1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
3 * added support for the following external formats: koi8-u,
4 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
5 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
6 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
7 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
8 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
9 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
10 * incompatible change: a threaded SBCL will no longer revert to
11 non-threaded mode on non-NPTL systems, but refuse to start entirely.
12 * bug fix: interrupts are disabled until startup is complete; no
13 more sigsegvs when receiving a signal to soon
14 * optimization: Faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
16 * bug fix: add a workaround for the memory randomization features in
17 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
18 workaround will only be in effect on systems with the proc filesystem
20 * bug fix: printing objects of type HASH-TABLE signals a
21 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
24 ** bug fix: parent thread now can be gc'ed even with a live
26 ** bug fix: binding a special with PROGV to no value is not
27 the same as that symbol not having been bound (thanks to
29 ** bug fix: binding specials is thread safe (thanks to
31 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
32 and SB-SPROF do not die with 'no handler for signal XX in
33 interrupt_handle_now(..)' anymore
34 * fixed some bugs revealed by Paul Dietz' test suite:
35 ** ENSURE-GENERIC-FUNCTION should take a method class object for
36 the :method-class keyword argument.
38 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
39 * new port: the Solaris operating system on x86 processors is now
40 mostly supported, though some rough edges in the environment
41 remain. (thanks to Daisuke Homma)
42 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
43 space for its heap. (thanks to Thiemo Seufer)
44 * optimization: SBCL on MIPS platforms now supports dynamic-extent
45 closures. (thanks to Thiemo Seufer)
46 * minor incompatible change: eof selects abort in the debugger.
47 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
48 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
50 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
51 deprecated, and will go away in a future revision of SBCL.
52 * minor incompatible change: GC-ON and GC-OFF are no longer
53 implemented with a counter, it does not matter how many times gc
55 * bug fix: discriminating functions for generic function classes
56 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
57 make invalid assumptions about method precedence order. (reported
59 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
60 platforms. (thanks to Thiemo Seufer)
61 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
62 * bug fix: correct pathname printing with printer escaping is on.
63 (thanks to Kevin Reid)
64 * bug fix: complex VOP definitions in "user-space" no longer trigger
65 package locks. (reported by Zach Beane)
66 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
67 for funcallable-instance functions, and is no different from
69 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
70 characters in the current readtable. (reported by Nicholas Neuss)
71 * bug fix: SUBTYPEP on various CONS types returns more a more
72 accurate acknowledgment of its certainty.
73 * optimizations: REMOVE-DUPLICATES now runs in linear time on
74 lists in some cases. This partially fixes bug 384.
75 * flush all standard streams before prompting in the REPL and the
77 * bug fix: signal handling and triggering gc do not conflict
78 directly anymore, in particular a high frequency sb-sprof does
79 not prevent gc from running
80 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
81 approximation for timezone and DST information between the
82 universal time epoch and the smallest negative 32-bit time_t.
83 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
84 year 1899 so long as the encoded time is non-negative.
85 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
86 the correct answer for non-constant rotate amounts when performing
87 an inline 32-bit rotation.
89 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
90 there is only one thread in the session
91 ** bug fix: memory leak for streams created in one thread and
93 ** bug fix: lockup when compiled with gcc4
94 ** bug fix: race that allows the gc to be triggered when gc is
96 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
97 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
99 ** bug fix: finished threads are gc'ed properly
100 * fixed some bugs revealed by Paul Dietz' test suite:
101 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
102 has arguments with a different set of applicable methods from
103 the orignal arguments.
104 ** The type error thrown by MAP now has a correct expected-type
106 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
107 name a compiled function.
108 ** Three MISC tests where a large bignum was improperly coerced to
109 a float that couldn't represent that bignum during type
110 derivation were fixed.
111 ** SUBTYPEP can now handle types involving the negation of a
112 list-form FUNCTION type.
113 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
114 as (COMPLEX (AND RATIO (NOT FIXNUM))).
115 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
117 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
118 * New feature: Experimental support for bivalent streams: streams
119 opened with :element-type :default now allow character and binary
120 (unsigned-byte 8) I/O
121 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
122 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
124 * bug fix: no more highly sporadic "couldn't check whether ~S is
125 readable" when reading a stream and an interrupt hits in the middle
126 of a select system call
127 * compiler better recognizes complex arrays (reported by Eduardo
129 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
130 for variables on SBCL built with linkage-tables. (reported by Luis
132 * various error reporting improvements.
133 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
134 (Thanks to Thiemo Seufer)
135 * bug fix: nested reader invokations work correctly
136 * bug fix: it is possible to have simultaneous references to foreign
137 code and foreign data with the same name.
139 ** added x86-64 support
140 ** incompatible change: the threading api now works with thread
141 objects instead of thread ids
142 ** bug fix: threads are protected from signals and interruption when
143 starting up or going down
144 ** bug fix: a race where an exiting thread could lose its stack to gc
145 ** bug fix: don't halt on infinite error in threads if possible
146 ** fixed numerous gc deadlocks introduced in the pthread merge
147 ** bug fix: fixed thread safety issues in read and print
148 ** bug fix: debugger doesn't hang on session lock if interrupted at
149 an inappropriate moment
150 ** bug fix: run-program is now thread safe(r)
151 ** bug fix: inner with-recursive-lock no longer releases the mutex
152 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
153 * fixed some bugs revealed by Paul Dietz' test suite:
154 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
155 EXPECTED-TYPE slots filled.
156 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
158 ** MAKE-CONDITION accepts classes as type-designators.
159 ** COMPILE may never return NIL.
160 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
161 range before calling Unix time functions
163 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
164 * numerous signal handling fixes to increase stability
165 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
166 * minor incompatible change: we now correctly canonize default
167 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
168 as per the MOP, rather than the historical (INITARG INITFUNCTION
169 INITFORM). (reported by Bruno Haible)
170 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
171 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
172 for more information.
173 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
174 pathname is a directory pathname.
175 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
176 * dynamic space size on PPC has been increased to 768Mb. (thanks to
178 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
179 :METACLASS argument in addition to a class name. (reported by
180 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
181 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
182 (reported by Bruno Haible; patch from Gerd Moellmann)
183 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
185 * bug fix: more cleanups to the floating point exception handling on
186 x86-64 (thanks to James Knight)
187 * bug fix: the compiler does not try to constant fold calls of
188 COERCE and BIT-* functions when they return freshly constructed
189 objects. (reported by Nikodemus Siivola and Paul Dietz)
190 * optimization: improved the MIPS versions of generic (in the
191 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
192 * optimization: direct conversion of (unsigned-byte 32) to floats on
193 the PowerPC platform.
194 * optimization: structure instances with raw slots now use less
195 memory, and probably show better memory locality. (thanks to
197 * optimization: DYNAMIC-EXTENT declarations for lists and closures
198 are treated as requests for stack allocation on the x86-64,
199 Alpha-32, PPC and SPARC platforms.
200 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
201 the wrong file descriptor; implementation of SOCKET-OPEN-P.
202 (thanks to Tony Martinez)
204 ** gcing a dead thread can no longer lead to lockups
205 ** threads block signals until they are set up properly
206 ** errno is no longer shared by threads
207 ** interrupt-thread restores the eflags register on x86
208 ** fixed some lockups due to gc/thread interaction
209 * fixed some bugs revealed by Paul Dietz' test suite:
210 ** invalid dotted lists no longer raise a read error when
212 ** signal an error if a symbol that names a declaration is used
213 as the name of a type, or vice versa
214 ** allow using the (declare (typespec var*)) abbreviation for
215 (declare (type typespec var*)) with all type specifiers
216 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
217 ** always return NIL from PPRINT-POP when OBJECT is NIL
218 ** don't signal errors when pretty-printing malformed LABELS,
219 FLET or MACROLET forms
220 ** declarations in a DOLIST body are also in scope for the
222 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
224 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
227 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
228 * fixed cross-compiler leakages that prevented building a 32-bit
229 target with a 64-bit host compiler.
230 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
231 opened with :IF-EXISTS :OVERWRITE.
232 * fixed bug 281: error for an invalid qualifier in a short-form method
233 combination method is not signalled until the faulty method is called.
234 * bug fix: iteration variable type inferrer failed to deal with open
235 intervals. (reported by Alan Shields)
236 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
237 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
238 or more runtime options were provided to the sbcl binary.
239 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
240 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
241 * optimization: there's now a fast-path for fixnum arguments in the
242 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
243 * optimization: the code generated on x86-64 is more compact thanks
244 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
245 * bug fix: floating-point exception handling now partly works on
246 x86-64. (Thanks to James Knight)
247 * improvement to the MIPS backend from Thiemo Seufer: C-style
248 64-bit long long arguments and return values to alien functions
250 * fixed some bugs revealed by Paul Dietz' test suite:
251 ** the type-error signalled from WARN has a filled-in DATUM slot.
252 ** the type-error required when a stream is not associated with
253 a file has the stream as its datum.
254 ** type-errors on single-floats on x86-64 no longer have
255 :INVALID-OBJECT as the datum
256 ** the type-errors signalled for invalid function names now have
257 a correct expected type
258 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
259 for structures defined with a :TYPE. Documentation strings for
260 typed structures are no longer immediately discarded
261 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
262 broadcast streams and synonym streams. FILE-LENGTH now also works
263 on broadcast streams.
265 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
266 * incompatible change: the --noprogrammer option, deprecated since
267 version 0.7.5, has been removed. Please use the equivalent
268 --disable-debugger option instead.
269 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
271 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
272 inoperational for a while now) has been completely removed.
273 * null lexical environments are now printed as #<NULL-LEXENV>,
274 significantly reducing the amount of clutter in typical backtraces.
275 * documentation on weak pointers, finalization, and after GC hooks
276 has been added to the manual.
277 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
278 as other specialized array types, is much faster. SUBSEQ and
279 COPY-SEQ on such arrays have also been sped up.
280 * optimization: EQL is now more efficient when at least other argument
281 is known to be of type (OR FIXNUM (NOT NUMBER)).
282 * fixed bug: the runtime is now less vulnerable to changes in the
283 size of the SBCL object on OS X, and virtual memory is reserved for
284 all spaces that need to be at a fixed address.
285 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
286 * fixed bug: finalizers and after GC hooks that cause consing are now
288 * fixed bug: compiler error messages and summaries are now printed to
289 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
290 * fixed inference of the upper bound of an iteration variable.
291 (reported by Rajat Datta).
292 * fixed bug 211e: calling local functions with duplicated constant
293 keyword argument no longer causes a bogus style warning about an
295 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
296 ftype declaration to be disregarded. (reported by Dave Roberts)
297 * fixed bug 373: caused by erronous compilation of references to alien
298 variables in the runtime on ppc/darwin.
299 * fixed bug 376: CONJUGATE type deriver.
300 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
301 (reported by Baughn on #lisp)
302 * a cross-compiler bug on non-x86 platforms has been identified and
303 fixed. (thanks to Bruno Haible)
304 * improvements to the MIPS runtime code for increased stability.
305 (thanks to Thiemo Seufer)
306 * increased the maximimum compact environment size to allow
307 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
308 * partial workaround for bug 135: don't name the function we're
309 calling for hairy cases of EVAL, so as not to accumulate one environment
310 entry per call to EVAL. (reported by Kevin Reid)
311 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
312 * optimization: type testing for non-vector arrays should be faster.
313 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
315 * bug fix: setting 31st element of a bit vector to zero did not work
317 * bug fix: redefining a class definition which failed due to a
318 previous accessor / function clash now works (but see BUGS entry
319 #380 for more problems in this area). (thanks to Zach Beane)
320 * the long form of DEFINE-METHOD-COMBINATION disables method group
321 checking when given a single method group with pattern *. (thanks
323 * on x86 compiler supports stack allocation of results of simple
324 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
325 * support for building and running on Mac OS X 10.4 "Tiger" has been added
326 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
327 * fixed some bugs related to Unicode integration:
328 ** the restarts for recovering from input and output encoding
329 errors only appear when there is in fact such an error to
331 * fixed some bugs revealed by Paul Dietz' test suite:
332 ** MISC.549 and similar: late transformation of unsafe type
333 assertions into derived types caused unexpected code
335 ** SCALE-FLOAT type deriver is less wrong.
336 ** type derivers for EXP, LOG and similar functions compute result
337 types for complex arguments better.
338 ** (MISC.563) CONJUGATE type deriver works for very restricted
340 ** out-of-line type testers for character strings are available.
341 ** EQUAL compiler transform understands specialness of objects
343 ** accessing double-floats stored on the stack now works on x86-64.
344 ** debugger internals could sometimes create invalid lispobjs,
345 resulting in GC crashes.
346 ** MISC.548: type check weakening can convert required type into
348 ** initialization forms for bindings are not in scope of free special
351 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
352 * incompatible change: thread support for non-NPTL systems has
353 been removed - locking is buggy and unreliable. A threaded
354 SBCL build will now warn at startup and refuse to create
355 new threads, unless futex support is detected in the kernel
356 * incompatible change: the top level REPL now has only an ABORT
357 restart associated with it, not TOPLEVEL and ABORT as it used to.
358 TOP and TOPLEVEL are now available as debugger commands for
359 returning to the top level.
360 * incompatible change: forms evaluated in the REPL now use the
361 global optimization policy.
362 * incompatible change: user- and system-initialization files are
363 no longer processed with LOAD, but by READ and EVAL; hence the
364 global optimization policy, startup package, readtable, etc,
366 * The .fasl file format number has been incremented because of
367 various incompatible changes.
368 * internal entry point details and argument counts no longer appear
369 in backtraces unless explicitly requested by setting
370 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
371 * built-in and standard functions no longer have names like "top
372 level local call to FOO".
373 * fixed bug 32: functions defined in non-null lexical environments
374 now have more legible printed representation
375 * fixed bug 33: functions defined in non-null lexical environemnts
376 are now more amenable to inspection by INSPECT.
377 * workaround for bug 354: XEPs no longer appear in backtraces unless
378 explicitly requested.
379 * fixed bug: receiving the signal which results from attempting to
380 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
381 SIGBUS on other *BSDs) on architectures where the C stack is also
382 the Lisp stack (x86 and x86-64) from foreign code no longer leads
383 to debugger confusion or wild instability. (reported by Cheuksan
385 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
386 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
387 (reported by Lutz Euler)
388 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
389 compiled code. (reported by Paul Dietz)
390 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
391 specialized parameter to a method to something that is not TYPEP
392 the specializer is now possible.
393 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
394 face of package deletion.
395 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
396 pathname. (reported by David Tolpin on c.l.l)
397 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
398 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
399 STRING-STREAM and FILE-STREAM, respectively.
400 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
401 than previously; in addition, its test suite is now run on build.
402 (thanks to Luke Gorrie)
403 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
404 * optimization: added a immediate representation for single-floats
406 * fixed some bugs revealed by Paul Dietz' test suite:
407 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
408 ** attempting to create a package with a colliding nickname causes
409 correctable errors to be signalled.
410 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
411 ** several x86-64 backend bugs related to sign-extension of immediate
414 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
415 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
416 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
418 * fixed loading of multiply forward-referenced layouts.
419 (thanks to Cheuksan Wang)
420 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
421 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
422 (see COMPILE-FILE documentation for details of :PRINT option)
423 * fixed bugs 19 and 317: fixed-format floating point printing is
424 more accurate. This also fixes a bug reported by Adam Warner
425 related to the ~@F format directive.
426 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
428 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
429 dispatch table if the from-char is a dispatch macro character.
430 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
431 by the wordsize no longer ignore the last word. (reported by Lutz
433 * fixed bug in type checking of dynamic-extent variables. (reported
435 * optimization: sequence traversal functions use their freedom to
436 coerce function designators to functions.
437 * optimization: code with many calls to CLOS methods specialized on
438 CLOS classes has had redundant type checks removed.
439 * optimization: added declarations to speed up operations that access
440 the internal character database (for example STRING-UPCASE)
441 * optimization: comparison operations between floats and sufficiently small
442 fixnums no longer create extra rationals
443 * fixed some bugs related to Unicode integration:
444 ** portions of multibyte characters at the end of buffers for
445 character-based file input are correctly transferred to the
446 start of the buffer at the next read.
447 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
448 passing it through to OPEN.
449 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
450 argument given to internal calls to OPEN.
451 ** the built-in comment readers (introduced by character sequences
452 ";" and "#|") are more forgiving to encoding errors; they will
453 STYLE-WARN and then attempt to resync the stream at a character
454 boundary. (thanks to Teemu Kalvas)
455 * fixed some bugs in the x86-64 port:
456 ** Negative short int return values from c-calls are sign-extended
458 ** The stack is aligned to 16-bytes for c-calls, as required by
459 the ABI. (thanks to Cheuksan Wang)
460 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
461 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
462 for enums. (thanks to Vincent Arkesteijn)
463 ** Multiple small optimizations and bugfixes for floating point
465 * fixed some bugs revealed by Paul Dietz' test suite:
466 ** Space, Tab, Linefeed, Return and Page have the invalid
467 secondary constituent character trait.
