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
21 ** bug fix: parent thread now can be gc'ed even with a live
23 ** bug fix: binding a special with PROGV to no value is not
24 the same as that symbol not having been bound (thanks to
26 ** bug fix: binding specials is thread safe (thanks to
28 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
29 and SB-SPROF do not die with 'no handler for signal XX in
30 interrupt_handle_now(..)' anymore
31 * fixed some bugs revealed by Paul Dietz' test suite:
32 ** ENSURE-GENERIC-FUNCTION should take a method class object for
33 the :method-class keyword argument.
35 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
36 * new port: the Solaris operating system on x86 processors is now
37 mostly supported, though some rough edges in the environment
38 remain. (thanks to Daisuke Homma)
39 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
40 space for its heap. (thanks to Thiemo Seufer)
41 * optimization: SBCL on MIPS platforms now supports dynamic-extent
42 closures. (thanks to Thiemo Seufer)
43 * minor incompatible change: eof selects abort in the debugger.
44 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
45 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
47 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
48 deprecated, and will go away in a future revision of SBCL.
49 * minor incompatible change: GC-ON and GC-OFF are no longer
50 implemented with a counter, it does not matter how many times gc
52 * bug fix: discriminating functions for generic function classes
53 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
54 make invalid assumptions about method precedence order. (reported
56 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
57 platforms. (thanks to Thiemo Seufer)
58 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
59 * bug fix: correct pathname printing with printer escaping is on.
60 (thanks to Kevin Reid)
61 * bug fix: complex VOP definitions in "user-space" no longer trigger
62 package locks. (reported by Zach Beane)
63 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
64 for funcallable-instance functions, and is no different from
66 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
67 characters in the current readtable. (reported by Nicholas Neuss)
68 * bug fix: SUBTYPEP on various CONS types returns more a more
69 accurate acknowledgment of its certainty.
70 * optimizations: REMOVE-DUPLICATES now runs in linear time on
71 lists in some cases. This partially fixes bug 384.
72 * flush all standard streams before prompting in the REPL and the
74 * bug fix: signal handling and triggering gc do not conflict
75 directly anymore, in particular a high frequency sb-sprof does
76 not prevent gc from running
77 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
78 approximation for timezone and DST information between the
79 universal time epoch and the smallest negative 32-bit time_t.
80 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
81 year 1899 so long as the encoded time is non-negative.
82 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
83 the correct answer for non-constant rotate amounts when performing
84 an inline 32-bit rotation.
86 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
87 there is only one thread in the session
88 ** bug fix: memory leak for streams created in one thread and
90 ** bug fix: lockup when compiled with gcc4
91 ** bug fix: race that allows the gc to be triggered when gc is
93 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
94 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
96 ** bug fix: finished threads are gc'ed properly
97 * fixed some bugs revealed by Paul Dietz' test suite:
98 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
99 has arguments with a different set of applicable methods from
100 the orignal arguments.
101 ** The type error thrown by MAP now has a correct expected-type
103 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
104 name a compiled function.
105 ** Three MISC tests where a large bignum was improperly coerced to
106 a float that couldn't represent that bignum during type
107 derivation were fixed.
108 ** SUBTYPEP can now handle types involving the negation of a
109 list-form FUNCTION type.
110 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
111 as (COMPLEX (AND RATIO (NOT FIXNUM))).
112 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
114 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
115 * New feature: Experimental support for bivalent streams: streams
116 opened with :element-type :default now allow character and binary
117 (unsigned-byte 8) I/O
118 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
119 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
121 * bug fix: no more highly sporadic "couldn't check whether ~S is
122 readable" when reading a stream and an interrupt hits in the middle
123 of a select system call
124 * compiler better recognizes complex arrays (reported by Eduardo
126 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
127 for variables on SBCL built with linkage-tables. (reported by Luis
129 * various error reporting improvements.
130 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
131 (Thanks to Thiemo Seufer)
132 * bug fix: nested reader invokations work correctly
133 * bug fix: it is possible to have simultaneous references to foreign
134 code and foreign data with the same name.
136 ** added x86-64 support
137 ** incompatible change: the threading api now works with thread
138 objects instead of thread ids
139 ** bug fix: threads are protected from signals and interruption when
140 starting up or going down
141 ** bug fix: a race where an exiting thread could lose its stack to gc
142 ** bug fix: don't halt on infinite error in threads if possible
143 ** fixed numerous gc deadlocks introduced in the pthread merge
144 ** bug fix: fixed thread safety issues in read and print
145 ** bug fix: debugger doesn't hang on session lock if interrupted at
146 an inappropriate moment
147 ** bug fix: run-program is now thread safe(r)
148 ** bug fix: inner with-recursive-lock no longer releases the mutex
149 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
150 * fixed some bugs revealed by Paul Dietz' test suite:
151 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
152 EXPECTED-TYPE slots filled.
153 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
155 ** MAKE-CONDITION accepts classes as type-designators.
156 ** COMPILE may never return NIL.
157 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
158 range before calling Unix time functions
160 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
161 * numerous signal handling fixes to increase stability
162 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
163 * minor incompatible change: we now correctly canonize default
164 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
165 as per the MOP, rather than the historical (INITARG INITFUNCTION
166 INITFORM). (reported by Bruno Haible)
167 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
168 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
169 for more information.
170 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
171 pathname is a directory pathname.
172 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
173 * dynamic space size on PPC has been increased to 768Mb. (thanks to
175 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
176 :METACLASS argument in addition to a class name. (reported by
177 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
178 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
179 (reported by Bruno Haible; patch from Gerd Moellmann)
180 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
182 * bug fix: more cleanups to the floating point exception handling on
183 x86-64 (thanks to James Knight)
184 * bug fix: the compiler does not try to constant fold calls of
185 COERCE and BIT-* functions when they return freshly constructed
186 objects. (reported by Nikodemus Siivola and Paul Dietz)
187 * optimization: improved the MIPS versions of generic (in the
188 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
189 * optimization: direct conversion of (unsigned-byte 32) to floats on
190 the PowerPC platform.
191 * optimization: structure instances with raw slots now use less
192 memory, and probably show better memory locality. (thanks to
194 * optimization: DYNAMIC-EXTENT declarations for lists and closures
195 are treated as requests for stack allocation on the x86-64,
196 Alpha-32, PPC and SPARC platforms.
197 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
198 the wrong file descriptor; implementation of SOCKET-OPEN-P.
