU CL "scavenger hook" extension has been removed.
3363 (It was undocumented and unused in the CMU CL sources that SBCL was
3364 derived from, and stale in sbcl-0.6.1.)
3365 * Various errors in the cross-compiler type system were detected
3366 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3367 (enabling various consistency checks). Many of them were fixed,
3368 but some hard problems remain, so the compiler is back to
3369 running without *TYPE-SYSTEM-INITIALIZED* for now.
3370 * As part of the cross-compiler type system cleanup, I implemented
3371 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3372 * I have started adding UNCROSS calls throughout the type system
3373 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3374 on cross-compiler input files will be able to go away, and various
3376 * CONSTANTP now returns true for quoted forms (as explicitly required
3379 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3381 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3382 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3383 (MAKE-LIST -1). In the longer term, it ought to become true
3384 that declarations are assertions even at SAFETY 1. For now, it's not
3385 quite true even at SAFETY 3, but it's at least more nearly true..
3386 (Note that this change seems to increases the size of the system by
3387 O(5%) and to decrease the speed of the compiler by 20% or more.)
3388 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3389 for the problem of printing dozens of lines of distracting information
3390 about low-level system machinery as part of the top stack frame
3391 on entry to the debugger when an undefined function was called.
3392 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3393 is not reset to COMMON-LISP-USER.
3394 * Compilation of stuff related to dyncount.lisp has been made conditional
3395 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3396 smaller. The various dyncount-related symbols have been moved into
3397 a new "SB-DYNCOUNT" package.
3398 * tty-inspect.lisp has been renamed to inspect.lisp.
3399 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3400 feature has gone away. (When we eventually port to other flavors of
3401 libc and/or Unix, we'll try to make the differences between flavors
3402 invisible at the user level.)
3403 * Various other *FEATURES* tags, and/or their associated conditionals,
3404 have been removed if obsolescent, or given better documentation, or
3405 sometimes given more-mnemonic names.
3407 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3409 * tidied up "make.sh" script
3410 * tidied up system directory structure
3411 * better "clean.sh" behavior
3412 * added doc/FOR-CMUCL-DEVELOPERS
3413 * many many small tweaks to output format, e.g. removing possibly-confusing
3414 trailing #\. character in DESCRIBE-INSTANCE
3415 * (EQUALP #\A 'A) no longer signals an error.
3416 * new hashing code, including EQUALP hashing
3417 * tidied up Lisp initialization and toplevel
3418 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3419 * command line argument processing
3420 * added POSIX-GETENV function to deal with Unix-ish environment variables
3421 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3422 terminating SBCL on EOF
3423 * non-verbose GC by default
3424 * There is no more "sbcl" shell script; the sbcl file is now the C
3425 runtime executable (just like CMU CL).
3426 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3428 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3429 symbol and package information as Lisp native symbol and package objects
3430 instead of strings naming symbols and strings naming packages. This way,
3431 whenever packages are renamed (as in warm init), debug information is
3432 transformed along with everything else.
3433 * tweaked the optimization policy declarations which control the building
3434 of SBCL itself. Now, among other things, the system no longer saves
3435 source location debugging information. (This helps two problems at once
3436 by reducing SBCL size and by keeping SBCL from trying to look for its
3437 sources -- which may not exist -- when reporting errors.)
3438 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3439 debugging and testing purposes
3440 * cleaned up printing, making the printer call PRINT-OBJECT for
3441 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3442 methods, giving nearly-ANSI behavior
3443 * converted almost all special variables to use *FOO* naming convention
3444 * deleted PARSE-TIME functionality, since it can be done portably
3445 * moved some files out of cold init into warm init
3446 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3448 * regularized formatting of source files
3449 * added an install.sh script
3450 * fixed ridiculous memory usage of cross-compiler by making
3451 compiler/alloc.lisp not try to do pooling unless it can hook
3452 itself into the GC of the cross-compilation host. Now the system
3453 builds nicely on my old laptop.
3454 * added :SB-ALLOC in target-features.lisp-expr
3455 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3456 was not implemented)
3457 * re-did condition handling and note reporting in the compiler. Notes
3458 are no longer handled by signalling conditions. Style warnings
3459 and warnings are handled more correctly and reported in such a way
3460 that it's easy to find one or the other in your output (so that you
3461 can e.g. figure out which of many problems caused COMPILE-FILE to
3463 * changed the severity of several compiler warnings from full WARNING
3464 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3465 compiler note reporting so that it doesn't use the condition system
3466 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3467 COMPILE-FILE command)
3468 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3469 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3470 no longer supported, since their implementation was deeply intertwingled
3471 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3472 * removed bogus "support" for compiler macros named (SETF FOO), and
3473 removed the compiler macro for SETF INFO (but only after making a fool
3474 of myself on the cmucl-imp mailing list by posting a bogus patch for
3475 DEFINE-COMPILER-MACRO..)
3476 * Compiled files containing forms which have side effects on the Lisp
3477 reader (such as DEFPACKAGE forms) are now handled more correctly.
3478 (Compiler queuing of top level lambdas has been suppressed by setting
3479 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3480 * deleted various currently-unused source files, e.g. gengc.lisp. They
3481 may be added back at some point e.g. when porting to other architectures,
3482 but until they are it's distracting to distribute them and to try to
3484 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3485 there were so many of them they're just distractions, and UNCROSS is
3486 known to be able to handle the current sources
3487 * moved PROFILE functionality into TRACE, so that it will be clear
3488 how the wrapping and unwrapping of functions when you profile them
3489 interacts with the wrapping and unwrapping of functions when you
3490 trace them. (Actually, the functionality isn't there yet, but at least
3491 the interface specification is there. Hopefully, the functionality will
3492 arrive with some future maintenance release.)
3493 * removed host-oops.lisp
3494 * changed signature of QUIT function to allow UNIX-CODE argument
3495 * fixed READ-SEQUENCE bug
3496 * tweaked verbose GC output so that it looks more like the progress
3497 output that ANSI specifies for functions like LOAD
3498 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3499 * added <http://sbcl.sourceforge.com> to the banner information printed by