468 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
470 ** WITH-INPUT-FROM-STRING should only update the index place on
472 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
474 ** Bit-array operations (BIT-AND and similar) worked incorrectly
475 with one-dimensional arrays with fill pointers.
476 ** TYPE-OF failed on a complex with an integer realpart and a
478 ** compiler failure during type inference for the code of form
479 (IF (EQL X (THE ...)) ...) (MISC.535).
481 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
482 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
483 platform. The port supports 61-bit fixnums, large memory spaces
484 and reloading shared object files.
485 * enhancement: saving cores with foreign code loaded is now
486 supported on ppc/Darwin in addition to the previously supported
488 * enhancement: the statistical profiler now walks deeper into the
489 call stack for more meaningful call-graphs and accrued time
490 reports (x86/x86-64 only). It also now reports time spent in
492 * enhancement: it is now possible to trace most individual methods
493 of a generic function in addition to tracing the generic function
495 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
497 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
498 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
499 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
500 produces an error. (thanks to Vincent Arkesteijn)
501 * bug fix: NAMESTRING on pathnames with :WILD components in their
502 directories works correctly. (thanks to Artem V. Andreev)
503 * fixed bug 125: compiler preserves identity of closures. (reported
505 * bug fixed: functions with &REST arguments sometimes failed with
506 "Undefined function" when compiled with (DEBUG 3). (reported by
508 * bug fix: overflow during compiling of setting element of a bit
509 vector with constant index and value. (reported by Timmy Douglas)
510 * build fix: fixed the dependence on *LOAD-PATHNAME* and
511 *COMPILE-FILE-PATHNAME* being absolute pathnames.
512 * on x86 compiler partially supports stack allocation of dynamic-extent
514 * GO and RETURN-FROM do not check the extent of their exit points
515 when compiled with SAFETY 0.
516 * fixed some bugs related to Unicode integration:
517 ** encoding and decoding errors are now much more robustly
518 handled; it should now be possible to recover even from invalid
519 input or output to the terminal. (thanks to Teemu Kalvas)
520 ** provided a first cut at implementing STRING-TO-OCTETS and
521 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
522 ** altered the SB-MD5 contributed module slightly, changing the
523 interface just enough for it to be supportable for builds where
524 lisp characters are not eight bits.
525 * fixed some bugs revealed by Paul Dietz' test suite:
526 ** the FORMATTER-generated functions for ~V[ conditionals require
527 the correct number of arguments.
528 ** READ-FROM-STRING returns the mandated second value when applied
529 to displaced strings.
530 ** the #\Rubout and #\Backspace characters are treated as invalid
531 constituent characters by the tokenizer.
533 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
534 * new feature: reloading changed shared object files with
535 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
536 * new feature: references to foreign variables and functions
537 can now be compiled and loaded before the corresponding shared
538 object file is loaded, as long as the foreign definitions are
539 available at runtime.
540 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
541 supported. (thanks to Dan Debertin)
542 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
543 just keywords. (thanks to Vincent Arkesteijn)
544 * fixed bug #331: structure-class instances corresponding to
545 DEFSTRUCT forms are now created eagerly.
546 * fixed bug #345: backtraces from calls to undefined functions work
547 on x86 as well. Related bug #61 is now also partially fixed on x86
548 (backtraces from throws to unknown catch tags.)
549 * bug fix: lambda-list parsing is now stricter vrt. order and number
550 of lambda-list keywords.
551 * bug fix: as specified by AMOP, an error is signalled if a
552 class-option appears multiple times in a DEFCLASS form. (reported
554 * bug fix: RANDOM can be compiled when the compiler derives the type
555 of its numeric argument as a disjoint set of small integers.
556 (reported by Paul Dietz)
557 * bug fix: starting a core saved with shared objects loaded when
558 those objects are not available no longer causes threaded SBCL to
559 hang. (reported by Sean Ross)
560 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
562 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
563 argument, which must always be NIL. (reported by Kalle Niemitalo)
564 * bug fix: printing 1.0d+23 no longer results in an error.
565 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
566 * bug fix: structure slot setters preserve evaluation order. (thanks
568 * fixed some bugs related to Unicode integration:
569 ** RUN-PROGRAM can allow its child to take input from a Lisp
570 stream. (reported by Stefan Scholl)
571 ** ASDF-INSTALL successfully downloads .tar.gz files in all
572 locales. (reported by Ken Causey)
573 * fixed some bugs revealed by Paul Dietz' test suite:
574 ** INCF, DECF and REMF evaluate their place form as specified in
576 ** FORMATTER expands ~{ ~} iteration directives with V or #
577 parameters correctly.
578 ** FORMATTER deals with the ~@[ ~] conditional directive where the
579 consequent uses no arguments correctly.
580 ** the system has a partial understanding of the (COMPLEX RATIO)
582 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
585 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
586 * new feature: a build-time option (controlled by the :SB-UNICODE
587 keyword feature, enabled by default) for building the system with
588 support for the entire 21-bit character space defined by the
590 * new feature: the system now has rudimentary external-format
591 support; the primary user-visible change at this time is that
592 characters with the high bit set (such as lower-case-e-acute) will
593 print correctly to a terminal in a UTF-8 environment.
594 * minor incompatible change: BASE-CHAR no longer names a class;
595 however, CHARACTER continues to do so, as required by ANSI.
596 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
597 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
598 has been moved to the SB-EXT package (temporarily re-exported from
600 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
601 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
603 * fixed bug #351: better error-handlind and reporting for malformed
605 * fixed bug #350: bignum-printing is now more memory-efficient,
606 allowing printing of very large bignums, eg. (expt 2 10000000).
607 (reported by Bruno Haible)
608 * fixed bug #302: better primitive-type selection for intersection
610 * fixed bug #308: non-graphic characters now all have names, as
611 required. (reported by Bruno Haible)
612 * bug fix: redefining a class with different superclasses now correctly
613 removes it from the direct-subclasses of its previous superclasses.
614 (reported by David Morse)
615 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
616 new value now works. (reported by Bruno Haible)
617 * bug fix: correct canonicalization of multiple non-standard slot
618 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
619 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
620 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
621 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
622 now exists, an signals an error.
623 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
624 during macroexpansion for non-iteration variables as well. (reported
625 by Bruno Haible for CMUCL)
626 * bug fix: Cyclic structures and unprintable objects in compiler
627 messages no longer cause errors. (reported by Bruno Haible)
628 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
629 and READ-FROM-STRING all now return a primary value of NIL if
630 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
631 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
632 (reported by Bruno Haible for CMUCL)
633 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
634 do not designate a condition. (reported by Bruno Haible for
636 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
637 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
638 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
639 permitting resolution in favour of any of the conflicting symbols.
640 (reported by Bruno Haible for CMUCL)
641 * bug fix: EQUAL compiler optimizations is less aggressive on
642 strings which can potentially compare true despite having distinct
643 specialized array element types.
644 * bug fix: unit enumerations can be defined without dividing by
645 zero. (thanks to Vincent Arkesteijn)
646 * FORMAT compile-time argument count checking has been enhanced.
647 (report from Bruno Haible for CMUCL)
648 * a partial workaround for the bug 262: the compiler does not try to
649 inline-expand a local function doing RETURN-FROM from a deleted
650 BLOCK. (thanks to Peter Denno for the bug report and to David
651 Wragg for the simple test case)
652 * fixed some bugs revealed by Paul Dietz' test suite:
653 ** RENAME-PACKAGE allows all package designators as new package
655 ** constraint propagation and lambda variable substitution are
656 more cautious in dealing with partially deleted code.
657 ** compiler sometimes lost reoptimization passes.
658 ** CERROR, given a condition as condition designator, passes its
659 remaining arguments to the continue format control without
661 ** Case-altering FORMAT directives work correctly on non-ASCII
663 ** The REAL type specifier handles bounds outside the floating
664 point ranges without signalling FLOATING-POINT-OVERFLOW.
665 ** Functions with IR1-transformations can create intercomponent
666 references to global functions.
667 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
669 ** FORMAT ~R treats a nil value for its first parameter correctly.
671 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
672 * enhancement: saving cores with foreign code loaded is now
673 supported on x86/NetBSD and sparc/Linux in addition to the previously
675 * bug fix: on some platforms repeated installations caused multiple
676 copies of HTML documentation to be installed -- should not happen
677 any more. (reported by Stefan Scholl)
678 * bug fix: parsing self-recursive alien record types multiple times
679 no longer causes infinite recursion. (reported by Thomas F. Burdick,
680 original patch by Helmut Eller for CMUCL)
681 * bug fix: stack-exhaustion detection works now on NetBSD as well.
682 (thanks to Richard Kreuter)
683 * bug fix: defining classes whose accessors are methods on existing
684 generic functions in other (locked) packages no longer signals
685 bogus package lock violations. (reported by François-René Rideau)
686 * bug fix: special variables as DEFMETHOD parameters no longer have
687 associated bogus type declarations. (reported by David Wragg and
689 * bug fix: read-write consistency on streams of element-type
690 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
691 * bug fix: redefiniton of the only method of a generic function with
692 no DEFGENERIC no longer emits a full WARNING. In addition,
693 redefinition of generic functions with no DEFGENERIC to an
694 incompatible lambda list now signals an error. (thanks to Zach
696 * bug fix: DEFGENERIC now works even when there's a function of the
697 same name in an enclosing lexical environment. (thanks to Zach
699 * fixed compiler failure, caused by instrumenting code during
700 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
701 * optimization: added loop analysis and improved register allocation
702 * fixed some bugs revealed by Paul Dietz' test suite:
703 ** POSITION on displaced vectors with non-zero displacement
704 returns the right answer.
705 ** (SIMPLE-STRING) is a valid type specifier for sequence
707 ** *PRINT-LEVEL* handling for slotless structures is pedantically
709 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
710 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
712 ** The justification version of the FORMAT ~< directive treats
713 non-zero minpad parameter correctly.
715 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
716 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
717 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
718 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
719 the supported interface.
720 * new feature: Single-stepping of code compiled with DEBUG 2 or
721 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
722 * new feature: saving cores with foreign code loaded is now
723 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
724 Timothy Moore's work for CMUCL)
725 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
726 parameters to * instead of NIL if no initform is supplied.
727 (reported by Johan Bockgaard)
728 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
729 correctly when FUNCALL appears as the car of the form. Note:
730 despite this FUNCALL forms are not currently subject to
731 compiler-macro expansion. (port of Raymond Toy's fix for the
732 same from CMUCL, reported by Johan Bockgaard)
733 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
734 (thanks for Teemu Kalvas)
735 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
736 case instead of terminating the iteration (thanks for Julian
737 Squires, Sean Champ and Raymond Toy)
738 * bug fix: incorrect expansion of defgeneric that caused a style
739 warning. (thanks for Zach Beane)
740 * bug fix: slot accessor effective method computation works properly
741 for classes with multiple non-standard applicable methods on
742 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
743 * on x86 compiler supports stack allocation of results of LIST and
744 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
745 CMUCL implementation by Gerd Moellmann)
746 * fixed some bugs revealed by Paul Dietz' test suite:
747 ** FORMAT strings with both the ~<~:;~> form of the justification
748 directive and pretty-printing directives cause an error.
750 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
751 * incompatible change: the internal functions
752 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
753 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
754 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
755 instead of the old functions.
756 * new feature: on platforms where "dladdr" is available foreign
757 function names now appear in backtraces. (based on Helmut Eller's
759 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
760 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
762 * documentation: ASDF manual is now distributed with SBCL.
763 * bug fix: SBCL can now load its contributed modules with REQUIRE
764 even if the system-provided entries have been removed from
765 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
767 * bug fix: backtraces involving undefined functions or assembly
768 routines are more informative. (thanks to Brian Downing)
769 * bug fix: mutually referent alien structures now work correctly.
770 (reported by Rick Taube)
771 * bug fix: structures defined by WITH-ALIEN can be referred to
772 within other definitions in the same WITH-ALIEN.
773 * bug fix: division operators (MOD, TRUNCATE and the like) with
774 constant zero divisors and integer dividends no longer generate
776 * bug fix: provide default methods for INPUT-STREAM-P and
777 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
778 * bug fix: improve the detection and resolution of MOP metacycles
779 (where effective method computation depends on the generic function
780 having its effective method computed). (reported by Bruno Haible)
781 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
782 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
783 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
784 * bug fix: pathnames with relative directory components can be
785 represented relative to default pathnames.
786 * optimization: in taking the GCD of bignums, reduce the two bignums
787 to approximately the same size (using Euclid's algorithm) before
788 applying the more sophisticated binary GCD. (thanks to Juho
790 * optimization: COUNT on bitvectors now operates word-at-a-time.
791 * optimization: ASH with a positive, but not necessarily constant,
792 (leftwards) shift, when in a modular context, is compiled to a
794 * fixed some bugs revealed by Paul Dietz' test suite:
795 ** FORMAT variable parameters ("~V<char>") are defaulted properly
796 if the corresponding argument is NIL.
797 ** FORMAT directives accepting a minpad parameter treat negative
799 ** literal commas as character parameters to FORMAT directives are
801 ** literal spaces directly after ~<Newline> directives within a
802 format-logical-block (~:< ~@:>) do not induce :FILL-style
803 conditional newlines.
804 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
805 as specified: it no longer includes conditional newlines).
806 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
808 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
809 :PER-LINE-PREFIX argument does not evaluate to a string.
811 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
812 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
813 the manual for details; for now, package locks can be disabled by
814 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
815 if no major problems are found then it is likely that they will be
816 compiled in unconditionally.
817 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
818 unsupported operators on all platforms. To load a shared library
819 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
820 object file, link it into a shared library outside of SBCL and
821 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
823 * bug workaround: ROOM T and the GC were not working together
824 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
825 reporting facility aren't preserved. That reporting has been
826 disabled, so now until and unless someone figures out how to make it
827 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
828 * minor incompatible change: as threatened around sbcl-0.8.0, the
829 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
830 The symbols are also exported from SB-PCL for backwards
831 compatibility, but more so than before SB-PCL should be treated as
832 an implementation-internal package.
833 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
835 * the SB-SPROF contrib now works on (most) non-x86 architectures.
836 It is known as of this release not to work on the Alpha, however.
837 * fixed bug #167: errors signalled due to illegal syntax in method
838 bodies are now more legible.
839 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
840 designators and can hence be used with TYPEP.
841 * fixed bug #333: CHECK-TYPE now ensures that the type error
842 signalled, if any, has the right object to be accessed by
843 TYPE-ERROR-DATUM. (reported by Tony Martinez)
844 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
845 5.1.2.3 for multiple-value place subforms. (reported by Kalle
847 * fixed bug #334: programmatic addition of slots using specialized
848 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
849 and :ALLOCATION :CLASS slots.
850 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
851 float exponents. (rereported by Peter Seibel)
852 * fixed a bug: #\Space (and other whitespace characters) are no
853 longer considered to be macro characters in standard syntax by
855 * fixed bug: initialization of condition class metaobjects no longer
856 causes an instance of the condition to be created. (reported by
858 * fixed bug: it is now possible to have slots such that
859 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
860 description is neither :INSTANCE nor :CLASS.
861 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
862 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
863 SLOT-VALUE-USING-CLASS), and no longer causes errors with
864 non-standard SLOT-DEFINITION-ALLOCATION values.
865 * fixed bugs: various race conditions handling exiting threads.
866 CL-PPCRE's thread torture test now passes.
867 * fixed arguably-a-bug: GC time proportional to number of threads in
868 system even when most of them are idle
869 * optimization: improved performance of BIT and SBIT on bit-vectors.
870 * .fasl file incompatibility: The fasl file version number has
871 been incremented because of changes associated with package locks.
873 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
874 * minor incompatible change: the system no longer provides
875 optimization hints (conditions which are TYPEP
876 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
877 mode; these hints are still emitted when higher SPEED optimization
879 * new contrib module: a sampling profiler (profiling by statistical
880 sampling, rather than by instrumenting functions) is available as
881 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
882 * the behaviour of the standard function ED is now customizeable by
883 third parties through a hook variable: see ED's documentation
884 string for information on the protocol.