199 (thanks to Tony Martinez)
201 ** gcing a dead thread can no longer lead to lockups
202 ** threads block signals until they are set up properly
203 ** errno is no longer shared by threads
204 ** interrupt-thread restores the eflags register on x86
205 ** fixed some lockups due to gc/thread interaction
206 * fixed some bugs revealed by Paul Dietz' test suite:
207 ** invalid dotted lists no longer raise a read error when
209 ** signal an error if a symbol that names a declaration is used
210 as the name of a type, or vice versa
211 ** allow using the (declare (typespec var*)) abbreviation for
212 (declare (type typespec var*)) with all type specifiers
213 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
214 ** always return NIL from PPRINT-POP when OBJECT is NIL
215 ** don't signal errors when pretty-printing malformed LABELS,
216 FLET or MACROLET forms
217 ** declarations in a DOLIST body are also in scope for the
219 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
221 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
224 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
225 * fixed cross-compiler leakages that prevented building a 32-bit
226 target with a 64-bit host compiler.
227 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
228 opened with :IF-EXISTS :OVERWRITE.
229 * fixed bug 281: error for an invalid qualifier in a short-form method
230 combination method is not signalled until the faulty method is called.
231 * bug fix: iteration variable type inferrer failed to deal with open
232 intervals. (reported by Alan Shields)
233 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
234 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
235 or more runtime options were provided to the sbcl binary.
236 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
237 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
238 * optimization: there's now a fast-path for fixnum arguments in the
239 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
240 * optimization: the code generated on x86-64 is more compact thanks
241 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
242 * bug fix: floating-point exception handling now partly works on
243 x86-64. (Thanks to James Knight)
244 * improvement to the MIPS backend from Thiemo Seufer: C-style
245 64-bit long long arguments and return values to alien functions
247 * fixed some bugs revealed by Paul Dietz' test suite:
248 ** the type-error signalled from WARN has a filled-in DATUM slot.
249 ** the type-error required when a stream is not associated with
250 a file has the stream as its datum.
251 ** type-errors on single-floats on x86-64 no longer have
252 :INVALID-OBJECT as the datum
253 ** the type-errors signalled for invalid function names now have
254 a correct expected type
255 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
256 for structures defined with a :TYPE. Documentation strings for
257 typed structures are no longer immediately discarded
258 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
259 broadcast streams and synonym streams. FILE-LENGTH now also works
260 on broadcast streams.
262 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
263 * incompatible change: the --noprogrammer option, deprecated since
264 version 0.7.5, has been removed. Please use the equivalent
265 --disable-debugger option instead.
266 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
268 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
269 inoperational for a while now) has been completely removed.
270 * null lexical environments are now printed as #<NULL-LEXENV>,
271 significantly reducing the amount of clutter in typical backtraces.
272 * documentation on weak pointers, finalization, and after GC hooks
273 has been added to the manual.
274 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
275 as other specialized array types, is much faster. SUBSEQ and
276 COPY-SEQ on such arrays have also been sped up.
277 * optimization: EQL is now more efficient when at least other argument
278 is known to be of type (OR FIXNUM (NOT NUMBER)).
279 * fixed bug: the runtime is now less vulnerable to changes in the
280 size of the SBCL object on OS X, and virtual memory is reserved for
281 all spaces that need to be at a fixed address.
282 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
283 * fixed bug: finalizers and after GC hooks that cause consing are now
285 * fixed bug: compiler error messages and summaries are now printed to
286 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
287 * fixed inference of the upper bound of an iteration variable.
288 (reported by Rajat Datta).
289 * fixed bug 211e: calling local functions with duplicated constant
290 keyword argument no longer causes a bogus style warning about an
292 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
293 ftype declaration to be disregarded. (reported by Dave Roberts)
294 * fixed bug 373: caused by erronous compilation of references to alien
295 variables in the runtime on ppc/darwin.
296 * fixed bug 376: CONJUGATE type deriver.
297 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
298 (reported by Baughn on #lisp)
299 * a cross-compiler bug on non-x86 platforms has been identified and
300 fixed. (thanks to Bruno Haible)
301 * improvements to the MIPS runtime code for increased stability.
302 (thanks to Thiemo Seufer)
303 * increased the maximimum compact environment size to allow
304 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
305 * partial workaround for bug 135: don't name the function we're
306 calling for hairy cases of EVAL, so as not to accumulate one environment
307 entry per call to EVAL. (reported by Kevin Reid)
308 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
309 * optimization: type testing for non-vector arrays should be faster.
310 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
312 * bug fix: setting 31st element of a bit vector to zero did not work
314 * bug fix: redefining a class definition which failed due to a
315 previous accessor / function clash now works (but see BUGS entry
316 #380 for more problems in this area). (thanks to Zach Beane)
317 * the long form of DEFINE-METHOD-COMBINATION disables method group
318 checking when given a single method group with pattern *. (thanks
320 * on x86 compiler supports stack allocation of results of simple
321 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
322 * support for building and running on Mac OS X 10.4 "Tiger" has been added
323 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
324 * fixed some bugs related to Unicode integration:
325 ** the restarts for recovering from input and output encoding
326 errors only appear when there is in fact such an error to
328 * fixed some bugs revealed by Paul Dietz' test suite:
329 ** MISC.549 and similar: late transformation of unsafe type
330 assertions into derived types caused unexpected code
332 ** SCALE-FLOAT type deriver is less wrong.
333 ** type derivers for EXP, LOG and similar functions compute result
334 types for complex arguments better.
335 ** (MISC.563) CONJUGATE type deriver works for very restricted
337 ** out-of-line type testers for character strings are available.
338 ** EQUAL compiler transform understands specialness of objects
340 ** accessing double-floats stored on the stack now works on x86-64.
341 ** debugger internals could sometimes create invalid lispobjs,
342 resulting in GC crashes.
343 ** MISC.548: type check weakening can convert required type into
345 ** initialization forms for bindings are not in scope of free special
348 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
349 * incompatible change: thread support for non-NPTL systems has
350 been removed - locking is buggy and unreliable. A threaded
351 SBCL build will now warn at startup and refuse to create
352 new threads, unless futex support is detected in the kernel
353 * incompatible change: the top level REPL now has only an ABORT
354 restart associated with it, not TOPLEVEL and ABORT as it used to.
355 TOP and TOPLEVEL are now available as debugger commands for
356 returning to the top level.
357 * incompatible change: forms evaluated in the REPL now use the
358 global optimization policy.
359 * incompatible change: user- and system-initialization files are
360 no longer processed with LOAD, but by READ and EVAL; hence the
361 global optimization policy, startup package, readtable, etc,
363 * The .fasl file format number has been incremented because of
364 various incompatible changes.
365 * internal entry point details and argument counts no longer appear
366 in backtraces unless explicitly requested by setting
367 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
368 * built-in and standard functions no longer have names like "top
369 level local call to FOO".
370 * fixed bug 32: functions defined in non-null lexical environments
371 now have more legible printed representation
372 * fixed bug 33: functions defined in non-null lexical environemnts
373 are now more amenable to inspection by INSPECT.
374 * workaround for bug 354: XEPs no longer appear in backtraces unless
375 explicitly requested.