885 * the compiler no longer emits efficiency notes for (FUNCALL X)
886 when the type of X is uncertain under default optimization
888 * fixed bug 276: mutating a binding of a specialized parameter to a
889 method to something that is not TYPEP the specializer is now
891 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
892 exhibit the required behaviour on the x86 platform. (thanks to
893 Peter van Eynde, Eric Marsden and Bruno Haible)
894 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
895 even for difficult arguments. (reported by Peter Graves)
896 * fixed bug 141a: the backquote printer now descends quoted
898 * fixed another bug in backquote printing: no more destructive
899 modification of the form's list structure. (reported by Brian
901 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
902 move between its address being taken and the call to
903 interrupt_thread, fixing a crashing race condition.
904 * the SB-POSIX contrib implementation has been adjusted so that it
905 no longer exhibits ridiculously poor performance when constructing
906 instances corresponding to C structs.
908 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
909 * minor incompatible change: the sb-grovel contrib now treats C
910 structures as alien (in the sense of SB-ALIEN) objects rather than
911 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
912 has implications for memory management of client code
913 (sb-grovel-returned objects must now be manually managed) and for
914 type safety (alien objects now have full types).
915 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
916 used to control emission of compiler diagnostics, rather than the
917 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
918 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
919 quality should be considered deprecated.
920 * install.sh now installs the user manual as well
921 * (not quite a new documentable feature, but worth considering in
922 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
923 beginnings of a semantically meaningful condition hierarchy is
924 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
926 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
927 designator as the defaults argument.
928 * fixed bug: Displaced arrays whose displaced-to array has become
929 too small now cause ARRAY-DIMENSION to signal an error, providing
930 for safer bounds-checking. (reported by Bruno Haible)
931 * fixed bug: DEFCLASS slot definitions with identical :READER and
932 :WRITER names now signal a reasonable error. (reported by Thomas
934 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
936 * fixed bug: Invalid slot specification errors now print correctly.
937 (thanks to Zach Beane)
938 * fixed bug 320: Shared to local slot value transfers in class
939 redefinitions now happen corrently from superclasses as
940 well. (reported by Bruno Haible)
941 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
943 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
944 improper lists correctly. (reported by Bruno Haible)
945 * fixed bug 313: source-transform for <fun-name> was erroneously
946 applied to a call of a value of a variable with name <fun-name>.
947 (reported by Antonio Menezes Leitao)
948 * fixed bug 307: The obsolete instance protocol ensures that
949 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
950 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
951 remove dead unknown-values globs from the middle of the stack.
952 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
953 easier to resume long computations after using *BREAK-ON-SIGNALS*
954 to diagnose and fix failures. (thanks to Nikodemus Siivola)
955 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
956 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
957 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
958 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
959 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
961 * fixed a (fairly theoretical) bug in string printing: if
962 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
963 does not have array-element-type equal to the most general string
965 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
966 function name. (thanks to Nikodemus Siivola)
967 * fixed bug: compile-time format string checker failed on
968 non-closed ~{. (reported by Thomas F Burdick)
969 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
970 don't warn on legal loop code involving more than one aggregate
971 boolean. (thanks to Nikodemus Siivola)
972 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
973 in keyword position to MAKE-PACKAGE was not being reported as
975 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
976 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
978 * fixed bugs 280 and 312: the checking for multiple definitions in a
979 file is less likely to become confused by uses of inline
981 * fixed bug: the #S reader macro performs the keyword coercion
982 specified for slot names. (reported by Kalle Niemitalo)
983 * fixed bug: lambda lists may contain symbols whose names start with
984 & but are not lambda-list-keywords; their occurrence triggers a
986 * fixed bug 321: define-method-combination argument lambda lists do
987 not cause invalid code to be generated when &optional or &aux
988 variables have default values. (reported by Bruno Haible)
989 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
990 have CLOS classes; this appears also to have fixed reported
991 instability in stack exhaustion detection.
992 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
993 MUFFLE-WARNING when no associated restart is present is now
995 * optimization: rearranged the expansion of various defining macros
996 so that each expands into only one top-level form in a
997 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
999 * optimization: used a previously little-used slot in symbols to
1000 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1002 * fixed some bugs revealed by Paul Dietz' test suite:
1003 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1005 ** Fixed an optimization bug related to inheritance of initforms
1006 from local to shared slots.
1007 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1008 ** CAST optimizer forgot to flush argument derived type.
1009 ** print/read consistency on floats is now orders of magnitude
1010 more likely. (thanks also to Bruno Haible for a similar report
1012 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1013 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1014 for objects of type REAL. Make it so.
1015 ** SXHASH is specified (infelicitously) to respect similarity,
1016 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1017 it so. (thanks to Markus Ziegler)
1018 ** on the Alpha, the compiler succeeds in compiling functions
1019 returning a known number of arguments greater than 63.
1020 ** fixed handling of invalid NIL arguments in keyword position
1022 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1024 ** provided a readably-printable representation for RANDOM-STATE
1026 ** ensured that pathnames, where they have a namestring, always
1027 print using #P"..." syntax.
1029 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1030 * Support for the forthcoming 2.0 version of the NetBSD kernel
1031 running on x86 hardware has been added. (thanks to Perry
1032 E. Metzger most immediately, and others for their past work)
1033 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1034 current release version; SBCL's OpenBSD support had been broken
1035 since about the time of OpenBSD's switch to ELF binary format.)
1036 * [placeholder for DX summary]
1037 ** user code with &REST lists declared dynamic-extent, under high
1038 speed or space and low safety and debug optimization policy.
1039 * The manual has been converted to Texinfo format and the debugger
1040 chapter from the cmucl manual has been added.
1041 * A facility has been added to extract documentation strings from
1042 sbcl and store them as Texinfo-formatted snippets for inclusion in
1043 the manual (via Texinfo's @include directive)
1044 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1045 the test case to Dave Roberts)
1046 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1047 pointers. (reported by Sean Ross)
1048 * bug fix: PROFILE output is printed nicely even for large numerical
1049 values. (thanks to Zach Beane)
1050 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1051 greater than 32 handle EOF correctly.
1052 * bug fix: on X86 an immediate argument of the IMUL instruction is
1053 correctly printed in disassembly. (thanks to Lutz Euler)
1054 * bug fix: class slots in redefined classes preserve their old
1055 values. (thanks to Bruno Haible and Nikodemus Siivola)
1056 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1057 succeeds. (reported by Marco Baringer)
1058 * fixed some bugs revealed by Paul Dietz' test suite:
1059 ** READ-SEQUENCE now works on ECHO-STREAMs.
1060 ** RATIONALIZE works more according to its specification. (thanks
1063 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1064 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1065 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1066 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1067 should matter to you only if you rebind the printer control
1068 variables and then find you want different bindings in the
1069 debugger than in the ordinary execution of your program.)
1070 * The runtime build system has been tweaked to support building
1071 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1072 assembler and linker. (thanks to Nikodemus Siivola)
1073 * Unbound, undefined, undeclared variables now trigger full
1074 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1075 more often programmer error than deliberate exploitation of undefined
1077 * optimization: the hash algorithm for strings has changed to one
1078 that is less vulnerable to spurious collisions. (thanks to Juho
1080 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1081 needless bounds checking. (thanks to Juho Snellman)
1082 * optimization: implemented multiplication as a modular
1083 (UNSIGNED-BYTE 32) operation on the PPC backend.
1084 * fixed some bugs revealed by Paul Dietz' test suite:
1085 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1087 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1088 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1089 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1091 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1093 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1095 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1096 their output stream on EOF from read.
1097 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1098 have been read to end-of-file.
1099 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1101 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1102 description of determination of which consecutive characters
1104 ** Printing the "Space" character with escaping on now yields "#\\ ",
1105 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1106 ** Reading floating-point numbers with *READ-BASE* set to a number
1107 less than 10 works correctly.
1108 ** Reading floating-point numbers with *READ-BASE* set to a number
1109 more than 10 works correctly.
1110 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1111 the readtable currently in effect.
1113 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1114 * minor incompatible change: parsing of namestrings on a physical
1115 (Unix) host has changed; numbers after the final #\. in a
1116 namestring are no longer interpreted as a version field. This is
1117 intented to be largely invisible to the user, except that the
1118 meaning of the namestring "*.*.*" has changed: it now refers to a
1119 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1120 should usually be replaced by
1121 (make-pathname :name :wild :type :wild :version :wild)
1122 with the added benefit that this is more likely to be portable.
1123 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1124 option to OPEN now signals an error if the file being opened
1125 exists; this may have an impact on existing code.
1126 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1127 signals received, on the PowerPC platforms (both Linux and
1128 Darwin). (thanks to Pierre Mai for pointing out the location of
1130 * several fixes on OS X: The system now builds and runs cleanly on
1131 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1132 (thanks to Brian Mastenbrook)
1133 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1134 second-resolution: integer multiples of 1/3600 between -24 and 24.
1135 (thanks to Vincent Arkesteijn)
1136 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1137 type when called with 1 argument; PEEK-CHAR checked type of
1138 PEEK-TYPE only after having read first character from a
1139 stream. (reported by Peter Graves)
1140 * bug fix: the garbage collector now has much better locality
1141 behaviour, and in particular no longer treats all memory as being
1142 exhausted when many small objects point to each other in a deeply
1144 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1145 recognized as being TYPEP their class.
1146 * bug fix: the PUSHNEW documentation string has been corrected.
1147 (thanks to Vincent Arkesteijn)
1148 * bug fix: defaulting of the value for the last of an atypically
1149 large number of multiple values being bound was not being
1150 performed correctly on the Alpha or PPC platforms
1151 * optimization: implemented multiplication as a modular
1152 (UNSIGNED-BYTE 32) operation on the x86 backend.
1153 * optimization: SEARCH on simple-base-strings can now be open-coded.
1154 (see also contrib/compiler-extras.lisp for inspiration for
1155 teaching the compiler about the Boyer-Moore algorithm).
1156 * value, returned by MAX (and MIN) called with several EQUALP, but
1157 not EQL, arguments now does not depend on compiler settings.
1158 * fixed some bugs revealed by Paul Dietz' test suite:
1159 ** in stack analysis liveness information is propagated from
1160 non-local entry points.
1161 ** pathwise CAST removing failed when the CAST node did not start
1163 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1164 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1166 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1167 host is already defined.
1168 ** RENAME-FILE works on streams instead of signalling an internal
1170 ** PEEK-CHAR uses the current readtable when determining whether
1171 or not a character is whitespace.
1172 ** MERGE-PATHNAMES handles the case when the pathname does not
1173 specify a name while the default-pathname specifies a version.
1174 ** Pathnames now stand a better chance of respecting print/read
1176 ** Attempting to use standardized file system operators with a
1177 pathname with invalid :DIRECTORY components signals a
1179 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1181 ** DIRECTORY on logical pathnames is more correct.
1182 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1183 signal a TYPE-ERROR if their argument is not a stream.
1184 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1185 designator argument does not designate a stream.
1186 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1187 examining the synonym.
1188 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1190 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1191 element-type larger than ([UN]SIGNED-BYTE 32).
1193 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1194 * When built with the :SB-FUTEX feature, threaded builds now take
1195 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1196 for faster/more reliable mutex and condition variable support.
1197 * Incompatible change (but one you probably shouldn't have been using
1198 anyway): the interface and code for arbitrating between multiple
1199 threads in the same user session has been redesigned.
1200 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1201 error when called without an explicit environment argument.
1202 (thanks to Nikodemus Siivola)
1203 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1204 confused about their position. (thanks to Adam Warner and Gerd
1206 * bug fix: Pretty printing backquoted forms with unquotations in the
1207 argument list position of various code constructs such as LAMBDA
1208 now works correctly. (reported by Paul Dietz)
1209 * bug fix: Pretty printing unquotations no longer loses all
1210 stream position information.
1211 * optimization: performance of string output streams is now less
1212 poor for multiple small sequence writes.
1213 * optimization: performance of CSUBTYPEP in the presence of complex
1214 expressions involving CONS and NOT many times has been improved.
1215 (reported by Paul Dietz)
1216 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1218 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1219 streams for the REPL, for the debugger, and so on. (thanks to
1221 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1222 * fixed some bugs revealed by Paul Dietz' test suite:
1223 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1224 is ignored if it is too small, rather than propagating through
1225 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1226 ** extremely complex negations of CONS types were not being
1227 sufficiently canonicalized, leading to inconsistencies in
1229 ** VALUES tranformer lost derived type.
1231 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1232 * fixed a bootstrapping bug: the build process no longer assumes
1233 that the various BOOLE-related constants have the same value in
1234 host and target lisps. (noted by Paul Dietz' test suite on an
1235 SBCL binary built from CLISP)
1236 * The system can now be dynamically linked on the MIPS platform,
1237 which enables dynamic loading of foreign code from Lisp. (thanks
1238 to Ralf Baechle for discussions on the MIPS ABI)
1239 * The system now records debugging information for its own source
1240 files in a filesystem-position-independent manner, relative to
1241 the "SYS" logical host.
1242 * fixed a compiler bug: MV-LET convertion did not check references
1243 to the "max args" entry point. (reported by Brian Downing)
1244 * tweaked disassembly notes to be less confident about proclaiming
1245 some instruction as an LRA. (thanks to Brian Downing)
1246 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1247 now each have their own history, command character, and other
1248 characteristics. (thanks to David Lichteblau)
1249 * fixed some bugs revealed by Paul Dietz' test suite:
1250 ** compiler failure in compiling LOGAND expressions including a
1252 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1253 shift greater than 32.
1254 ** FLUSH-DEST did not mark blocks for type check regeneration.
1255 ** HANDLER-CASE failed to accept declarations in handler clauses
1256 in some circumstances.
1258 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1259 * New code in contrib/sb-introspect (still probably not entirely
1260 stable yet) provides some support for smart Lisp development
1261 environments like SLIME.
1262 * The conditions signalled for errors occurring when loading .fasl
1263 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1264 in a way which should help ASDF recover gracefully.
1265 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1266 stuff has been cleaned up. If you code contrib/ stuff, this might
1267 affect you, and you can look at contrib/README, contrib/STANDARDS,
1268 and/or the 0.8.4.27 diff to check.
1269 * In full calls the compiler now does not generate checks for declared
1270 argument types for all arguments.
1271 * various threading fixes
1272 ** and some experimental patches which didn't make it into
1273 the main tree for this release, but which are shipped in
1274 contrib/experimental-thread.patch as a possible fix for some
1275 failures (deadlock, spinning...) in GC-intensive multithreaded
1277 * fixed PPC build problem (source code incompatibility of different
1278 library versions): added offsetof() hackery which attempts to divine
1279 where glibc maintainers put uc_mcontext today
1280 * fixed bug 282: compiler does not trust type assertions while passing
1281 arguments to a full call.
1282 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1283 &OPTIONAL VALUES type parameter.
1284 * fix bug 214: algorithm for noting rejected templates is now more
1285 similar to that of template seletion. (also reported by rydis on
1287 * fixed bug 141b: printing backquoted information readably and prettily
1288 inserts a space where necessary.
1289 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1290 (CHARACTER 10) are now reported as errors, rather than propagated
1291 as unknown types. (reported by piso on #lisp)
1292 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1293 with values NIL and :ERROR. (thanks to Milan Zamazal)
1294 * fixed bug 191c: CLOS now does proper keyword argument checking as
1295 described in CLHS 7.6.5 and 7.6.5.1.
1296 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1297 longer raise an error; further, using a list as a for-as-arithmetic
1298 counter now raises a meaningful error.
1299 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1300 now understood by sequence creation functions such as MAKE-SEQUENCE
1302 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1303 a STRING-INPUT-STREAM.
1304 * compiler enhancement: SIGNUM is now better able to derive the type
1306 * type declarations inside WITH-SLOTS are checked. (reported by
1308 * fixed some bugs revealed by Paul Dietz' test suite:
1309 ** incorrect optimization of TRUNCATE for a positive first
1310 argument and negative second.
1311 ** compiler failure in let-convertion during flushing dead code.
1312 ** compiler failure while deriving type of TRUNCATE on an
1313 interval, containing 0.
1314 ** ASH of a negative bignum by a negative bignum count now returns
1316 ** intersection of CONS types now canonicalizes properly, fixing
1317 inconsistencies in SUBTYPEP.
1319 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1320 * incompatible change: The --disable-debugger command line
1321 option now clobbers the debugger at a more fundamental
1322 level, by redefining #'INVOKE-DEBUGGER instead of by
1323 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1324 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1325 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1326 enter the debugger REPL and then suffer recursive errors
1327 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1328 while under the new system, BREAK in --disable-debugger mode
1329 terminates the system just as an unhandled error would.
1330 * fixed compiler performance when processing loops with a step >1;
1331 * bug fix: DOCUMENTATION now retrieves generic function
1332 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1333 support has been systematized, and now supports the methods
1334 specified by ANSI, along with a default method and a method for
1335 slot documentation. (reported by Nathan Froyd)
1336 * bug fix: effective methods associated with a generic function are
1337 no longer cached over a change of that generic function's method
1338 combination. (reported by Andreas Fuchs)
1339 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1340 names a non-existent directory. (thanks to Andreas Fuchs)
1341 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1342 the PPC platform, lead to stack corruption; this has been fixed.
1343 (reported by Rainer Joswig)
1344 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1345 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1346 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1347 arithmetic in general, is now much more reliable on the PPC
1349 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1350 platform now returns the right answer.
1351 * optimization: restored some effective method precomputation in
1352 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1353 precomputation is now tunable.
1354 * optimization: compiler-internal data structure use has been
1355 reviewed, and changes have been made that should improve the
1356 performance of the compiler by about 20%.
1357 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1358 simple-base-strings and simple-bit-vectors is improved.
1359 * optimization: the optimization of 32-bit logical and arithmetic
1360 functions introduced in version 0.8.3 on the x86 has been
1361 implemented on the mips, ppc and sparc platforms; an
1362 implementation of the same facility, but for 64-bit arithmetic,
1363 has been added for the alpha.
1364 * microoptimization: the compiler is better able to make use of the
1365 x86 LEA instruction for multiplication by constants.
1366 * bug fix: in some situations compiler did not report usage of
1367 generic arithmetic in (SPEED 3) policy.
1368 * bug 145b fix: compiler used wrong type specifier while converting
1369 MEMBER-types to numeric.
1370 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1372 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1373 index argument now works properly on the Alpha platform.
1374 * bug fix: floating point exception treatment on the Alpha platform
1376 * bug fix: FILE-POSITION works much better on string input and
1377 output streams. (thanks to Nikodemus Siivola)
1378 * bug fix: many threading/garbage collection symptoms sorted.
1379 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1380 might be pseudo-atomic.
1381 * internal change: Stopping for GC is now done with signals not
1382 ptrace. GC is now done in whichever thread wanted it, instead of
1384 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1386 * bug fix: result form in DO is not contained in the implicit
1388 * incompatible change: ICR structure is changed; the value part of
1389 CONTINUATION is now called LVAR; corresponding functions are
1390 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1391 * added type deriver for ISQRT (thanks to Robert E. Brown).
1392 * bug fix: better support for loading from the command line when an
1393 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1395 * fixed some bugs revealed by Paul Dietz' test suite:
1396 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1397 ** ROUND and FROUND now give the right answer when given very
1398 small float arguments.
1399 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1401 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1402 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1403 ** LCM should return a non-negative integer.
1404 ** PARSE-INTEGER returned the index of a terminator instead of the
1405 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1406 ** PARSE-INTEGER returned an incorrect index being applied to a
1408 ** LCM with two arguments of 0 returns 0 rather than signalling
1410 ** unsigned addition of a 32-bit constant with the high bit set no
1411 longer causes an internal compiler error on the x86.
1412 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1413 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1414 with negative last argument.
1415 ** byte specifiers with zero size and position no longer cause
1416 an error during type derivation.
1417 ** bignum multiplication on the Alpha platform now returns the
1419 * porting: The system now builds on SuSE AMD64, although it still
1420 generates a 32-bit binary.
1421 * .fasl file incompatibility: The fasl file version number has
1422 been incremented (because of the changes to internal compiler
1423 data structures referred to above).
1425 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1426 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1427 more accurately, on the Darwin kernel running on PowerPC hardware.
1428 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1429 * Compiler code deletion notes now signal a condition of type
1430 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1431 an associated MUFFLE-WARNING restart.
1432 * The compiler now performs limited argument count validation of
1433 constant format strings in FORMAT, and where appropriate in ERROR,
1434 CERROR and WARN. (thanks to Gerd Moellmann)
1435 * New ASDF-INSTALL contrib can be used for automatic download and
1436 installation of third-party Lisp code from CCLAN or other sites
1438 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1439 which forces another thread to execute a function supplied by the
1441 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1442 now accept and act upon their :ELEMENT-TYPE keyword argument.
1443 (reported by Martin Atzmueller, Edi Weitz)
1444 * bug fix: FILE-POSITION now accepts position designators up to
1445 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1446 is the greater. (thanks to Patrik Nordebo)
1447 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1448 Kalle Olavi Niemitalo)
1449 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1450 printable. (reported by Eric Marsden)
1451 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1452 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1453 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1454 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1455 (thanks to Stig Sandoe)
1456 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1457 UNREAD-CHAR now correctly includes the unread character in the
1458 target sequence. (thanks to Gerd Moellmann)
1459 * bug fix in threaded builds: the system can now be suspended and
1460 resumed by shell job control with minimal disruption.
1461 * bug fixes in times and timezones >2038AD
1462 * better handling of "where is GNU make?" problem in build scripts
1463 (thanks to Nikodemus Siivola)
1464 * new optimization: inside a named function any reference to a
1465 function with the same name is considered to be a self-reference;
1466 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1467 optimization quality.
1468 * new optimization on x86: logical functions and + now have
1469 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1470 used when the result is truncated to 32 bits.
1471 * VALUES declaration is partially enabled.
1472 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1473 * fixed some bugs revealed by Paul Dietz' test suite:
1474 ** The system now obeys the constraint imposed by
1475 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1476 types form a lattice under type intersection.
1477 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1478 ** ASSOC now ignores NIL elements in an alist.
1479 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1480 and (1+ MOST-POSITIVE-FIXNUM) answers.
1481 ** The addition of a method with invalid qualifiers to a generic
1482 function does not cause an error to be signalled immediately;
1483 a warning is signalled, and the error is generated only on
1484 calling the generic function.
1485 * changed .fasl file version number, in part to add type codes for
1486 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1487 obscure ANSI requirements
1489 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1490 * fixed bug 148: failure to inline-expand a local function left
1491 garbage, confusing the compiler.
1492 * fixed bugs 3cd: structure slot readers perform type check if the
1493 slot can have an invalid value (i.e. it is either not initialized
1494 or can be written with a less specific slot writer).
1495 * bug fix: the compiler now traps array references to elements off
1496 the end of an array; previously, the bounds checking in some
1497 circumstances could go off-by-one.
1498 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1500 * type declarations for array element types now obey the description
1501 on the CLHS page "Declaration TYPE", as per discussions on
1502 sbcl-help around 2003-05-08. This means that a declaration
1503 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1504 declaration, all references to BAR will be asserted or assumed
1505 (with THE, so dependent on compiler policy) to involve objects of
1506 type FOO. Note that no such declaration is implied in
1507 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1508 * declared types of functions from the "Conditions"
1509 chapter. (reported by Paul Dietz)
1510 * bug fix: CERROR accepts a function as its first argument.
1511 * bug fix: NTH an NTHCDR accept a bignum as index
1512 arguments. (reported by Adam Warner)
1513 * optimization: character compare routines now optimize comparing
1514 against a constant character. (reported by Gilbert Baumann)
1515 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1516 argument now works properly on the MIPS platform.
1517 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1518 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1519 declared to have a functional type, does not cause an error
1521 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1522 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1523 * fixed bug: if last continuation of a deleted block has a
1524 destination, this destination should be deleted too. (reported by
1526 * fixed a bug in the bootstrap process: the host compiler's values
1527 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1528 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1529 test case from Patrik Nordebo)
1530 * improved the ability of the disassembler on the PPC platform to
1531 provide helpful disassembly notes.
1532 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1533 the class in more cases than previously.
1534 * bug fix: FILE-POSITION now understands :START and :END for
1535 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1536 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1537 8) data. (thanks to David Lichteblau)
1538 * bug fix: it is possible to add a method to a generic function
1539 without lambda list.
1540 * bug fix: reader failed to signal END-OF-FILE inside an
1541 object representation. (reported by Nikodemus Siivola)
1542 * fixed some bugs revealed by Paul Dietz' test suite:
1543 ** LAST and [N]BUTLAST should accept a bignum.
1544 ** condition slot accessors are methods.
1545 ** (VECTOR NIL) is a subtype of STRING.
1547 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1548 * minor incompatible change: some nonsensical specialized lambda
1549 lists (used in DEFMETHOD) which were previously ignored now signal
1551 * minor incompatible change: the system is now aware of the types of
1552 variables in the COMMON-LISP package, and will signal errors for
1553 most violations of these type constraints (where previously they
1554 were silently accepted).
1555 * minor incompatible change: COMPILE-FILE now uses the freedom
1556 afforded (ANSI 3.2.2.3) to use derived function types for
1557 functions defined in the same file. This also permits the system
1558 to warn on static type mismatches and function
1559 redefinition. (Currently it does not work with high DEBUG level.)
1560 * minor incompatible change: VALUES declaration is disabled.
1561 * When issuing notes, the compiler now signals a condition of type
1562 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1563 restart for use in user handlers. It is expected that the
1564 COMPILER-NOTE condition will eventually become a condition
1565 supertype to a hierarchy of note types, which will then be
1566 handleable in a similar fashion. However, at the moment, no such
1567 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1568 but it's an implementation detail, not a classification for the
1570 * Changes in type checking closed the following bugs:
1571 ** type checking of unused values (192b, 194d, 203);
1572 ** template selection based on unsafe type assertions (192c, 236);
1573 ** type checking in branches (194bc).
1574 * A short form of VALUES type specifier has ANSI meaning (it has
1575 increased the number of situations when SBCL cannot perform type
1577 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1578 keywords or constants is permissible.
1579 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1580 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1581 classes. (thanks to Antonio Martinez)
1582 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1583 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1584 argument to a pretty-print stream. Instead, it leaves any such
1585 operations to DESCRIBE-OBJECT methods.
1586 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1588 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1589 lambda lists. (thanks to David Lichteblau)
1590 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1591 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1592 (thanks to Gerd Moellmann)
1593 * fixed bug 246: increased compilation speed of long
1594 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1596 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1598 * DEFCONSTANT now throws a condition of type
1599 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1600 constant to a non-EQL value; CONTINUE and ABORT restarts
1601 respectively change and preserve the value.
1602 * fixed bug 63: The code walker, part of the implementation of CLOS,
1603 is now better at handling symbol macros.
1604 * bug fix: There is no longer an internal implementation type named
1605 CL:LENGTH. (reported by Raymond Toy)
1606 * bug fix: In macro-like defining macros/special operators the
1607 implicit block does not enclose the lambda list.
1608 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1609 enclosed in parentheses) are not suitable as type specifiers, and
1610 their use properly signals an error now.
1611 * bug fix: An argument count mismatch for a type specifier in code
1612 being compiled no longer causes an unhandled error at compile
1613 time, but signals a compile-time warning.
1614 * fixed simple vector readable printing
1615 * bug fix: DESCRIBE takes more care over whether the class
1616 precedence list slot of a class is bound before accessing it.
1617 (reported by Markus Krummenacker)
1618 * bug fix: FORMATTER can successfully compile pretty-printer format
1619 strings which use variants of the ~* directive inside.
1620 * bug fix: SEARCH now applies its TEST predicate to the elements of
1621 the arguments in the correct order. (thanks to Wolfhard Buss)
1622 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1623 in type inference. (thanks to Robert E. Brown)
1624 * bug fix: Reading in symbols with an explicit package name of ""
1625 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1626 * fixed some bugs revealed by Paul Dietz' test suite:
1627 ** NIL is now allowed as a structure slot name.
1628 ** Arbitrary numbers, not just REALs, are allowed in certain
1629 circumstances in LOOP for-as-arithmetic clauses.
1630 ** Multiple class redefinitions before slot access no longer
1631 causes a type error.
1632 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1633 association between the name and a class.
1634 ** Generic functions with non-standard method-combination and over
1635 five methods all of which return constants no longer return NIL
1636 after the first few invocations. (thanks to Gerd Moellmann)
1637 ** CALL-NEXT-METHOD with no arguments now passes the original
1638 values of the arguments, even in the presence of assignment.
1639 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1641 ** DISASSEMBLE works with closures and funcallable instances.
1642 ** ADD-METHOD now returns the generic function, not the new method.
1643 ** FIND-METHOD signals an error if the lengths of the specializers
1644 is incompatible with the generic function, even if the ERRORP
1646 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1647 which its argument is a member.
1648 ** DEFCLASS only redefines the class named by its class-name
1649 argument if that name is the proper name of the class;
1650 otherwise, it creates a new class.
1651 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1652 of the UNBOUND-SLOT condition to the name of the slot.
1653 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1654 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1655 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1656 SLOT-MAKUNBOUND in the specified fashion.
1658 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1659 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1660 cross-compilation host. As a consequence, we can now bootstrap our
1661 way up to SBCL starting with a bare gcc toolchain and human-readable
1662 source code (first the source to CLISP, then the source to SBCL).
1663 * A contributed module containing a partial implementation of the
1664 simple-streams interface has been included. (thanks to Rudi
1666 * A contributed module implementing the RFC1321 Message Digest
1667 Algorithm, known as MD5, has been included.
1668 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1669 no longer has any effect, as the code controlled by this feature
1670 has been deleted. (As far as we know, no-one has ever built using
1671 this feature, and its semantics were confused in any case).
1672 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1673 work on conditions (as required by the ANSI specification),
1674 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1675 expected behaviour on conditions. Users should note, however,
1676 that such behaviour is not required by the ANSI specification,
1677 and so use of this behaviour may render their code unportable.
1678 * fixed some bugs revealed by Paul Dietz' test suite:
1679 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1681 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1682 ** SXHASH on condition objects no longer returns NIL.
1683 ** :ALLOCATION :CLASS slots are better treated; their values are
1684 updated on class redefinition, and initforms inherited from
1685 superclasses are applied.
1686 ** REMOVE-METHOD returns its generic function argument even when
1687 no method was removed.
1688 ** SHARED-INITIALIZE now initializes the values of the requested
1689 slots, including those with :ALLOCATION :CLASS.
1690 ** ALLOCATE-INSTANCE now works on structure classes defined via
1691 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1693 ** SLOT-EXISTS-P now works on conditions, as well as structures
1695 ** MAKE-LOAD-FORM now has the required methods on
1696 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1697 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1698 keyword, and now implements the SLOT-NAMES argument.
1699 ** methods with &OPTIONAL arguments no longer allow too many
1700 arguments to be passed in the call without error.
1701 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1702 option is consistent with the required arguments of the generic
1703 function lambda list.
1704 * bug fix: REQUIRE accepts a string designator. (Thanks to
1706 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1707 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1708 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1709 * bug fix: The compiler now checks for duplicated variables in macro
1711 * bug fix: SETQ on globals returns the correct value.
1712 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1713 not a COMPILER-ERROR (followed by some other strange error on
1714 choosing the CONTINUE restart).
1715 * bug fix: make.sh and friends are now more consistent in the way that
1716 they look for GNU "make".
1718 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1719 * experimental native threads support (on x86 Linux >=2.4 only).
1720 This is not compiled in by default: you need to add :SB-THREAD to
1721 the target features. See the "Beyond ANSI" chapter of the manual
1723 * fix for longstanding nonANSIism: The old distinction between
1724 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1725 The return value from CL:FIND-CLASS is now a CLOS class, and
1726 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1727 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1728 * An interface to the MetaObject Protocol, as described in Kiczales,
1729 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1730 MIT Press, 1991, is available from the SB-MOP package.
1731 * incompatible change: the SB-PCL package should now be considered
1732 a private implementation detail, and no longer a semi-private MOP
1734 * minor incompatible change: due to rearrangement for threads, the
1735 control stack and binding stack are now allocated at arbitrary
1736 addresses instead of being hardcoded per-port. Users affected by
1737 this probably have to be doing advanced things with shared
1738 libraries, and will know who they are.
1739 * minor incompatible change: Previously, all --eval forms used were
1740 processed with READ before any of them were processed with EVAL.
1741 Now each --eval form is processed with both READ and EVAL before
1742 the next --eval form is processed. (Thus package operations like
1743 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1744 work as the user might reasonably expect.)
1745 * minor incompatible change: *STANDARD-INPUT* is now only an
1746 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1748 * minor incompatible change: Y-OR-N-P is now character-oriented,
1749 not line oriented. Also, YES-OR-NO-P now works without errors.
1750 (thanks to Antonio Martinez)
1751 * sb-aclrepl module improvements: an integrated inspector, added
1752 repl features, and a bug fix to :trace command.
1753 * Known functions, which cannot be open coded by the backend, are
1754 considered to be able to check types of their arguments. (fixing
1755 a bug report by Nathan J. Froyd)
1756 * fixed a bug in computing method discriminating functions: It is
1757 now possible to define methods specialized on classes which have
1758 forward-referenced superclasses. (thanks to Gerd Moellmann)
1759 * fixed evaluation order in optional entries (reported by Gilbert
1761 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1762 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1763 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1764 required-by-AMOP NAME argument, as well as CLASS and
1765 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1766 * fixed bug 20: DEFMETHOD can define methods using names that are
1767 not the proper names of classes to designate class specializers.