376 * fixed bug: receiving the signal which results from attempting to
377 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
378 SIGBUS on other *BSDs) on architectures where the C stack is also
379 the Lisp stack (x86 and x86-64) from foreign code no longer leads
380 to debugger confusion or wild instability. (reported by Cheuksan
382 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
383 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
384 (reported by Lutz Euler)
385 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
386 compiled code. (reported by Paul Dietz)
387 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
388 specialized parameter to a method to something that is not TYPEP
389 the specializer is now possible.
390 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
391 face of package deletion.
392 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
393 pathname. (reported by David Tolpin on c.l.l)
394 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
395 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
396 STRING-STREAM and FILE-STREAM, respectively.
397 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
398 than previously; in addition, its test suite is now run on build.
399 (thanks to Luke Gorrie)
400 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
401 * optimization: added a immediate representation for single-floats
403 * fixed some bugs revealed by Paul Dietz' test suite:
404 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
405 ** attempting to create a package with a colliding nickname causes
406 correctable errors to be signalled.
407 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
408 ** several x86-64 backend bugs related to sign-extension of immediate
411 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
412 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
413 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
415 * fixed loading of multiply forward-referenced layouts.
416 (thanks to Cheuksan Wang)
417 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
418 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
419 (see COMPILE-FILE documentation for details of :PRINT option)
420 * fixed bugs 19 and 317: fixed-format floating point printing is
421 more accurate. This also fixes a bug reported by Adam Warner
422 related to the ~@F format directive.
423 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
425 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
426 dispatch table if the from-char is a dispatch macro character.
427 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
428 by the wordsize no longer ignore the last word. (reported by Lutz
430 * fixed bug in type checking of dynamic-extent variables. (reported
432 * optimization: sequence traversal functions use their freedom to
433 coerce function designators to functions.
434 * optimization: code with many calls to CLOS methods specialized on
435 CLOS classes has had redundant type checks removed.
436 * optimization: added declarations to speed up operations that access
437 the internal character database (for example STRING-UPCASE)
438 * optimization: comparison operations between floats and sufficiently small
439 fixnums no longer create extra rationals
440 * fixed some bugs related to Unicode integration:
441 ** portions of multibyte characters at the end of buffers for
442 character-based file input are correctly transferred to the
443 start of the buffer at the next read.
444 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
445 passing it through to OPEN.
446 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
447 argument given to internal calls to OPEN.
448 ** the built-in comment readers (introduced by character sequences
449 ";" and "#|") are more forgiving to encoding errors; they will
450 STYLE-WARN and then attempt to resync the stream at a character
451 boundary. (thanks to Teemu Kalvas)
452 * fixed some bugs in the x86-64 port:
453 ** Negative short int return values from c-calls are sign-extended
455 ** The stack is aligned to 16-bytes for c-calls, as required by
456 the ABI. (thanks to Cheuksan Wang)
457 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
458 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
459 for enums. (thanks to Vincent Arkesteijn)
460 ** Multiple small optimizations and bugfixes for floating point
462 * fixed some bugs revealed by Paul Dietz' test suite:
463 ** Space, Tab, Linefeed, Return and Page have the invalid
464 secondary constituent character trait.
465 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
467 ** WITH-INPUT-FROM-STRING should only update the index place on
469 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
471 ** Bit-array operations (BIT-AND and similar) worked incorrectly
472 with one-dimensional arrays with fill pointers.
473 ** TYPE-OF failed on a complex with an integer realpart and a
475 ** compiler failure during type inference for the code of form
476 (IF (EQL X (THE ...)) ...) (MISC.535).
478 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
479 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
480 platform. The port supports 61-bit fixnums, large memory spaces
481 and reloading shared object files.
482 * enhancement: saving cores with foreign code loaded is now
483 supported on ppc/Darwin in addition to the previously supported
485 * enhancement: the statistical profiler now walks deeper into the
486 call stack for more meaningful call-graphs and accrued time
487 reports (x86/x86-64 only). It also now reports time spent in
489 * enhancement: it is now possible to trace most individual methods
490 of a generic function in addition to tracing the generic function
492 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
494 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
495 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
496 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
497 produces an error. (thanks to Vincent Arkesteijn)
498 * bug fix: NAMESTRING on pathnames with :WILD components in their
499 directories works correctly. (thanks to Artem V. Andreev)
500 * fixed bug 125: compiler preserves identity of closures. (reported
502 * bug fixed: functions with &REST arguments sometimes failed with
503 "Undefined function" when compiled with (DEBUG 3). (reported by
505 * bug fix: overflow during compiling of setting element of a bit
506 vector with constant index and value. (reported by Timmy Douglas)
507 * build fix: fixed the dependence on *LOAD-PATHNAME* and
508 *COMPILE-FILE-PATHNAME* being absolute pathnames.
509 * on x86 compiler partially supports stack allocation of dynamic-extent
511 * GO and RETURN-FROM do not check the extent of their exit points
512 when compiled with SAFETY 0.
513 * fixed some bugs related to Unicode integration:
514 ** encoding and decoding errors are now much more robustly
515 handled; it should now be possible to recover even from invalid
516 input or output to the terminal. (thanks to Teemu Kalvas)
517 ** provided a first cut at implementing STRING-TO-OCTETS and
518 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
519 ** altered the SB-MD5 contributed module slightly, changing the
520 interface just enough for it to be supportable for builds where
521 lisp characters are not eight bits.
522 * fixed some bugs revealed by Paul Dietz' test suite:
523 ** the FORMATTER-generated functions for ~V[ conditionals require
524 the correct number of arguments.
525 ** READ-FROM-STRING returns the mandated second value when applied
526 to displaced strings.
527 ** the #\Rubout and #\Backspace characters are treated as invalid
528 constituent characters by the tokenizer.
530 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
531 * new feature: reloading changed shared object files with
532 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
533 * new feature: references to foreign variables and functions
534 can now be compiled and loaded before the corresponding shared
535 object file is loaded, as long as the foreign definitions are
536 available at runtime.
537 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
538 supported. (thanks to Dan Debertin)
539 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
540 just keywords. (thanks to Vincent Arkesteijn)
541 * fixed bug #331: structure-class instances corresponding to
542 DEFSTRUCT forms are now created eagerly.
543 * fixed bug #345: backtraces from calls to undefined functions work
544 on x86 as well. Related bug #61 is now also partially fixed on x86
545 (backtraces from throws to unknown catch tags.)
546 * bug fix: lambda-list parsing is now stricter vrt. order and number
547 of lambda-list keywords.
548 * bug fix: as specified by AMOP, an error is signalled if a
549 class-option appears multiple times in a DEFCLASS form. (reported
551 * bug fix: RANDOM can be compiled when the compiler derives the type
552 of its numeric argument as a disjoint set of small integers.
553 (reported by Paul Dietz)
554 * bug fix: starting a core saved with shared objects loaded when
555 those objects are not available no longer causes threaded SBCL to
556 hang. (reported by Sean Ross)
557 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
559 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
560 argument, which must always be NIL. (reported by Kalle Niemitalo)
561 * bug fix: printing 1.0d+23 no longer results in an error.
562 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
563 * bug fix: structure slot setters preserve evaluation order. (thanks
565 * fixed some bugs related to Unicode integration:
566 ** RUN-PROGRAM can allow its child to take input from a Lisp
567 stream. (reported by Stefan Scholl)
568 ** ASDF-INSTALL successfully downloads .tar.gz files in all
569 locales. (reported by Ken Causey)
570 * fixed some bugs revealed by Paul Dietz' test suite:
571 ** INCF, DECF and REMF evaluate their place form as specified in
573 ** FORMATTER expands ~{ ~} iteration directives with V or #
574 parameters correctly.
575 ** FORMATTER deals with the ~@[ ~] conditional directive where the
576 consequent uses no arguments correctly.
577 ** the system has a partial understanding of the (COMPLEX RATIO)
579 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
582 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
583 * new feature: a build-time option (controlled by the :SB-UNICODE
584 keyword feature, enabled by default) for building the system with
585 support for the entire 21-bit character space defined by the
587 * new feature: the system now has rudimentary external-format
588 support; the primary user-visible change at this time is that
589 characters with the high bit set (such as lower-case-e-acute) will
590 print correctly to a terminal in a UTF-8 environment.
591 * minor incompatible change: BASE-CHAR no longer names a class;
592 however, CHARACTER continues to do so, as required by ANSI.
593 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
594 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
595 has been moved to the SB-EXT package (temporarily re-exported from
597 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
598 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
600 * fixed bug #351: better error-handlind and reporting for malformed
602 * fixed bug #350: bignum-printing is now more memory-efficient,
603 allowing printing of very large bignums, eg. (expt 2 10000000).
604 (reported by Bruno Haible)
605 * fixed bug #302: better primitive-type selection for intersection
607 * fixed bug #308: non-graphic characters now all have names, as
608 required. (reported by Bruno Haible)
609 * bug fix: redefining a class with different superclasses now correctly
610 removes it from the direct-subclasses of its previous superclasses.
611 (reported by David Morse)
612 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
613 new value now works. (reported by Bruno Haible)
614 * bug fix: correct canonicalization of multiple non-standard slot
615 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
616 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
617 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
618 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
619 now exists, an signals an error.
620 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
621 during macroexpansion for non-iteration variables as well. (reported
622 by Bruno Haible for CMUCL)
623 * bug fix: Cyclic structures and unprintable objects in compiler
624 messages no longer cause errors. (reported by Bruno Haible)
625 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
626 and READ-FROM-STRING all now return a primary value of NIL if
627 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
628 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
629 (reported by Bruno Haible for CMUCL)
630 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
631 do not designate a condition. (reported by Bruno Haible for
633 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
634 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
635 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
636 permitting resolution in favour of any of the conflicting symbols.
637 (reported by Bruno Haible for CMUCL)
638 * bug fix: EQUAL compiler optimizations is less aggressive on
639 strings which can potentially compare true despite having distinct
640 specialized array element types.
641 * bug fix: unit enumerations can be defined without dividing by
642 zero. (thanks to Vincent Arkesteijn)
643 * FORMAT compile-time argument count checking has been enhanced.
644 (report from Bruno Haible for CMUCL)
645 * a partial workaround for the bug 262: the compiler does not try to
646 inline-expand a local function doing RETURN-FROM from a deleted
647 BLOCK. (thanks to Peter Denno for the bug report and to David
648 Wragg for the simple test case)
649 * fixed some bugs revealed by Paul Dietz' test suite:
650 ** RENAME-PACKAGE allows all package designators as new package
652 ** constraint propagation and lambda variable substitution are
653 more cautious in dealing with partially deleted code.
654 ** compiler sometimes lost reoptimization passes.
655 ** CERROR, given a condition as condition designator, passes its
656 remaining arguments to the continue format control without
658 ** Case-altering FORMAT directives work correctly on non-ASCII
660 ** The REAL type specifier handles bounds outside the floating
661 point ranges without signalling FLOATING-POINT-OVERFLOW.
662 ** Functions with IR1-transformations can create intercomponent
663 references to global functions.
664 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
666 ** FORMAT ~R treats a nil value for its first parameter correctly.
668 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
669 * enhancement: saving cores with foreign code loaded is now
670 supported on x86/NetBSD and sparc/Linux in addition to the previously
672 * bug fix: on some platforms repeated installations caused multiple
673 copies of HTML documentation to be installed -- should not happen
674 any more. (reported by Stefan Scholl)
675 * bug fix: parsing self-recursive alien record types multiple times
676 no longer causes infinite recursion. (reported by Thomas F. Burdick,
677 original patch by Helmut Eller for CMUCL)
678 * bug fix: stack-exhaustion detection works now on NetBSD as well.
679 (thanks to Richard Kreuter)
680 * bug fix: defining classes whose accessors are methods on existing
681 generic functions in other (locked) packages no longer signals
682 bogus package lock violations. (reported by François-René Rideau)
683 * bug fix: special variables as DEFMETHOD parameters no longer have
684 associated bogus type declarations. (reported by David Wragg and
686 * bug fix: read-write consistency on streams of element-type
687 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
688 * bug fix: redefiniton of the only method of a generic function with
689 no DEFGENERIC no longer emits a full WARNING. In addition,
690 redefinition of generic functions with no DEFGENERIC to an
691 incompatible lambda list now signals an error. (thanks to Zach
693 * bug fix: DEFGENERIC now works even when there's a function of the
694 same name in an enclosing lexical environment. (thanks to Zach
696 * fixed compiler failure, caused by instrumenting code during
697 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
698 * optimization: added loop analysis and improved register allocation
699 * fixed some bugs revealed by Paul Dietz' test suite:
700 ** POSITION on displaced vectors with non-zero displacement
701 returns the right answer.
702 ** (SIMPLE-STRING) is a valid type specifier for sequence
704 ** *PRINT-LEVEL* handling for slotless structures is pedantically
706 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
707 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
709 ** The justification version of the FORMAT ~< directive treats
710 non-zero minpad parameter correctly.
712 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
713 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
714 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
715 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
716 the supported interface.
717 * new feature: Single-stepping of code compiled with DEBUG 2 or
718 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
719 * new feature: saving cores with foreign code loaded is now
720 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
721 Timothy Moore's work for CMUCL)
722 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
723 parameters to * instead of NIL if no initform is supplied.