1768 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1769 Unix file descriptors, instead of blowing up. (thanks to Antonio
1771 * Garbage collection refactoring: user-visible change is that a
1772 call to the GC function during WITHOUT-GCING will not do garbage
1773 collection until the end of the WITHOUT-GCING. If you were doing
1774 this you were probably losing anyway.
1775 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1776 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1777 TYPEP the latter but not the former.
1778 * The compiler issues a full WARNING for calls to undefined functions
1779 with names from the CL package.
1780 * MAP-INTO for a vector destination is open coded. (reported by
1781 Brian Downing on c.l.l)
1782 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1783 documentation string.
1784 * fixed some bugs revealed by Paul Dietz' test suite:
1785 ** COPY-ALIST now signals an error if its argument is a dotted
1787 ** Condition slots are now accessed more correctly in the presence
1788 of multiple initargs for a given slot.
1789 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1790 exclude restarts of the same name associated with a different
1792 ** DEFCLASS of forward-referenced classes with another
1793 forward-referenced class in the superclasses list no longer
1795 ** Condition slots are now initialized once each, not multiple
1796 times. (thanks to Gerd Moellmann)
1797 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1799 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1800 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1801 arguments contain duplicated elements.
1802 ** RESTART-CASE understands local macros.
1803 ** RESTART-CASE associates exactly its own restarts with a condition.
1804 ** ENDP in safe mode checks its argument to be of type LIST.
1805 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1806 in question is unbound.
1807 ** Optimized MAKE-INSTANCE functions no longer cause internal
1808 assertion failures in the presence of duplicate initargs.
1809 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1810 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1811 propagated even in the ctor optimized implementation of
1813 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1815 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1816 * a better implementation of SXHASH on (simple) bit vectors,
1817 measured both in execution speed and in distribution of results
1818 over the positive fixnums, has been installed. Likewise, a better
1819 implementation of EQUAL for simple bit vectors is now available.
1820 * fixed CEILING optimization for a divisor of form 2^k.
1821 * fixed bug 240 (emitting extra style warnings "using the lexical
1822 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1823 by Antonio Martinez)
1824 * fixed SXHASH, giving different results for NIL depending on type
1825 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1826 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1827 variable. (found by Rolf Wester)
1828 * SBCL does not ignore type declarations for special
1829 variables. (reported by rif on c.l.l 2003-03-05)
1830 * some bug fixes in contrib/sb-aclrepl/
1831 * fixed some bugs revealed by Paul Dietz' test suite:
1832 ** a bug in the CONS type specifier, whereby the CAR and CDR
1833 types got intertwined, has been fixed;
1834 ** the type system is now able to reason about the interaction
1835 between INTEGER and RATIO types more completely;
1836 ** APPEND, [N]REVERSE and NRECONC check that those their
1837 arguments, which must be proper lists, are really so;
1838 ** An array specialized to be unable to hold elements has been
1839 implemented, as required -- yes, really -- by ANSI;
1840 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1841 on malformed property lists;
1843 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1844 * incompatible packaging change: in line with Unix convention,
1845 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1846 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1847 * REQUIRE and PROVIDE are now optionally capable of doing something
1848 useful. See the documentation string for REQUIRE.
1849 * infrastructure for a managed SBCL contrib system: contributed
1850 modules in this release include:
1851 ** the ASDF system definition facility;
1852 ** an interface to the BSD Sockets API;
1853 ** an ACL-like convenience interface to the repl;
1854 (thanks to Kevin Rosenberg)
1855 ** an implementation of ROTATE-BYTE, with efficient implementation
1857 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1858 gives a closer approximation to the right answer than previously.
1859 (thanks to Raymond Toy)
1860 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1861 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1862 optional environment arguments, as required by ANSI.
1863 * fixed bugs in other functions taking environment objects, allowing
1864 calls with an explicit NIL environment argument to be compiled
1866 * fixed bug 228: primary return values from
1867 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1868 COMPILE or FUNCTION.
1869 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1870 :TYPE will no longer signal errors on innocuous objects.
1871 * fixed bug 231b: SETQ is better at respecting type declarations in
1872 the lexical environment.
1873 * fixed a bug in DEFCLASS: classes named by symbols with no or
1874 unprintable packages can now be defined.
1875 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1876 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1877 * fixed compiler failure related to checking types of functions.
1878 (reported by Robert E. Brown)
1879 * the compiler is now much more consistent in its error-checking
1880 treatment of bounding index arguments to sequence functions: in
1881 (SAFETY 3) code, errors will be signalled in almost all cases if
1882 invalid sequence bounding indices are passed to functions defined
1883 by ANSI to operate on sequences.
1884 * fixed a bug in the build procedure: documentation of SBCL-specific
1885 packages is now preserved and available in the final Lisp image.
1886 * lifted FDEFINITION lookup out of loops in the implementation of
1887 many list operations. (thanks to Robert E. Brown)
1888 * fixed a bug in the reader: the #n# reader macro now works for
1889 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1890 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1891 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1892 E. Brown and Tony Martinez respectively)
1893 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1894 SLOT-BOUNDP: in particular, it is now able to optimize them much
1895 better, and is now not vulnerable to having packages renamed.
1896 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1897 always called when a slot is not present in an instance. (thanks
1899 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1900 longer signal an error on unreading a character following EOF on
1901 the previous constituent stream. (thanks to Tony Martinez)
1902 * fixed some bugs revealed by Paul Dietz' test suite:
1903 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1904 not just nonnegative fixnums;
1905 ** the logical bit-array operators such as BIT-AND now accept an
1906 explicit NIL for their "opt-arg" argument (to indicate a
1907 freshly-consed result bit-array);
1908 ** ELT now signals an error on an invalid sequence index in safe
1910 ** the type system is now cleverer about negations of numeric
1911 types, and consequently understands the BIGNUM and RATIO types
1913 ** the type system is now cleverer about the interaction between
1914 INTEGER and RATIO types: while bugs still remain, many more
1915 cases are accurately computed;
1916 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1917 if it is in the last clause;
1918 ** CONSTANTLY now correctly returns a side-effect-free function in
1920 ** DECLARE is no longer treated as a special-operator; in
1921 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1922 * incremented fasl file version number due to the change in the
1923 DEFSTRUCT-SLOT-DESCRIPTION structure.
1925 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1926 * minor incompatible change: code processed by the "interpreter" or
1927 EVAL now has a compilation optimization policy of (DEBUG 2)
1928 (changed from (DEBUG 1)) to improve debuggability of interactive
1929 development, and to allow the use of the debug RETURN command in
1931 * an experimental implementation of the RETURN command for the
1932 debugger has been included. (thanks to Frederik Kuivinen)
1933 * fixed bug 62: constraints were not propagated into a loop.
1934 * fixed bug in embedded calls of SORT (reported and investigated by
1936 * fixed some bugs revealed by Paul Dietz' test suite:
1937 ** printing and reading of arrays with some dimensions having
1938 length 0 (thanks to Gerd Moellmann);
1939 ** BOA constructor with &AUX argument without a default value does
1940 not cause a type error;
1941 ** CONSTANTP now returns true for all self-evaluating objects.
1943 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1944 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1945 accessors that are related by inheritance, as specified in the
1946 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1947 Valtteri Vuorikoski)
1948 * The compiler is now able to inline functions that were defined in
1949 a complex lexical environment (e.g. inside a MACROLET).
1950 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1952 * Support for the upcoming FreeBSD-5.0 release has been included.
1953 (thanks to Dag-Erling Smorgrav)
1954 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1955 effect when it is not in a toplevel context.
1956 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1957 stand a better chance of being correct. (thanks to Gerd
1959 * fixed bug in COERCE, which now signals an error on coercing a
1960 rational to a bounded real type which excludes the expected
1962 * The compiler is now able to derive types more accurately from the
1963 COERCE and COMPILE functions.
1964 * fixed bug 223: functional binding is considered to be constant
1965 only for symbols in the CL package.
1966 * fixed bug 231: SETQ did not check the type of a variable being set
1967 (reported by Robert E. Brown)
1968 * A new optimization for MAKE-INSTANCE has been included, fixing
1969 various bugs (including relating to :ALLOCATION :CLASS slots and
1970 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1972 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1973 ** As required by ANSI, LOOP now disallows anonymous collection
1974 clauses such as COLLECT I in conjunction with aggregate boolean
1975 clauses such as THEREIS (= I 1);
1976 ** LOOP now signals an error when any variable is reused in the
1977 same loop (including the potentially useful construct analogous
1978 to WITH A = 1 WITH A = (1+ A);
1979 ** IT is only a special loop symbol within the first clause of a
1980 conditional loop clause;
1981 ** LOOP with a typed iteration variable over a hashtable now
1982 signals a type error iff it should.
1983 * fixed some other bugs revealed by Paul Dietz' test suite:
1984 ** FILE-STREAM now names the class previously known as FD-STREAM;
1985 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1986 argument) no longer signals an error;
1987 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1988 of the slot symbol, rather than using the current package
1989 ((:CONC-NAME "") continues to intern the slot's name in the
1991 * incremented fasl file version number, because of the incompatible
1992 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1993 of the new implementation of DEFINE-COMPILER-MACRO.
1995 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1996 * Support for building SBCL for MIPS platforms running in
1997 little-endian mode has now been checked in, and basic
1998 functionality on said platforms verified.
1999 * minor incompatible change: PCL now records the pathname of a file
2000 in which methods and the like are defined, rather than its
2002 * minor incompatible change: TRUENAME now considers the truename of
2003 a file naming a directory to be the pathname with :DIRECTORY
2004 component indicating that directory.
2005 * minor incompatible change: a NAMED clause in the extended form of
2006 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2007 reason for the previous behaviour is unclear.
2008 * more systematization and improvement of CLOS and MOP conformance
2009 in PCL (thanks to Gerd Moellman and Pierre Mai):
2010 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2012 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2013 primary methods with no specializers;
2014 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2016 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2017 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2018 FORWARD-REFERENCED-CLASSes; error reporting on
2019 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2021 ** SXHASH on CLOS instances now uses a slot internal to the
2022 instance to return different numbers on distinct instances,
2023 while preserving the same return value through invocations of
2025 ** DEFMETHOD signals errors when methods with longer incongruent
2026 lambda lists are added to generic functions;
2027 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2028 CLASS, as specified in AMOP;
2029 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2030 on the order returned by the primary method for classes of
2031 class STANDARD-CLASS;
2032 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2033 * fixed some bugs shown by Paul Dietz' test suite:
2034 ** DOLIST puts its body in TAGBODY;
2035 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2037 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2038 value producing form;
2039 ** if more variables are given to PROGV than values, extra
2040 variables are bound and made to have no value;
2041 ** NSUBSTITUTE on list arguments gets the right answer with
2043 ** ELT signals an error of type TYPE-ERROR when the index argument
2044 is not a valid sequence index;
2045 ** LOOP signals (at macroexpansion time) an error of type
2046 PROGRAM-ERROR when duplicate variable names are found;
2047 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2048 ** FUNCALL of special-operators now cause an error of type
2050 ** PSETQ now works as required in the presence of side-effecting
2051 symbol-macro places;
2052 ** NCONC accepts any object as its last argument;
2053 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2055 ** loop-for-as-package does not require a package to be explicitely
2057 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2059 * fixed bug 166: compiler preserves "there is a way to go"
2060 invariant when deleting code.
2061 * fixed bug 172: macro lambda lists with required arguments after
2062 &REST arguments now cause an error to be signalled. (thanks to
2064 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2065 bug. (thanks to Matthew Danish)
2066 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2068 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2069 arguments are lexically rebound. (thanks to Gerd Moellmann and
2071 * fixed bug 194: error messages are now more informative when there
2072 is no primary method applicable in a call to a generic
2073 function. (thanks to Gerd Moellmann)
2074 * fixed bug in command line argument checking (thanks to Julian
2076 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2077 (thanks to Matthew Danish)
2078 * incremented fasl file version number, because of the
2079 SXHASH-related changes in the layout of CLOS data structures
2081 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2082 * minor incompatible change: The runtime (the Unix executable named
2083 "sbcl") is now much pickier about the .core files it will load.
2084 Essentially it now requires .core files to descend from the same
2085 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2086 as the runtime does. (The intent is to prevent the crashes which
2087 can occur, and which can even be reported as mysterious failures,
2088 when people patch the sources or change the build parameters
2089 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2090 sbcl and .core files.)
2091 * fixed bug: VALUES-LIST is no longer optimized away.
2092 * fixed bug 142: The FFI conversion of C string values to Lisp
2093 string values no longer conses excessively. (thanks to Nathan
2094 Froyd porting Raymond Toy's fix to CMU CL)
2095 * began to systematize and improve MOP conformance in PCL (thanks to
2096 Nathan Froyd, Gerd Moellman and Pierre Mai):
2097 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2099 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2100 ** FINALIZE-INHERITANCE is now called on class finalization;
2101 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2102 argument precedence order.
2103 * fixed bug 202: The compiler no longer fails on functions whose
2104 derived types contradict their declared type.
2105 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2106 so it can be non-toplevel.
2107 * The fasl file version number has changed (because of the new
2108 implementation of DEFMACRO).
2109 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2110 safe code, that any length requirement by their type-specifier
2111 argument is valid. The exceptions to this are described in bug
2113 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2114 functions now check the types of their inputs as required by ANSI.
2115 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2116 introduced symbol is DECLAREd to be SPECIAL.
2117 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2118 * fixed bug: :COUNT argument to sequence functions may be negative
2119 * fixed bug: body of DO-SYMBOLS may contain declarations
2120 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2121 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2122 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2123 symbol macro only once
2124 * fixed printing of call frame when argument list is unavailable
2125 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2126 * compiler no longer signals WARNING on unknown keyword
2129 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2130 * A beta-quality port to the mips architecture running Linux,
2131 based on the old CMUCL backend, has been made. It has been tested
2132 on a big-endian kernel, and works sufficiently well to be able to
2133 rebuild itself; it has not been tested in little-endian mode.
2134 * fixed an inconsistency between gencgc.c and purify.c which made
2135 dumping/loading .core files unreliable
2136 * fixed bug 120a: The compiler now deals correctly with IFs where
2137 the consequent is the same as the alternative, instead of
2138 misderiving the return type. (thanks to Alexey Dejneka)
2139 * fixed bug 113: Logical pathnames are now dumpable (the logical
2140 host is resolved at load-time, throwing an error if it is not
2142 * fixed bug 174: FORMAT's error message is slightly clearer when a
2143 non-printing character is used in a format directive.
2144 * fixed several bugs in compiler checking of type declarations, i.e.
2145 violations of the Python "declarations are assertions" principle
2146 (thanks to Alexey Dejneka)
2147 * fixed several bugs in PCL's error checking (thanks to Gerd
2149 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2151 * fixed bug in compilation of functions as first class values
2152 (thanks to Antonio Martinez-Shotton)
2153 * The compiler's handling TYPE-ERRORs which it can prove will
2154 inevitably happen at runtime has been cleaned up and corrected
2155 in several ways. (thanks to Alexey Dejneka)
2156 * improved argument type checking for various basic arithmetic
2157 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2158 lavished on them in the past that they can be compiled in many
2159 ways in different special cases
2160 * fixed bug 181: compiler checks validity of user supplied type
2162 * cleaned up code flushing in optimization: Function calls which
2163 should signal errors for safety purposes (e.g. which ANSI says
2164 should signal errors when their arguments are of incorrect type)
2165 are no longer optimized away.
2166 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2167 * incremented fasl file version number, because changes in the
2168 implementation of sequence functions like COERCE caused
2169 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2170 in old inline expansions) to become undefined. (Actually these
2171 changes were later undone, so we might very well be binary
2172 compatible with 0.7.7 after all, but leaving the version number
2173 incremented seemed like the simplest and most conservative
2176 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2177 * An alpha-quality port to the parisc architecture running Linux,
2178 based on the old CMUCL backend, has been made. This, even more so
2179 than the other backends, should be considered still a work in
2180 progress; known problems include that the Linux kernel in 64-bit
2181 mode does not propagate the correct sigcontext structure to
2182 userspace, and consequently SBCL on a parisc64 kernel will not
2184 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2185 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2186 suppressed.) Also now that the compiler is looking at declarations
2187 in the environment, it checks optimization declarations as well,
2188 and suppresses inlining when (> DEBUG SPEED).
2189 * More fixes have been made to treatment of floating point exception
2190 treatment and other Unix signals. In particular, floating point
2191 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2192 * The detection and handling of control stack exhaustion (infinite
2193 or very deeply nested recursion) has changed. Stack exhaustion
2194 detection is now done by write-protecting pages at the OS level
2195 and applies at all optimization settings; when found, a
2196 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2197 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2198 be caught using IGNORE-ERRORS.