724 (reported by Johan Bockgaard)
725 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
726 correctly when FUNCALL appears as the car of the form. Note:
727 despite this FUNCALL forms are not currently subject to
728 compiler-macro expansion. (port of Raymond Toy's fix for the
729 same from CMUCL, reported by Johan Bockgaard)
730 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
731 (thanks for Teemu Kalvas)
732 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
733 case instead of terminating the iteration (thanks for Julian
734 Squires, Sean Champ and Raymond Toy)
735 * bug fix: incorrect expansion of defgeneric that caused a style
736 warning. (thanks for Zach Beane)
737 * bug fix: slot accessor effective method computation works properly
738 for classes with multiple non-standard applicable methods on
739 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
740 * on x86 compiler supports stack allocation of results of LIST and
741 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
742 CMUCL implementation by Gerd Moellmann)
743 * fixed some bugs revealed by Paul Dietz' test suite:
744 ** FORMAT strings with both the ~<~:;~> form of the justification
745 directive and pretty-printing directives cause an error.
747 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
748 * incompatible change: the internal functions
749 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
750 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
751 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
752 instead of the old functions.
753 * new feature: on platforms where "dladdr" is available foreign
754 function names now appear in backtraces. (based on Helmut Eller's
756 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
757 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
759 * documentation: ASDF manual is now distributed with SBCL.
760 * bug fix: SBCL can now load its contributed modules with REQUIRE
761 even if the system-provided entries have been removed from
762 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
764 * bug fix: backtraces involving undefined functions or assembly
765 routines are more informative. (thanks to Brian Downing)
766 * bug fix: mutually referent alien structures now work correctly.
767 (reported by Rick Taube)
768 * bug fix: structures defined by WITH-ALIEN can be referred to
769 within other definitions in the same WITH-ALIEN.
770 * bug fix: division operators (MOD, TRUNCATE and the like) with
771 constant zero divisors and integer dividends no longer generate
773 * bug fix: provide default methods for INPUT-STREAM-P and
774 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
775 * bug fix: improve the detection and resolution of MOP metacycles
776 (where effective method computation depends on the generic function
777 having its effective method computed). (reported by Bruno Haible)
778 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
779 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
780 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
781 * bug fix: pathnames with relative directory components can be
782 represented relative to default pathnames.
783 * optimization: in taking the GCD of bignums, reduce the two bignums
784 to approximately the same size (using Euclid's algorithm) before
785 applying the more sophisticated binary GCD. (thanks to Juho
787 * optimization: COUNT on bitvectors now operates word-at-a-time.
788 * optimization: ASH with a positive, but not necessarily constant,
789 (leftwards) shift, when in a modular context, is compiled to a
791 * fixed some bugs revealed by Paul Dietz' test suite:
792 ** FORMAT variable parameters ("~V<char>") are defaulted properly
793 if the corresponding argument is NIL.
794 ** FORMAT directives accepting a minpad parameter treat negative
796 ** literal commas as character parameters to FORMAT directives are
798 ** literal spaces directly after ~<Newline> directives within a
799 format-logical-block (~:< ~@:>) do not induce :FILL-style
800 conditional newlines.
801 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
802 as specified: it no longer includes conditional newlines).
803 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
805 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
806 :PER-LINE-PREFIX argument does not evaluate to a string.
808 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
809 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
810 the manual for details; for now, package locks can be disabled by
811 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
812 if no major problems are found then it is likely that they will be
813 compiled in unconditionally.
814 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
815 unsupported operators on all platforms. To load a shared library
816 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
817 object file, link it into a shared library outside of SBCL and
818 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
820 * bug workaround: ROOM T and the GC were not working together
821 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
822 reporting facility aren't preserved. That reporting has been
823 disabled, so now until and unless someone figures out how to make it
824 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
825 * minor incompatible change: as threatened around sbcl-0.8.0, the
826 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
827 The symbols are also exported from SB-PCL for backwards
828 compatibility, but more so than before SB-PCL should be treated as
829 an implementation-internal package.
830 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
832 * the SB-SPROF contrib now works on (most) non-x86 architectures.
833 It is known as of this release not to work on the Alpha, however.
834 * fixed bug #167: errors signalled due to illegal syntax in method
835 bodies are now more legible.
836 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
837 designators and can hence be used with TYPEP.
838 * fixed bug #333: CHECK-TYPE now ensures that the type error
839 signalled, if any, has the right object to be accessed by
840 TYPE-ERROR-DATUM. (reported by Tony Martinez)
841 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
842 5.1.2.3 for multiple-value place subforms. (reported by Kalle
844 * fixed bug #334: programmatic addition of slots using specialized
845 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
846 and :ALLOCATION :CLASS slots.
847 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
848 float exponents. (rereported by Peter Seibel)
849 * fixed a bug: #\Space (and other whitespace characters) are no
850 longer considered to be macro characters in standard syntax by
852 * fixed bug: initialization of condition class metaobjects no longer
853 causes an instance of the condition to be created. (reported by
855 * fixed bug: it is now possible to have slots such that
856 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
857 description is neither :INSTANCE nor :CLASS.
858 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
859 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
860 SLOT-VALUE-USING-CLASS), and no longer causes errors with
861 non-standard SLOT-DEFINITION-ALLOCATION values.
862 * fixed bugs: various race conditions handling exiting threads.
863 CL-PPCRE's thread torture test now passes.
864 * fixed arguably-a-bug: GC time proportional to number of threads in
865 system even when most of them are idle
866 * optimization: improved performance of BIT and SBIT on bit-vectors.
867 * .fasl file incompatibility: The fasl file version number has
868 been incremented because of changes associated with package locks.
870 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
871 * minor incompatible change: the system no longer provides
872 optimization hints (conditions which are TYPEP
873 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
874 mode; these hints are still emitted when higher SPEED optimization
876 * new contrib module: a sampling profiler (profiling by statistical
877 sampling, rather than by instrumenting functions) is available as
878 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
879 * the behaviour of the standard function ED is now customizeable by
880 third parties through a hook variable: see ED's documentation
881 string for information on the protocol.
882 * the compiler no longer emits efficiency notes for (FUNCALL X)
883 when the type of X is uncertain under default optimization
885 * fixed bug 276: mutating a binding of a specialized parameter to a
886 method to something that is not TYPEP the specializer is now
888 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
889 exhibit the required behaviour on the x86 platform. (thanks to
890 Peter van Eynde, Eric Marsden and Bruno Haible)
891 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
892 even for difficult arguments. (reported by Peter Graves)
893 * fixed bug 141a: the backquote printer now descends quoted
895 * fixed another bug in backquote printing: no more destructive
896 modification of the form's list structure. (reported by Brian
898 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
899 move between its address being taken and the call to
900 interrupt_thread, fixing a crashing race condition.
901 * the SB-POSIX contrib implementation has been adjusted so that it
902 no longer exhibits ridiculously poor performance when constructing
903 instances corresponding to C structs.
905 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
906 * minor incompatible change: the sb-grovel contrib now treats C
907 structures as alien (in the sense of SB-ALIEN) objects rather than
908 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
909 has implications for memory management of client code
910 (sb-grovel-returned objects must now be manually managed) and for
911 type safety (alien objects now have full types).
912 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
913 used to control emission of compiler diagnostics, rather than the
914 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
915 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
916 quality should be considered deprecated.
917 * install.sh now installs the user manual as well
918 * (not quite a new documentable feature, but worth considering in
919 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
920 beginnings of a semantically meaningful condition hierarchy is
921 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
923 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
924 designator as the defaults argument.
925 * fixed bug: Displaced arrays whose displaced-to array has become
926 too small now cause ARRAY-DIMENSION to signal an error, providing
927 for safer bounds-checking. (reported by Bruno Haible)
928 * fixed bug: DEFCLASS slot definitions with identical :READER and
929 :WRITER names now signal a reasonable error. (reported by Thomas
931 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
933 * fixed bug: Invalid slot specification errors now print correctly.
934 (thanks to Zach Beane)
935 * fixed bug 320: Shared to local slot value transfers in class
936 redefinitions now happen corrently from superclasses as
937 well. (reported by Bruno Haible)
938 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
940 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
941 improper lists correctly. (reported by Bruno Haible)
942 * fixed bug 313: source-transform for <fun-name> was erroneously
943 applied to a call of a value of a variable with name <fun-name>.
944 (reported by Antonio Menezes Leitao)
945 * fixed bug 307: The obsolete instance protocol ensures that
946 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
947 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
948 remove dead unknown-values globs from the middle of the stack.
949 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
950 easier to resume long computations after using *BREAK-ON-SIGNALS*
951 to diagnose and fix failures. (thanks to Nikodemus Siivola)
952 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
953 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
954 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
955 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
956 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
958 * fixed a (fairly theoretical) bug in string printing: if
959 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
960 does not have array-element-type equal to the most general string
962 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
963 function name. (thanks to Nikodemus Siivola)
964 * fixed bug: compile-time format string checker failed on
965 non-closed ~{. (reported by Thomas F Burdick)
966 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
967 don't warn on legal loop code involving more than one aggregate
968 boolean. (thanks to Nikodemus Siivola)
969 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
970 in keyword position to MAKE-PACKAGE was not being reported as
972 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
973 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
975 * fixed bugs 280 and 312: the checking for multiple definitions in a
976 file is less likely to become confused by uses of inline
978 * fixed bug: the #S reader macro performs the keyword coercion
979 specified for slot names. (reported by Kalle Niemitalo)
980 * fixed bug: lambda lists may contain symbols whose names start with
981 & but are not lambda-list-keywords; their occurrence triggers a
983 * fixed bug 321: define-method-combination argument lambda lists do
984 not cause invalid code to be generated when &optional or &aux
985 variables have default values. (reported by Bruno Haible)
986 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
987 have CLOS classes; this appears also to have fixed reported
988 instability in stack exhaustion detection.
989 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
990 MUFFLE-WARNING when no associated restart is present is now
992 * optimization: rearranged the expansion of various defining macros
993 so that each expands into only one top-level form in a
994 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
996 * optimization: used a previously little-used slot in symbols to
997 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
999 * fixed some bugs revealed by Paul Dietz' test suite:
1000 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1002 ** Fixed an optimization bug related to inheritance of initforms
1003 from local to shared slots.
1004 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1005 ** CAST optimizer forgot to flush argument derived type.
1006 ** print/read consistency on floats is now orders of magnitude
1007 more likely. (thanks also to Bruno Haible for a similar report
1009 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1010 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1011 for objects of type REAL. Make it so.
1012 ** SXHASH is specified (infelicitously) to respect similarity,
1013 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1014 it so. (thanks to Markus Ziegler)
1015 ** on the Alpha, the compiler succeeds in compiling functions
1016 returning a known number of arguments greater than 63.
1017 ** fixed handling of invalid NIL arguments in keyword position
1019 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1021 ** provided a readably-printable representation for RANDOM-STATE
1023 ** ensured that pathnames, where they have a namestring, always
1024 print using #P"..." syntax.
1026 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1027 * Support for the forthcoming 2.0 version of the NetBSD kernel
1028 running on x86 hardware has been added. (thanks to Perry
1029 E. Metzger most immediately, and others for their past work)
1030 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1031 current release version; SBCL's OpenBSD support had been broken
1032 since about the time of OpenBSD's switch to ELF binary format.)
1033 * [placeholder for DX summary]
1034 ** user code with &REST lists declared dynamic-extent, under high
1035 speed or space and low safety and debug optimization policy.
1036 * The manual has been converted to Texinfo format and the debugger
1037 chapter from the cmucl manual has been added.
1038 * A facility has been added to extract documentation strings from
1039 sbcl and store them as Texinfo-formatted snippets for inclusion in
1040 the manual (via Texinfo's @include directive)
1041 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1042 the test case to Dave Roberts)
1043 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1044 pointers. (reported by Sean Ross)
1045 * bug fix: PROFILE output is printed nicely even for large numerical
1046 values. (thanks to Zach Beane)
1047 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1048 greater than 32 handle EOF correctly.
1049 * bug fix: on X86 an immediate argument of the IMUL instruction is
1050 correctly printed in disassembly. (thanks to Lutz Euler)
1051 * bug fix: class slots in redefined classes preserve their old
1052 values. (thanks to Bruno Haible and Nikodemus Siivola)
1053 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1054 succeeds. (reported by Marco Baringer)
1055 * fixed some bugs revealed by Paul Dietz' test suite:
1056 ** READ-SEQUENCE now works on ECHO-STREAMs.
1057 ** RATIONALIZE works more according to its specification. (thanks
1060 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1061 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1062 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1063 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1064 should matter to you only if you rebind the printer control
1065 variables and then find you want different bindings in the
1066 debugger than in the ordinary execution of your program.)
1067 * The runtime build system has been tweaked to support building
1068 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1069 assembler and linker. (thanks to Nikodemus Siivola)
1070 * Unbound, undefined, undeclared variables now trigger full
1071 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1072 more often programmer error than deliberate exploitation of undefined
1074 * optimization: the hash algorithm for strings has changed to one
1075 that is less vulnerable to spurious collisions. (thanks to Juho
1077 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1078 needless bounds checking. (thanks to Juho Snellman)
1079 * optimization: implemented multiplication as a modular
1080 (UNSIGNED-BYTE 32) operation on the PPC backend.
1081 * fixed some bugs revealed by Paul Dietz' test suite:
1082 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1084 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1085 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1086 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1088 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1090 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1092 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1093 their output stream on EOF from read.
1094 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1095 have been read to end-of-file.
1096 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1098 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1099 description of determination of which consecutive characters
1101 ** Printing the "Space" character with escaping on now yields "#\\ ",
1102 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1103 ** Reading floating-point numbers with *READ-BASE* set to a number
1104 less than 10 works correctly.
1105 ** Reading floating-point numbers with *READ-BASE* set to a number
1106 more than 10 works correctly.
1107 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1108 the readtable currently in effect.