2199 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2200 correctly when transforming recursive LABELS functions to LETs.
2201 (thanks to Alexey Dejneka)
2202 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2203 that are names of constants or global variables.
2204 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2205 alien routines with docstrings.
2206 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2207 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2209 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2210 (thanks to Eric Marsden)
2211 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2212 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2213 * Bug fix: The SPARC backend can now compile functions involving
2214 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2215 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2216 to a functions expecting a PATHNAME-DESIGNATOR.
2217 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2218 lambda lists. (thanks to Alexey Dejneka)
2219 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2220 functions so that EXPORTing the name of the class doesn't cause
2221 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2222 to Antonio Martinez for reporting this)
2223 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2224 constant-folded without failing an assertion. (thanks to Einar
2225 Floystad Dorum for reporting this)
2226 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2227 types are now dealt with more correctly by the compiler.
2228 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2229 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2230 behaviour analogous to RENAME-FILE. This puts its behaviour more
2231 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2233 * The fasl file version number has changed again. (because of the
2234 bug fix involving the names of PCL MAKE-INSTANCE functions)
2236 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2237 * bug fix: Floating point exceptions are treated much more
2238 consistently on the x86/Linux and PPC/Linux platforms.
2239 * Array initialization with :INITIAL-ELEMENT is now much faster for
2240 cases when the compiler cannot open code the array creation, but
2241 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2242 array accesses have also seen a speed increase.
2243 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2244 specified. (This was at the root of some bad interactions between
2245 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2247 * bug fix: Internal error arguments for undefined functions are now
2248 computed correctly on the PPC/Linux platform.
2249 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2250 Raymond Toy's patch for CMU CL)
2251 * Support for the Solaris 9 operating environment has been included
2252 (thanks to Daniel Merritt)
2253 * A very ugly but hopefully complete draft of the missing FFI chapter
2254 of the manual has been created by reformatting the corresponding
2255 CMU CL manual chapter into (currently very ugly and incoherent)
2256 DocBook and bringing it up to date for SBCL behavior. Thus, the
2257 manual is now essentially complete, at least by my extreme
2258 once-and-only-once standards, whereby it's acceptable to refer to
2259 the doc strings of SB-EXT functions as the primary documentation.
2260 * The fasl file version number has changed again, due to cleanup of
2261 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2262 is no longer a static symbol.)
2264 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2265 * SBCL now builds with OpenMCL (version 0.12) as the
2266 cross-compilation host; also, more progress has been made toward
2267 bootstrapping under CLISP.
2268 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2270 * bug 158 fixed: The compiler can now deal with integer loop
2271 increments different from 1; fixing this turned out also to fix
2273 * bug 169 fixed: no more bogus warnings about using lexical bindings
2274 despite the presence of perfectly good SPECIAL declarations (thanks
2275 to David Lichteblau)
2276 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2277 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2278 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2280 * bug 180 fixed: Method combination specifications no longer ignore
2281 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2282 * bug fix: Structure type predicate functions now check their argument
2283 count as they should.
2284 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2285 correctly. (thanks to Pierre Mai)
2286 * minor incompatible change: The --noprogrammer option is deprecated
2287 in favor of the new --disable-debugger option, which is very similar.
2288 (The major difference is that it takes effect at a slightly different
2289 time at startup, causing handling of errors in --sysinit and
2290 --userinit files will be affected differently.) The
2291 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2292 been added to allow this functionality to be controlled from ordinary
2293 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2294 maintainers, who might want to run non-interactive scripts to
2295 build SBCL cores which will later be used interactively.)
2296 * minor incompatible change: The LOAD function no longer, when given
2297 a wild pathname to load, loads all files matching that pathname.
2298 Instead, an error of type FILE-ERROR is signalled.
2300 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2301 * bug 147 fixed: The compiler preserves its block link/count
2302 invariants more correctly now so that it doesn't crash. (thanks
2304 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2306 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2307 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2308 no-value-for-this-character as NIL (as specified by ANSI).
2309 * HOST-NAMESTRING on physical pathnames now returns a string that is
2310 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2311 (thanks to Christophe Rhodes)
2312 * The Alpha port handles icache flushing more correctly. (thanks to
2314 * More progress has been made toward bootstrapping under CLISP. (thanks
2315 to Christophe Rhodes)
2316 * The fasl file format has changed again, because dynamic loading
2317 on OpenBSD (which has non-ELF object files) motivated some cleanups
2318 in the way that foreign symbols are transformed and passed around.
2319 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2320 is no longer treated as whitespace by the reader, but instead as
2321 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2322 |A
\7fB|, instead of A as it used to.
2324 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2325 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2326 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2327 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2328 this as of 0.7.1.45, but was left out of the previous news section
2329 (thanks to Dan Barlow)
2330 * SBCL now runs on the Solaris operating system on SPARC architectures
2331 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2332 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2333 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2334 * SPARC backend cleanups, allowing builds of cores optimized for V8
2335 and V9 SPARCS, and also emission of code targeted to a particular
2336 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2338 * SBCL is closer to bootstrapping under CLISP, thanks to various
2339 fixes by Christophe Rhodes.
2340 * The fasl file format has changed again, to allow the compiler's
2341 INFO database to support symbol macros.
2342 * The user manual (in doc/) is formatted into HTML more nicely.
2343 (thanks to coreythomas)
2344 * The system is smarter about SUBTYPEP relationships, especially
2345 those involving NOT types (including types such as ATOM which are
2346 represented internally using NOT types). Thus SUBTYPEP is less
2347 likely to return (VALUES NIL NIL) in general, and in particular
2348 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2350 * The fasl file format has changed again, because the internal
2351 representation of types now includes a new slot to support the new
2352 SUBTYPEP-of-NOT-types logic.
2353 * (not a change in the main branch of SBCL, but a related prototype
2354 which can hopefully be merged into the main branch of SBCL in the
2355 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2356 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2357 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2358 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2359 dispatch macro character combinations. (thanks to Alexey Dejneka)
2360 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2361 that superficially look like logical namestrings correctly.
2362 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2364 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2365 default supplied-p) for &optional and &key arguments. (thanks to
2368 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2369 * incompatible change: The compiler is now less aggressive about
2370 tail call optimization, doing it only when (> SPACE DEBUG) or
2371 (> SPEED DEBUG). (This is an incompatible change because there are
2372 programs which relied on the old CMU-CL-style behavior to optimize
2373 away their unbounded recursion which will now die of stack overflow.)
2374 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2375 for non-GENCGC systems has been increased to 20M (since that
2376 seems much closer to the likely performance optimum for modern
2377 systems than the old 4M value was)
2378 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2379 and *DEBUG-PRINT-LEVEL*
2380 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2381 of CMU CL's support for SPARC, and various endianness and other
2382 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2383 * new syntactic sugar for the Unix command line: --load foo.bar is now
2384 an alternate notation for --eval '(load "foo.bar")'.
2386 ** The system now detects stack overflow and handles it gracefully,
2387 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2388 optimization settings. (This is a good thing in general, and
2389 its introduction in this version should be particularly timely
2390 for anyone whose code fails because of suppression of tail
2392 ** The system now hunts for the C variable "environ" in a more
2393 devious way, to avoid segfaults when the C library version differs
2394 between compile time and run time. (thanks to Christophe Rhodes)
2395 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2396 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2397 ** The compiler no longer issues bogus style warnings for undefined
2398 classes in the same source file as the DEFCLASSes which defined
2399 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2401 ** fixes in CONDITION class precedence list for undefined function
2402 errors (thanks to Alexei Dejneka)
2403 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2404 correctly. (thanks to Dan Barlow)
2405 ** portability fixes aiming at bootstrapping under CLISP (thanks
2406 to Dave McDonald and Christophe Rhodes)
2407 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2408 ** fixes in type translation and and type inference (thanks to
2410 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2411 ** various fixes in the new ports (thanks to Dan Barlow)
2412 * several changes related to debugging:
2413 ** suppression of tail recursion, as noted above
2414 ** stack overflow detection, as noted above
2415 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2416 is now the default. (For some time encapsulation has been more
2417 reliable than the breakpoint-based :ENCAPSULATE NIL
2418 implementation, at least on X86 systems; and I just noticed that
2419 encapsulation also seems closer to the spirit of the ANSI
2422 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2424 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2425 up properly again. (There was a packaging bug in 0.7.0 which
2426 left their definitions in SB-SYS::LOAD-FOREIGN and
2427 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2428 vital for most things which interface to C-level interfaces,
2429 like extensions working with sockets or databases or
2430 Perl-compatible regexes or whatever, and the need to fix
2431 this bug is the main reason that 0.7.1 was released so
2433 ** DEFGENERIC is now choosier about the methods it redefines, so that
2434 reLOADing a previously-LOADed file containing DEFGENERICs does
2435 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2436 cycle now works as it should. (thanks to Alexey Dejneka)
2437 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2438 fixed by Christophe Rhodes. (He actually submitted this patch
2439 months ago, and I delayed until after 0.7.0.)
2440 ** Bug 111 (internal compiler confusion about runtime checks on
2441 FUNCTION types) was fixed by Alexey Dejneka.
2442 * Some internal cleanups (getting rid of variables which aren't
2443 needed now that the byte interpreter is gone) caused the fasl
2444 file format number to change again.
2446 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2447 * major incompatible change: The default fasl file extension, i.e. the
2448 default extension for files produced by COMPILE-FILE, has changed
2449 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2451 ** There are many changes in the implementation of the compiler.
2452 SBCL is now essentially a compiler-only implementation of ANSI
2453 Common Lisp. EVAL still "interprets" a few special cases, but
2454 almost all the interesting cases are handled by creating
2455 a LAMBDA expression, calling COMPILE on it, then calling
2456 FUNCALL on the result.
2457 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2458 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2459 Since the code is newer, there might still be some new bugs
2460 (though not as many as before Martin Atzmueller's fixes:-). But
2461 the new code is substantially simpler and clearer, and hopefully
2462 any remaining bugs will be simpler, less fundamental, and more
2463 fixable then the bugs in the old code.
2464 ** The revised compiler is still a little unsteady on its feet.
2466 *** The debugging information it produces (particularly the names
2467 of FUNCTION objects) is sometimes much less useful than what
2468 the old compiler produced.
2469 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2470 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2471 within a MACROLET) has been temporarily weakened.
2472 ** There are new compiler optimizations for various functions:
2473 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2474 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2475 *** the math functions TRUNCATE, FLOOR, and CEILING
2476 *** the function-of-all-trades COERCE
2477 Mostly these should be transparent, but there's one
2478 potentially-annoying problem (bug 117): when the compiler
2479 inline-expands a function and does type analysis on the result,
2480 it can create control paths which have type mismatches, and
2481 when it can't prove that those control paths aren't taken,
2482 it will issue WARNINGs about the type mismatches. This is
2483 a particular problem in practice for the new sequence functions.
2484 It's not clear how this should be fixed, and for now, a
2485 workaround is given in the entry for 117 in the BUGS file.
2486 ** (Because of the interaction between the two previous items --
2487 occasional inlining problems and new inline expansions -- some
2488 of the new sequence function optimizations won't really kick in
2489 completely until debugging information, and then inlining, are
2490 straightened out in some future version.)
2491 * minor incompatible changes:
2492 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2493 DIRECTORY no longer implicitly promotes NIL slots of its
2494 pathname argument to :WILD. In particular, when you ask for the
2495 contents of a directory (which you used to be able to do without
2496 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2497 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2498 ** changes in behavior that ANSI explicitly defines to be
2499 implementation dependent:
2500 *** The new compiler-only implementation still conforms with ANSI,
2501 but acts a little different than before. Besides the obvious
2502 changes in performance tradeoffs (that the cost per form passed
2503 to EVAL has gone up, and the cost per form executed by EVAL
2504 has gone down), the behavior of the system changes a little
2505 because there are no longer any interpreted function objects.
2506 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2507 e.g. doing COMPILE on the output of interactive DEFUN is
2509 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2511 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2512 has changed from (:CL) to NIL.
2513 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2514 e.g. #\Newline and #\Tab, don't have names specified in the
2515 ANSI Common Lisp standard, is now based on their ASCII symbolic
2516 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2517 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2518 are no longer used for output.
2519 ** changes in internal implementation constants:
2520 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2521 4 million. (If your application spends a lot of time GCing and
2522 you have a lot of RAM, you might want to experiment with
2523 increasing it even more.)
2524 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2525 However, almost all old code should still continue to work without
2526 immediate update, as SB-C-CALL is now a (deprecated) nickname
2528 ** Old operator names in the style DEF-FOO are now deprecated in
2529 favor of new corresponding names DEFINE-FOO, for consistency with
2530 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2531 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2532 internal symbols, but a few supported extensions like
2533 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2534 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2535 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2536 etc. as you get deeper into recursive calls to the debugger
2537 command loop, instead of the old "5]", "5]]", "5]]]"
2538 sequence. (I was motivated to do this when squabbles between
2539 ILISP and SBCL left me very deeply nested in the debugger. In the
2540 short term, this change will probably provoke more ILISP/SBCL
2541 squabbles, but hopefully it will be an improvement in the long run.)
2542 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2543 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2544 function, since declaiming return types involving aliens is
2545 (1) annoyingly messy to do by hand and (2) vital to efficient
2546 compilation of code which calls such functions.
2547 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2548 longer reexported by the SB-EXT package. They're solely useful
2549 for alien code, so it seems more logical that you should get
2550 them from the SB-ALIEN package, not in SB-EXT.
2551 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2552 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2553 features. Instead, the code that they used to control is always
2554 built into the system.
2555 * many other bug fixes
2556 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2557 advantage of the new EVAL-WHEN stuff and to clean them up in
2558 general, and they are now more ANSI-compliant in a number of
2559 ways. Martin Atzmueller is responsible for a lot of this.
2560 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2562 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2563 *** correct ERROR type for various file operations
2564 *** some fixes for Lisp streams
2565 *** DEFMETHOD syntax checking
2566 *** changing old weird representation of debug information as
2567 strings (which, among their other deficiencies, don't transform
2568 correctly when you rename packages, and don't change their
2569 print representation when you change things like *PACKAGE*
2570 and *PRINT-LENGTH*) to symbols and lists of symbols
2571 He also made several improvements and fixed several bugs in DESCRIBE.
2572 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2574 *** misbehavior of WRITE-STRING/WRITE-LINE
2575 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2576 and several other LOOP problems as well
2577 *** DIRECTORY when similar filenames are present
2578 *** DEFGENERIC with :METHOD options
2579 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2580 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2581 *** argument ordering in FIND with :TEST option
2582 *** mishandled package designator argument in APROPOS-LIST
2583 *** various problems in the backquote readmacro
2584 *** a bug in APROPOS
2585 *** probably some others that I'm not describing very well here,
2586 since the CVS log documents them by reference to sbcl-devel
2587 messages, and the SourceForge archives aren't working well.:-(
2588 ** Dan Barlow improved the Alpha port (and is making progress on the
2589 PPC port, for those of you who think different).
2590 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2591 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2592 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2593 continues to work on the SPARC port (for those of you in a position
2594 to look down upon our little PC-compatible boxes from a great height).
2595 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2596 and uses that copy, rather than the current dynamic value, when
2597 it's trying to decide whether to truncate output. Thus e.g.
2598 (let ((*print-lines* 50))
2599 (pprint-logical-block (stream nil)
2601 (let ((*print-lines* 8))
2602 (print (aref possiblybigthings i) stream)))))
2603 should now truncate the logical block only at 50 lines, instead of
2604 often truncating it at 8 lines, as it did before.
2605 * The doc/cmucl/ directory, containing old CMU CL documentation
2606 from the time of the fork, is no longer part of the base system.
2607 SourceForge has shut down its anonymous FTP service, and with it
2608 my original plan for distributing the old CMU CL documentation
2609 there. For now, if you need these files you can download an old
2610 SBCL source release and extract them from it.
2611 * The fasl file version number changed again, for dozens of reasons,
2612 some of which are apparent above.
2614 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2615 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2616 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2617 patch, so that DISASSEMBLE output is much nicer.
2618 * The code in the SB-PROFILE package now seems reasonably stable.
2619 I still haven't decided what the final interface should look like
2620 (I'd like PROFILE to interact cleanly with TRACE, since both
2621 facilities use function encapsulation) but if you have a need
2622 for profiling now, you can probably use it successfully with
2623 the current CMU-CL-style interface.