1110 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1111 * minor incompatible change: parsing of namestrings on a physical
1112 (Unix) host has changed; numbers after the final #\. in a
1113 namestring are no longer interpreted as a version field. This is
1114 intented to be largely invisible to the user, except that the
1115 meaning of the namestring "*.*.*" has changed: it now refers to a
1116 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1117 should usually be replaced by
1118 (make-pathname :name :wild :type :wild :version :wild)
1119 with the added benefit that this is more likely to be portable.
1120 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1121 option to OPEN now signals an error if the file being opened
1122 exists; this may have an impact on existing code.
1123 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1124 signals received, on the PowerPC platforms (both Linux and
1125 Darwin). (thanks to Pierre Mai for pointing out the location of
1127 * several fixes on OS X: The system now builds and runs cleanly on
1128 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1129 (thanks to Brian Mastenbrook)
1130 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1131 second-resolution: integer multiples of 1/3600 between -24 and 24.
1132 (thanks to Vincent Arkesteijn)
1133 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1134 type when called with 1 argument; PEEK-CHAR checked type of
1135 PEEK-TYPE only after having read first character from a
1136 stream. (reported by Peter Graves)
1137 * bug fix: the garbage collector now has much better locality
1138 behaviour, and in particular no longer treats all memory as being
1139 exhausted when many small objects point to each other in a deeply
1141 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1142 recognized as being TYPEP their class.
1143 * bug fix: the PUSHNEW documentation string has been corrected.
1144 (thanks to Vincent Arkesteijn)
1145 * bug fix: defaulting of the value for the last of an atypically
1146 large number of multiple values being bound was not being
1147 performed correctly on the Alpha or PPC platforms
1148 * optimization: implemented multiplication as a modular
1149 (UNSIGNED-BYTE 32) operation on the x86 backend.
1150 * optimization: SEARCH on simple-base-strings can now be open-coded.
1151 (see also contrib/compiler-extras.lisp for inspiration for
1152 teaching the compiler about the Boyer-Moore algorithm).
1153 * value, returned by MAX (and MIN) called with several EQUALP, but
1154 not EQL, arguments now does not depend on compiler settings.
1155 * fixed some bugs revealed by Paul Dietz' test suite:
1156 ** in stack analysis liveness information is propagated from
1157 non-local entry points.
1158 ** pathwise CAST removing failed when the CAST node did not start
1160 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1161 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1163 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1164 host is already defined.
1165 ** RENAME-FILE works on streams instead of signalling an internal
1167 ** PEEK-CHAR uses the current readtable when determining whether
1168 or not a character is whitespace.
1169 ** MERGE-PATHNAMES handles the case when the pathname does not
1170 specify a name while the default-pathname specifies a version.
1171 ** Pathnames now stand a better chance of respecting print/read
1173 ** Attempting to use standardized file system operators with a
1174 pathname with invalid :DIRECTORY components signals a
1176 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1178 ** DIRECTORY on logical pathnames is more correct.
1179 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1180 signal a TYPE-ERROR if their argument is not a stream.
1181 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1182 designator argument does not designate a stream.
1183 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1184 examining the synonym.
1185 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1187 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1188 element-type larger than ([UN]SIGNED-BYTE 32).
1190 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1191 * When built with the :SB-FUTEX feature, threaded builds now take
1192 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1193 for faster/more reliable mutex and condition variable support.
1194 * Incompatible change (but one you probably shouldn't have been using
1195 anyway): the interface and code for arbitrating between multiple
1196 threads in the same user session has been redesigned.
1197 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1198 error when called without an explicit environment argument.
1199 (thanks to Nikodemus Siivola)
1200 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1201 confused about their position. (thanks to Adam Warner and Gerd
1203 * bug fix: Pretty printing backquoted forms with unquotations in the
1204 argument list position of various code constructs such as LAMBDA
1205 now works correctly. (reported by Paul Dietz)
1206 * bug fix: Pretty printing unquotations no longer loses all
1207 stream position information.
1208 * optimization: performance of string output streams is now less
1209 poor for multiple small sequence writes.
1210 * optimization: performance of CSUBTYPEP in the presence of complex
1211 expressions involving CONS and NOT many times has been improved.
1212 (reported by Paul Dietz)
1213 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1215 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1216 streams for the REPL, for the debugger, and so on. (thanks to
1218 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1219 * fixed some bugs revealed by Paul Dietz' test suite:
1220 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1221 is ignored if it is too small, rather than propagating through
1222 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1223 ** extremely complex negations of CONS types were not being
1224 sufficiently canonicalized, leading to inconsistencies in
1226 ** VALUES tranformer lost derived type.
1228 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1229 * fixed a bootstrapping bug: the build process no longer assumes
1230 that the various BOOLE-related constants have the same value in
1231 host and target lisps. (noted by Paul Dietz' test suite on an
1232 SBCL binary built from CLISP)
1233 * The system can now be dynamically linked on the MIPS platform,
1234 which enables dynamic loading of foreign code from Lisp. (thanks
1235 to Ralf Baechle for discussions on the MIPS ABI)
1236 * The system now records debugging information for its own source
1237 files in a filesystem-position-independent manner, relative to
1238 the "SYS" logical host.
1239 * fixed a compiler bug: MV-LET convertion did not check references
1240 to the "max args" entry point. (reported by Brian Downing)
1241 * tweaked disassembly notes to be less confident about proclaiming
1242 some instruction as an LRA. (thanks to Brian Downing)
1243 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1244 now each have their own history, command character, and other
1245 characteristics. (thanks to David Lichteblau)
1246 * fixed some bugs revealed by Paul Dietz' test suite:
1247 ** compiler failure in compiling LOGAND expressions including a
1249 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1250 shift greater than 32.
1251 ** FLUSH-DEST did not mark blocks for type check regeneration.
1252 ** HANDLER-CASE failed to accept declarations in handler clauses
1253 in some circumstances.
1255 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1256 * New code in contrib/sb-introspect (still probably not entirely
1257 stable yet) provides some support for smart Lisp development
1258 environments like SLIME.
1259 * The conditions signalled for errors occurring when loading .fasl
1260 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1261 in a way which should help ASDF recover gracefully.
1262 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1263 stuff has been cleaned up. If you code contrib/ stuff, this might
1264 affect you, and you can look at contrib/README, contrib/STANDARDS,
1265 and/or the 0.8.4.27 diff to check.
1266 * In full calls the compiler now does not generate checks for declared
1267 argument types for all arguments.
1268 * various threading fixes
1269 ** and some experimental patches which didn't make it into
1270 the main tree for this release, but which are shipped in
1271 contrib/experimental-thread.patch as a possible fix for some
1272 failures (deadlock, spinning...) in GC-intensive multithreaded
1274 * fixed PPC build problem (source code incompatibility of different
1275 library versions): added offsetof() hackery which attempts to divine
1276 where glibc maintainers put uc_mcontext today
1277 * fixed bug 282: compiler does not trust type assertions while passing
1278 arguments to a full call.