2624 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2625 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2626 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2627 link now returns the dangling link itself, and for similar
2628 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2629 link itself. (In these cases the old code signalled an error and
2630 looped endlessly, respectively.) Thus, DIRECTORY now works even
2631 in the presence of dangling and cyclic symbolic links.
2632 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2633 is now a supported extension again, since the consensus on
2634 sbcl-devel was that it can be useful for ordinary development
2635 work, not just for debugging SBCL itself.
2636 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2637 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2638 that currently-defined functions might be redefined later with
2639 different return types.
2640 * Hash tables can be printed readably, as inspired by CMU CL code
2641 of Eric Marsden and SBCL code of Martin Atzmueller.
2642 * better error handling in CLOS method combination, thanks to
2643 Martin Atzmueller porting Pierre Mai's CMU CL patches
2644 * more overflow fixes for >16Mbyte I/O buffers
2645 * A bug in READ has been fixed, so that now a single Ctrl-D
2646 character suffices to cause end-of-file on character streams.
2647 In particular, now you only need one Ctrl-D at the command
2648 line (not two) to exit SBCL.
2649 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2651 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2652 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2653 does the right thing.
2654 * The implementation of some type tests, especially for CONDITION
2655 types, is now tidier and maybe faster, due to CMU CL code
2656 originally by Douglas Crosher, ported by Martin Atzmueller.
2657 * Some math functions have been fixed, and there are new
2658 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2659 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2660 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2661 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2662 likely to be built into sbcl-0.7.0.)
2663 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2664 CL, ported to SBCL by Martin Atzmueller)
2665 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2666 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2667 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2668 * There's a new slam.sh hack to shorten the edit/compile/debug
2669 cycle for low-level changes to SBCL itself, and a new
2670 :SB-AFTER-XC-CORE target feature to control the generation of
2671 the after-xc.core file needed by slam.sh.
2672 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2673 COMPILE-FILE is no longer supported, so that now every function
2674 gets an entry point, so that block compilation looks a little
2675 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2676 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2677 returns the number of bytes consed since the system started,
2678 rather than the number consed since the first time the function
2679 was called. (The new definition parallels ANSI functions like
2680 CL:GET-INTERNAL-RUN-TIME.)
2681 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2682 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2683 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2684 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2685 * Fasl file version numbers are now independent of the target CPU,
2686 since historically most system changes which required version
2687 number changes have affected all CPUs equally. Similarly,
2688 the byte fasl file version is now equal to the ordinary
2691 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2692 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2693 is no longer recognized. I apologize for this, because it was
2694 listed in SB-EXT as a supported extension, but I found that
2695 its existing behavior was poorly specified, as well as incorrectly
2696 specified, and it looked like too much of a mess to straighten it
2697 out. I have enough on my hands trying to get ANSI stuff to work..
2698 * many patches ported from CMU CL by Martin Atzmueller, with
2699 half a dozen bug fixes in pretty-printing and the debugger, and
2700 half a dozen others elsewhere
2701 * fixed bug 13: Floating point infinities are now supported again.
2702 They might still be a little bit flaky, but thanks to bug reports
2703 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2704 as flaky as they were.
2705 * The --noprogrammer command line option is now supported. (Its
2706 behavior is slightly different in detail from what the old man
2707 page claimed it would do, but it's still appropriate under the
2708 same circumstances that the man page talks about.)
2709 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2710 are now supported, and enabled by default. Thus, the compiler can
2711 handle many floating point and complex operations much less
2712 inefficiently. (Thus e.g. you can implement a complex FFT
2714 * The compiler now detects type mismatches between DECLAIM FTYPE
2715 and DEFUN better, and implements CHECK-TYPE more correctly, and
2716 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2717 thanks to patches from Martin Atzmueller.
2718 * various fixes to make the cross-compiler more portable to
2719 ANSI-conforming-but-different cross-compilation hosts (notably
2720 Lispworks for Windows, following bug reports from Arthur Lemmens)
2721 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2722 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2723 fixed thanks to Pierre Mai's CMU CL patches.
2724 * fixes to keep the system from overflowing internal counters when
2725 it tries to use i/o buffers larger than 16M bytes
2726 * fixed bug 45a: Various internal functions required to support
2727 complex special functions have been merged from CMU CL sources.
2728 (When I was first setting up SBCL, I misunderstood a compile-time
2729 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2730 * improved support for type intersection and union, fixing bug 12
2731 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2732 more obscure bugs as well
2733 * some steps toward byte-compiling non-performance-critical
2734 parts of the system, courtesy of patches from Martin Atzmueller
2735 * Christophe Rhodes has made some debian packages of sbcl at
2736 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2737 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2738 stable, but are nonetheless usable. When he's ready, I'd be happy
2739 to add them to the SourceForge "File Releases" section. (And if
2740 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2741 * new fasl file format version number (because of changes in
2742 internal representation of (OR ..) types to accommodate the new
2743 support for (AND ..) types, among other things)
2745 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2746 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2748 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2750 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2751 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2752 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2753 * DESCRIBE now works on structure objects again.
2754 * Most function call argument type mismatches are now handled as
2755 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2756 know whether the function will be redefined before the call is
2757 executed. (The compiler could flag local calls with full WARNINGs,
2758 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2759 it doesn't keep track of enough information to know whether calls
2760 are local in this sense.)
2761 * Compiler output is now more verbose, with messages truncated
2762 later than before. (There should be some supported way for users
2763 to override the default verbosity, but I haven't decided how to
2764 provide it yet, so this behavior is still controlled by the internal
2765 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2766 src/compiler/ir1util.lisp.)
2767 * Fasl file format version numbers have increased again, because
2768 support for the Gray streams extension changes the layout of the
2769 system's STREAM objects.
2770 * The Gray subclassable streams extension now works, thanks to a
2771 patch from Martin Atzmueller.
2772 * The full LOAD-FOREIGN extension (not just the primitive
2773 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2774 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2775 but like most other programs, it defaults to copying the Unix
2776 environment from the original process instead of starting the
2777 new process in an empty environment.
2778 * Extensions which manipulate the Unix environment now support
2779 an :ENVIRONMENT keyword option which doesn't smash case or
2780 do other bad things. The CMU-CL-style :ENV option is retained
2781 for porting convenience.
2782 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2783 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2785 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2787 * A patch from Martin Atzmueller seems to have solved the SIGINT
2788 problem, and as far as we know, signal-handling now works cleanly.
2789 (If you find any new bugs, please report them!)
2790 * The system no longer defaults Lisp source file names to types
2791 ".l", ".cl", or ".lsp", but only to ".lisp".
2792 * The compiler no longer uses special default file extensions for
2793 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2794 seems to expect a single default extension for all compiled code,
2795 and there's no compelling reason to try to stretch the standard
2796 to allow two different extensions.) Instead, byte-compiled files
2797 default to the same extension as native-compiled files.
2798 * Fasl file format version numbers have increased again, because
2799 a rearrangement of internal implementation packages made some
2800 dumped symbols in old fasl files unreadable in new cores.
2801 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2802 many fewer weird special cases.
2803 * Bug #17 (differing COMPILE-FILE behavior between logical and
2804 physical pathnames) has been fixed, and some related misbehavior too,
2805 thanks to a patch from Martin Atzmueller.
2806 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2807 by Tim Moore, ported to SBCL by Martin Atzmueller.
2808 * Martin Atzmueller fixed several filesystem-related problems,
2809 including bug #36, in part by porting CMU CL patches, which were
2810 written in part by Paul Werkowski.
2811 * More compiler warnings in src/runtime/ are gone, thanks to
2812 more patches from Martin Atzmueller.
2813 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2816 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2818 * DESCRIBE now works on CONDITION objects.
2819 * The debugger now handles errors which arise when trying to print
2820 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2822 * The build system now uses an additional file, customize-target-features.lisp,
2823 to allow local modifications to the target *FEATURES* list. (The point of
2824 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2825 debugging features enabled, without having to worry about propagating it
2826 into everyone's system when I do a "cvs update".) When no
2827 customize-target-features.lisp file exists, the target *FEATURES* list
2828 should be constructed the same way as before.
2829 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2830 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2831 warn about behavior which is undefined under the ANSI spec. Note: This
2832 is specified by ANSI, but it's not very popular with programmers.
2833 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2834 macro in the SBCL sources for an example of a workaround which you
2835 might use to make portable ANSI-standard code which does what you want.
2836 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2837 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2838 when it's used as a non-toplevel form. (This is required in order
2839 to implement the DEFCONSTANT-EQX macro.)
2840 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2841 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2842 funny things, due to bugs in the implementation of EVAL-WHEN
2843 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2844 non-toplevel DEFCONSTANTs are uncommon.)
2845 * The core file version number and fasl file version number have been
2846 incremented, because the old noncompliant DEFCONSTANT behavior involved
2847 calling functions which no longer exist, and because I also took the
2848 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2849 * fixed bug 1 (error handling before read-eval-print loop starts), and
2850 redid debugger restarts and related debugger commands somewhat while
2852 ** The QUIT debugger command is gone, since it did something
2853 rather different than the SB-EXT:QUIT command, and since it never
2854 worked properly outside the main toplevel read/eval/print loop.
2855 Invoking the new TOPLEVEL restart provides the same functionality.
2856 ** The GO debugger command is also gone, since you can just invoke
2857 the CONTINUE restart directly instead.
2858 ** The TOP debugger command is also gone, since it's redundant with the
2859 FRAME 0 command, and since it interfered with abbreviations for the
2861 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2863 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2864 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2865 some porting work by Martin Atzmueller)
2866 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2867 with a boatload of patches which he ported to SBCL. Now that those
2869 ** The system tries to make sure that its low-priority messages
2870 are prefixed by semicolons, to help people who like to use
2871 syntax highlighting in their ILISP buffer. (This patch
2872 was originally due to Raymond Toy.)
2873 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2874 patches originally written by Raymond Toy.
2875 ** The compiler understands coercion between single-value and
2876 multiple-VALUES type expressions better, getting rid of some very
2877 weird behavior, thanks to patches originally by Robert MacLachlan
2878 and Douglas Crosher.
2879 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2880 lambda lists, thanks to a patch originally by Pierre Mai.
2881 ** The system no longer bogusly warns about "abbreviated type
2883 ** The compiler gets less confused by inlining and RETURN-FROM,
2884 thanks to some patches originally by Tim Moore.
2885 ** The system no longer hangs when dumping circular lists to fasl
2886 files, thanks to a patch originally from Douglas Crosher.
2887 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2888 undefined function error.
2889 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2890 in structure slot initforms) for now, documented workaround instead:-|
2891 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2892 * fixed bug 5: added stubs for various Gray stream functions called
2893 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2894 installed, at least appropriate type errors are generated
2895 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2896 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2897 consistently in DEFMETHOD forms.
2898 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2899 it doesn't seem to affect SBCL after all
2900 * The C runtime system now builds with better optimization and many
2901 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2903 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2905 * The system is now under CVS at SourceForge (instead of the
2906 CVS repository on my home machine).
2907 * The new signal handling code has been tweaked to treat register
2908 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2909 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2910 downstream, e.g. in debug-int.lisp, has implicit dependencies
2911 on the unsignedness of integer representation of machine words,
2912 and that caused the system to bomb out with infinite regress
2913 when trying to recover from type errors involving signed values,
2914 e.g. (BUTLAST '(1 2 3) -1).)
2915 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2916 (This was one of the bugs Peter Van Eynde reported back in July.)
2917 * The system now uses code inspired by Colin Walters' O(N)
2918 implementation of MAP (from the cmucl-imp@cons.org mailing
2919 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2920 inline the MAP operation, and there is more than one
2921 sequence argument to the MAP call (so that it can't just
2922 do ETYPECASE once and for all based on the type of the
2923 single sequence argument). (The old non-inline implementation
2924 of the general M-argument sequence-of-length-N case required
2925 O(M*N*N) time when any of the sequence arguments were LISTs.)
2926 * The QUIT :UNIX-CODE keyword argument has been renamed to
2927 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2928 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2930 * Raymond Wiker's patches to port dynamic loading from Linux to
2931 FreeBSD have been added.
2932 * The BUGS file is now more nearly up to date, thanks in large part
2933 to Martin Atzmueller's review of it.
2934 * The debugger now flushes standard output streams before it begins
2935 its output ("debugger invoked" and so forth).
2936 * The core version number and fasl file version number have both
2937 been incremented, because of incompatible changes in the layout
2939 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2940 used to not be called for a saved Lisp image.)
2941 * Martin Atzmueller's version of a patch to fix a compiler crash,
2942 as posted on sbcl-devel 13 September 2000, has been installed.
2943 * Instead of installing Martin Atzmueller's patch for the
2944 compiler transform for SUBSEQ, I deleted the compiler transform,
2945 and transforms for some similar consing operations.
2946 * A bug in signal handling which kept TRACE from working on OpenBSD
2948 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2949 away by constant folding
2950 * The system now defines its address space constants in one place
2951 (in the Lisp sources), and propagates them automatically elsewhere
2952 (through GENESIS and the sbcl.h file). Therefore, patching the
2953 address map is less unnecessarily tedious and error-prone. The
2954 Lisp names of address space constants have also been systematized.
2955 * CVS tags like dollar-Header-dollar have been removed from
2956 the sources, because they have never saved me trouble and
2957 they've been source of trouble working with patches and other
2958 diff-related operations.
2959 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2960 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2962 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2964 * The system has been ported to OpenBSD.
2965 * The system now compiles with a simple "sh make.sh" on the systems
2966 that it's supported on. I.e., now you no longer need to tweak
2967 text in the target-features.lisp-expr and symlinks in src/runtime/
2968 by hand, the make.sh takes care of it for you.
2969 * The system is no longer so grossly inefficient when compiling code
2970 involving vectors implemented as general (not simple) vectors (VECTOR T),
2971 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2972 which dares to use the various sequence functions on non-simple
2973 vectors, takes less of a performance hit.
2974 * There is now a primitive type predicate VECTOR-T-P
2975 to test for the (VECTOR T) type, so that e.g.
2976 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2977 can now be compiled with some semblance of efficiency. (The old code
2978 turned the type declaration into a full call to %TYPEP at runtime!)
2979 * AREF on (VECTOR T) is still not fast, since it's still compiled
2980 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2981 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2982 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2983 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2984 without making full calls to SUBTYPEP at runtime.
2985 (Some analogous efficiency issues for non-simple vectors specialized to
2986 element types other than T, or for non-simple multidimensional arrays,
2987 have not been addressed. They could almost certainly be handled the
2988 same way if anyone is motivated to do so.)
2989 * The changes in array handling break binary compatibility, so
2990 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2991 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2992 instead of (VALUES T T).
2993 * By following the instructions that Dan Barlow posted to sbcl-devel
2994 on 2 July 2000, I was able to enable primitive dynamic object
2995 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2996 functionality is not implemented (since it calls ld to resolve
2997 library references automatically, requiring RUN-PROGRAM for its
2998 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2999 try to resolve library references) is now supported.
3000 * The system now flushes the standard output streams when it terminates,
3001 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3002 them at several other probably-convenient times, e.g. in each pass of
3003 the toplevel read-eval-print loop, and after evaluating a form given
3004 as an "--eval" command-line option. (These changes were motivated by a
3005 discussion of stream flushing issues on cmucl-imp in August 2000.)
3006 * The source transform for TYPEP of array types no longer assumes
3007 that an array whose element type is a not-yet-defined type
3008 is implemented as an array of T, but instead punts, so that the
3009 type will be interpreted at runtime.
3010 * There is now some support for cross-compiling in make.sh: each of
3011 the phases of make.sh has its own script. (This should be transparent
3012 to people doing ordinary, non-cross-compile builds.)
3013 * Since my laptop doesn't have hundreds of megabytes of memory like
3014 my desktop machine, I became more motivated to do some items on
3015 my to-do list in order to reduce the size of the system a little:
3016 ** Arrange for various needed-only-at-cold-init things to be
3017 uninterned after cold init. To support this, those things have
3018 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3019 symbols with such names are now uninterned after cold init).
3020 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3021 fasl files for cold load.
3022 ** Remove the old compiler structure pooling code (which used to
3023 be conditional on the target feature :SB-ALLOC) completely.
3024 ** Redo the representation of some data in cold init to be more compact.
3025 (I also looked into supporting byte compiled code at bootstrap time,
3026 which would probably reduce the size of the system a lot, but that
3027 looked too complicated, so I punted for now.)
3028 * The maximum signal nesting depth in the src/runtime/ support code has
3029 been reduced from 4096 to 256. (I don't know any reason for the very
3030 large old value. If the new smaller value turns out to break something,
3031 I'll probably just bump it back up.)
3032 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3034 * Many, many bugs reported by Peter Van Eynde have been added to
3035 the BUGS list; some have even been fixed.
3036 * While enabling dynamic object file loading, I tried to make the
3037 code easier to understand, renaming various functions and variables
3038 with less ambiguous names, and changing some function calling
3039 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3040 * While trying to figure out how to do the OpenBSD port, I tried to
3041 clean up some of the code in src/runtime/. In particular, I dropped
3042 support for non-POSIX signal handling, added various comments,
3043 tweaked the code to reduce the number of compilation warnings, and
3044 renamed some files to increase consistency.
3045 * To support the new automatic configuration functionality in make.sh,
3046 the source file target-features.lisp-expr has been replaced with the
3047 source file base-target-features.lisp-expr and the machine-generated
3048 file local-target-features.lisp-expr.
3049 * fixed a stupid quoting error in make.sh so that using CMU CL
3050 "lisp -batch" as cross-compilation host works again
3052 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3054 * DESCRIBE no longer tries to call itself recursively to describe
3055 bound/fbound values, so that it no longer fails on symbols which are
3056 bound to themselves (like keywords, T, and NIL).
3057 * DESCRIBE now works on generic functions.
3058 * The printer now prints less-screwed-up representations of closures
3059 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3060 * A private symbol is used instead of the :EMPTY keyword previously
3061 used to mark empty slots in hash tables. Thus
3062 (DEFVAR *HT* (MAKE-HASH-TABLE))
3063 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3064 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3065 now does what ANSI says that it should. (You can still get
3066 similar noncompliant behavior if bang on the hash table
3067 implementation with all the symbols you get back from
3068 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3069 This breaks binary compatibility, since tests for equality to
3070 :EMPTY are wired into things like the macroexpansion of
3071 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3073 * There's now a minimal placeholder implementation for CL:STEP,
3074 as required by ANSI.
3075 * An obscure bug in the interaction of the normal compiler, the byte
3076 compiler, inlining, and structure predicates has been patched
3077 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3078 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3080 * Missing ordinary arguments in a macro call are now detected even
3081 when the macro lambda list contains &KEY or &REST.
3082 * The debugger no longer complains about encountering the top of the
3083 stack when you type "FRAME 0" to explicitly instruct it to go to
3084 the top of the stack. And it now prints the frame you request even
3085 if it's the current frame (instead of saying "You are here.").
3086 * As specified by ANSI, the system now always prints keywords
3087 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3089 * The default initial SIZE of HASH-TABLEs is now smaller.
3090 * Type information from CLOS class dispatch is now propagated
3091 into DEFMETHOD bodies, so that e.g.
3092 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3094 is now basically equivalent to
3095 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3096 (DECLARE (TYPE SINGLE-FLOAT X))
3098 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3099 operation, without having to do run-time type dispatch.
3100 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3101 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3102 * Since I don't seem to be making big file reorganizations very often
3103 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3104 is overflowing my ability to conveniently back them up), I've finally
3105 checked the system into CVS. (The CVS repository is on my home system,
3106 not at SourceForge -- putting it on SourceForge might come later.)
3107 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3108 high-level GC-NOTIFY-FOO functions send their output. (There's
3109 still very little control of where low-level verbose GC functions
3110 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3111 less than it used to -- the GC-NOTIFY-FOO functions are now under
3112 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3113 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3114 in only one place in the source code, and propagates it automatically
3115 everywhere that it's needed. Thus e.g. when I bump the version from
3116 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3117 * The C source files now include boilerplate legalese and documentation
3118 at the head of each file (just as the Lisp source files already did).
3119 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3120 to his page will be replaced with a link to his new CLiki service.
3122 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3124 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3125 * The build process now looks for GNU make under the default name "gmake",
3126 instead of "make" as it used to. If GNU make is not available as "gmake"
3127 on your system, you can change this default behavior by setting the
3128 GNUMAKE environment variable.
3129 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3130 can build without error under CMU CL.
3132 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3134 * There is now a partial SBCL user manual (with some new text and some
3135 text cribbed from the CMU CL manual).
3136 * The beginnings of a profiler have been added (starting with the
3137 CMU CL profiler and simplifying and cleaning up). Eventually the
3138 main interface should be through the TRACE macro, but for now,
3139 it's still accessed through vaguely CMU-CL-style functions and macros
3140 exported from the package SB-PROFILE.
3141 * Some problems left over from porting CMU CL to the new
3142 cross-compilation bootstrap process have been cleaned up:
3143 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3144 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3145 part when getting rid of the compiler *BACKEND* stuff.)
3146 ** The value of *NULL-TYPE* was screwed up, because it was
3147 being initialized before the type system knew the final
3148 definition of the 'NULL type. This screwed up several key
3149 optimizations in the compiler, causing inefficiency in all sorts
3150 of places. (I found it because I wanted to understand why
3151 GET-INTERNAL-RUN-TIME was consing.)
3152 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3153 PROCLAIM FTYPE information. Unfortunately this broke binary
3154 compatibility again, since now the forms output by DEFGENERIC
3155 to refer to functions which didn't exist in 0.6.3.
3156 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3157 can use the new (as of 0.6.3) transform for SOME into MAP into
3159 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3160 values to (INTEGER 0 1000000), so that the time code will no longer
3161 occasionally get blown up by Linux returning 1000000 microseconds
3162 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3163 its output conform to the ANSI spec. (Alas, this makes its output
3164 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3165 * A full call to MAP NIL with a single sequence argument no longer conses.
3166 * fixes to problems pointed out by Martin Atzmueller:
3167 * The manual page no longer talks about multiprocessing as though
3168 it were currently supported.
3169 * The ILISP support patches have been removed from the distribution,
3170 because as of version 5.10.1, ILISP now supports SBCL without us
3171 having to maintain patches.
3172 * added a modified version of Raymond Toy's recent CMU CL patch for
3173 EQUALP comparison of HASH-TABLE
3175 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3177 * The system still can't cross-compile itself with
3178 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3179 entails), but at least it can compile more of itself that way
3180 than it used to be able to, and various buglets which were uncovered
3181 by trying to cross-compile itself that way have now been fixed.
3182 * This release breaks binary compatibility again. This time
3183 at least I've incremented the FASL file format version to 2, so that the
3184 problem can be detected reliably instead of just causing weird errors.
3185 * various new style warnings:
3186 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3187 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3188 ** using the lexical binding of a variable named in the *FOO* style
3189 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3190 as specified by ANSI.
3191 * *RANDOM-STATE* is no longer automatically initialized from
3192 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3193 default behavior of the system is to repeat its behavior every time
3194 it's run. If you'd like to change this behavior, you can always
3195 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3196 old convention there was no comparably easy way to get the system to
3197 repeat its behavior every time it was run.
3198 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3199 FTYPE declarations has been removed, in favor of their ANSI
3200 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3201 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3202 the types of their sequence arguments can be determined at compile time.
3203 This is done through a new open code expansion for MAP which eliminates
3204 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3205 argument types can be determined at compile time.
3206 * The optimizer now transforms COERCE into an identity operation when it
3207 can prove that the coerced object is already of the correct type. (This
3208 can be a win for machine generated code, including the output of other
3209 optimization transforms, such as the MAP transform above.)
3210 * Credit information has been moved from source file headers into CREDITS.
3211 * Source file headers have been made more standard.
3212 * The CASE macro now compiles without complaining even when it has
3215 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3217 * (Note that the way that the PCL macroexpansions were rewritten
3218 to accommodate the change in DEFGENERIC below breaks binary
3219 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3220 not run under sbcl-0.6.2. Once we get out of alpha releases,
3221 i.e. hit release 1.0.0, we'll probably try to maintain binary
3222 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3223 and sbcl-1.4.4. Until then, however, it might be fairly common
3224 for maintenance releases to break binary compatibility.)
3225 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3227 * The reporting of unhandled signals has been changed to print some
3228 explanatory text as well as the report form. (Previously only
3229 the report form was printed.)
3230 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3231 it defines, so that the compiler no longer issues undefined function
3232 warnings for compiled-but-not-yet-loaded generic functions.
3233 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3234 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3235 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3236 section of the ANSI spec are supported.
3237 * The "" nickname for the "KEYWORD" package has been removed.
3238 The reader still handles symbol tokens which begin with a package marker
3239 as keywords, but it doesn't expose its mechanism for doing so in the
3240 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3241 * The system now issues STYLE-WARNINGs for contradictory TYPE
3242 proclamations. (Warnings for contradictory FTYPE proclamations would
3243 be nice too, but those can't be done usefully unless the type system
3244 is made smarter about FUNCTION types.)
3245 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3246 systematized, so that "*target-*.lisp is supposed to exist only on the
3247 target and imply that there's a related file which exists on the
3248 host, and *host-*.lisp is supposed to exist only on the host and imply
3249 that there's a related file which exists on the target. This involves a
3250 lot of renaming. Hopefully the acute confusion caused by the renaming
3251 will be justified by the reduction in chronic confusion..
3252 ** runtime-type.lisp -> early-target-type.lisp
3253 ** target-type.lisp -> late-target-type.lisp
3254 ** early-host-format.lisp -> early-format.lisp
3255 ** late-host-format.lisp -> late-format.lisp
3256 ** host-error.lisp -> misc-error.lisp
3257 ** early-error.lisp -> early-target-error.lisp
3258 ** late-error.lisp -> late-target-error.lisp
3259 ** host-defboot.lisp -> early-defboot.lisp
3260 ** code/misc.lisp -> code/target-misc.lisp
3261 ** code/host-misc.lisp -> code/misc.lisp
3262 ** code/numbers.lisp -> code/target-numbers.lisp
3263 ** code/early-numbers.lisp -> numbers.lisp
3264 ** early-host-type.lisp -> early-type.lisp
3265 ** late-host-type.lisp -> late-type.lisp
3266 ** host-typep.lisp -> typep.lisp
3267 ** load.lisp -> target-load.lisp
3268 ** host-load.lisp -> load.lisp
3269 ** host-disassem.lisp -> disassem.lisp
3270 ** host-insts.lisp -> insts.lisp
3271 ** byte-comp.lisp -> target-byte-comp.lisp
3272 ** host-byte-comp.lisp -> byte-comp.lisp
3273 ** host-signal.lisp -> signal.lisp
3274 ** host-defstruct.lisp -> defstruct.lisp
3275 ** late-target-type.lisp -> deftypes-for-target.lisp
3276 Furthermore, several other previously target-only files foo.lisp (e.g.
3277 hash-table.lisp and random.lisp) have been split into a target-and-host
3278 foo.lisp file and a target-only target-foo.lisp file, with their key type
3279 definitions in the target-and-host part, so that the cross-compiler will
3280 know more about target types.
3281 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3282 gone away. In their place are various *BACKEND-FOO* variables
3283 corresponding to the slots of the old structure.
3284 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3285 into the target SBCL has been fixed.
3286 * The chill.lisp system for loading cold code into a running SBCL
3288 * Support for the CMU CL "scavenger hook" extension has been removed.
3289 (It was undocumented and unused in the CMU CL sources that SBCL was
3290 derived from, and stale in sbcl-0.6.1.)
3291 * Various errors in the cross-compiler type system were detected
3292 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3293 (enabling various consistency checks). Many of them were fixed,
3294 but some hard problems remain, so the compiler is back to
3295 running without *TYPE-SYSTEM-INITIALIZED* for now.
3296 * As part of the cross-compiler type system cleanup, I implemented
3297 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3298 * I have started adding UNCROSS calls throughout the type system
3299 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3300 on cross-compiler input files will be able to go away, and various
3302 * CONSTANTP now returns true for quoted forms (as explicitly required
3305 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3307 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3308 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3309 (MAKE-LIST -1). In the longer term, it ought to become true
3310 that declarations are assertions even at SAFETY 1. For now, it's not
3311 quite true even at SAFETY 3, but it's at least more nearly true..
3312 (Note that this change seems to increases the size of the system by
3313 O(5%) and to decrease the speed of the compiler by 20% or more.)
3314 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3315 for the problem of printing dozens of lines of distracting information
3316 about low-level system machinery as part of the top stack frame
3317 on entry to the debugger when an undefined function was called.
3318 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3319 is not reset to COMMON-LISP-USER.
3320 * Compilation of stuff related to dyncount.lisp has been made conditional
3321 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3322 smaller. The various dyncount-related symbols have been moved into
3323 a new "SB-DYNCOUNT" package.
3324 * tty-inspect.lisp has been renamed to inspect.lisp.
3325 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3326 feature has gone away. (When we eventually port to other flavors of
3327 libc and/or Unix, we'll try to make the differences between flavors
3328 invisible at the user level.)
3329 * Various other *FEATURES* tags, and/or their associated conditionals,
3330 have been removed if obsolescent, or given better documentation, or
3331 sometimes given more-mnemonic names.
3333 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3335 * tidied up "make.sh" script
3336 * tidied up system directory structure
3337 * better "clean.sh" behavior
3338 * added doc/FOR-CMUCL-DEVELOPERS
3339 * many many small tweaks to output format, e.g. removing possibly-confusing
3340 trailing #\. character in DESCRIBE-INSTANCE
3341 * (EQUALP #\A 'A) no longer signals an error.
3342 * new hashing code, including EQUALP hashing
3343 * tidied up Lisp initialization and toplevel
3344 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3345 * command line argument processing
3346 * added POSIX-GETENV function to deal with Unix-ish environment variables
3347 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3348 terminating SBCL on EOF
3349 * non-verbose GC by default
3350 * There is no more "sbcl" shell script; the sbcl file is now the C
3351 runtime executable (just like CMU CL).
3352 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3354 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3355 symbol and package information as Lisp native symbol and package objects
3356 instead of strings naming symbols and strings naming packages. This way,
3357 whenever packages are renamed (as in warm init), debug information is
3358 transformed along with everything else.
3359 * tweaked the optimization policy declarations which control the building
3360 of SBCL itself. Now, among other things, the system no longer saves
3361 source location debugging information. (This helps two problems at once
3362 by reducing SBCL size and by keeping SBCL from trying to look for its
3363 sources -- which may not exist -- when reporting errors.)
3364 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3365 debugging and testing purposes
3366 * cleaned up printing, making the printer call PRINT-OBJECT for
3367 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3368 methods, giving nearly-ANSI behavior
3369 * converted almost all special variables to use *FOO* naming convention
3370 * deleted PARSE-TIME functionality, since it can be done portably
3371 * moved some files out of cold init into warm init
3372 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3374 * regularized formatting of source files
3375 * added an install.sh script
3376 * fixed ridiculous memory usage of cross-compiler by making
3377 compiler/alloc.lisp not try to do pooling unless it can hook
3378 itself into the GC of the cross-compilation host. Now the system
3379 builds nicely on my old laptop.
3380 * added :SB-ALLOC in target-features.lisp-expr
3381 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3382 was not implemented)
3383 * re-did condition handling and note reporting in the compiler. Notes
3384 are no longer handled by signalling conditions. Style warnings
3385 and warnings are handled more correctly and reported in such a way
3386 that it's easy to find one or the other in your output (so that you
3387 can e.g. figure out which of many problems caused COMPILE-FILE to
3389 * changed the severity of several compiler warnings from full WARNING
3390 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3391 compiler note reporting so that it doesn't use the condition system
3392 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3393 COMPILE-FILE command)
3394 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3395 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3396 no longer supported, since their implementation was deeply intertwingled
3397 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3398 * removed bogus "support" for compiler macros named (SETF FOO), and
3399 removed the compiler macro for SETF INFO (but only after making a fool
3400 of myself on the cmucl-imp mailing list by posting a bogus patch for
3401 DEFINE-COMPILER-MACRO..)
3402 * Compiled files containing forms which have side effects on the Lisp
3403 reader (such as DEFPACKAGE forms) are now handled more correctly.
3404 (Compiler queuing of top level lambdas has been suppressed by setting
3405 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3406 * deleted various currently-unused source files, e.g. gengc.lisp. They
3407 may be added back at some point e.g. when porting to other architectures,
3408 but until they are it's distracting to distribute them and to try to
3410 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3411 there were so many of them they're just distractions, and UNCROSS is
3412 known to be able to handle the current sources
3413 * moved PROFILE functionality into TRACE, so that it will be clear
3414 how the wrapping and unwrapping of functions when you profile them
3415 interacts with the wrapping and unwrapping of functions when you
3416 trace them. (Actually, the functionality isn't there yet, but at least
3417 the interface specification is there. Hopefully, the functionality will
3418 arrive with some future maintenance release.)
3419 * removed host-oops.lisp
3420 * changed signature of QUIT function to allow UNIX-CODE argument
3421 * fixed READ-SEQUENCE bug
3422 * tweaked verbose GC output so that it looks more like the progress
3423 output that ANSI specifies for functions like LOAD
3424 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3425 * added <http://sbcl.sourceforge.com> to the banner information printed by