1279 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1280 &OPTIONAL VALUES type parameter.
1281 * fix bug 214: algorithm for noting rejected templates is now more
1282 similar to that of template seletion. (also reported by rydis on
1284 * fixed bug 141b: printing backquoted information readably and prettily
1285 inserts a space where necessary.
1286 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1287 (CHARACTER 10) are now reported as errors, rather than propagated
1288 as unknown types. (reported by piso on #lisp)
1289 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1290 with values NIL and :ERROR. (thanks to Milan Zamazal)
1291 * fixed bug 191c: CLOS now does proper keyword argument checking as
1292 described in CLHS 7.6.5 and 7.6.5.1.
1293 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1294 longer raise an error; further, using a list as a for-as-arithmetic
1295 counter now raises a meaningful error.
1296 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1297 now understood by sequence creation functions such as MAKE-SEQUENCE
1299 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1300 a STRING-INPUT-STREAM.
1301 * compiler enhancement: SIGNUM is now better able to derive the type
1303 * type declarations inside WITH-SLOTS are checked. (reported by
1305 * fixed some bugs revealed by Paul Dietz' test suite:
1306 ** incorrect optimization of TRUNCATE for a positive first
1307 argument and negative second.
1308 ** compiler failure in let-convertion during flushing dead code.
1309 ** compiler failure while deriving type of TRUNCATE on an
1310 interval, containing 0.
1311 ** ASH of a negative bignum by a negative bignum count now returns
1313 ** intersection of CONS types now canonicalizes properly, fixing
1314 inconsistencies in SUBTYPEP.
1316 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1317 * incompatible change: The --disable-debugger command line
1318 option now clobbers the debugger at a more fundamental
1319 level, by redefining #'INVOKE-DEBUGGER instead of by
1320 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1321 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1322 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1323 enter the debugger REPL and then suffer recursive errors
1324 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1325 while under the new system, BREAK in --disable-debugger mode
1326 terminates the system just as an unhandled error would.
1327 * fixed compiler performance when processing loops with a step >1;
1328 * bug fix: DOCUMENTATION now retrieves generic function
1329 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1330 support has been systematized, and now supports the methods
1331 specified by ANSI, along with a default method and a method for
1332 slot documentation. (reported by Nathan Froyd)
1333 * bug fix: effective methods associated with a generic function are
1334 no longer cached over a change of that generic function's method
1335 combination. (reported by Andreas Fuchs)
1336 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1337 names a non-existent directory. (thanks to Andreas Fuchs)
1338 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1339 the PPC platform, lead to stack corruption; this has been fixed.
1340 (reported by Rainer Joswig)
1341 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1342 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1343 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1344 arithmetic in general, is now much more reliable on the PPC
1346 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1347 platform now returns the right answer.
1348 * optimization: restored some effective method precomputation in
1349 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1350 precomputation is now tunable.
1351 * optimization: compiler-internal data structure use has been
1352 reviewed, and changes have been made that should improve the
1353 performance of the compiler by about 20%.
1354 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1355 simple-base-strings and simple-bit-vectors is improved.
1356 * optimization: the optimization of 32-bit logical and arithmetic
1357 functions introduced in version 0.8.3 on the x86 has been
1358 implemented on the mips, ppc and sparc platforms; an
1359 implementation of the same facility, but for 64-bit arithmetic,
1360 has been added for the alpha.
1361 * microoptimization: the compiler is better able to make use of the
1362 x86 LEA instruction for multiplication by constants.
1363 * bug fix: in some situations compiler did not report usage of
1364 generic arithmetic in (SPEED 3) policy.
1365 * bug 145b fix: compiler used wrong type specifier while converting
1366 MEMBER-types to numeric.
1367 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1369 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1370 index argument now works properly on the Alpha platform.
1371 * bug fix: floating point exception treatment on the Alpha platform
1373 * bug fix: FILE-POSITION works much better on string input and
1374 output streams. (thanks to Nikodemus Siivola)
1375 * bug fix: many threading/garbage collection symptoms sorted.
1376 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1377 might be pseudo-atomic.
1378 * internal change: Stopping for GC is now done with signals not
1379 ptrace. GC is now done in whichever thread wanted it, instead of
1381 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1383 * bug fix: result form in DO is not contained in the implicit
1385 * incompatible change: ICR structure is changed; the value part of
1386 CONTINUATION is now called LVAR; corresponding functions are
1387 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1388 * added type deriver for ISQRT (thanks to Robert E. Brown).
1389 * bug fix: better support for loading from the command line when an
1390 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1392 * fixed some bugs revealed by Paul Dietz' test suite:
1393 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1394 ** ROUND and FROUND now give the right answer when given very
1395 small float arguments.
1396 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1398 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1399 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1400 ** LCM should return a non-negative integer.
1401 ** PARSE-INTEGER returned the index of a terminator instead of the
1402 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1403 ** PARSE-INTEGER returned an incorrect index being applied to a
1405 ** LCM with two arguments of 0 returns 0 rather than signalling
1407 ** unsigned addition of a 32-bit constant with the high bit set no
1408 longer causes an internal compiler error on the x86.
1409 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1410 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1411 with negative last argument.
1412 ** byte specifiers with zero size and position no longer cause
1413 an error during type derivation.
1414 ** bignum multiplication on the Alpha platform now returns the
1416 * porting: The system now builds on SuSE AMD64, although it still
1417 generates a 32-bit binary.
1418 * .fasl file incompatibility: The fasl file version number has
1419 been incremented (because of the changes to internal compiler
1420 data structures referred to above).
1422 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1423 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1424 more accurately, on the Darwin kernel running on PowerPC hardware.
1425 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1426 * Compiler code deletion notes now signal a condition of type
1427 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1428 an associated MUFFLE-WARNING restart.
1429 * The compiler now performs limited argument count validation of
1430 constant format strings in FORMAT, and where appropriate in ERROR,
1431 CERROR and WARN. (thanks to Gerd Moellmann)
1432 * New ASDF-INSTALL contrib can be used for automatic download and
1433 installation of third-party Lisp code from CCLAN or other sites
1435 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1436 which forces another thread to execute a function supplied by the
1438 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1439 now accept and act upon their :ELEMENT-TYPE keyword argument.
1440 (reported by Martin Atzmueller, Edi Weitz)
1441 * bug fix: FILE-POSITION now accepts position designators up to
1442 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1443 is the greater. (thanks to Patrik Nordebo)
1444 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1445 Kalle Olavi Niemitalo)
1446 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1447 printable. (reported by Eric Marsden)
1448 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1449 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1450 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1451 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1452 (thanks to Stig Sandoe)
1453 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1454 UNREAD-CHAR now correctly includes the unread character in the
1455 target sequence. (thanks to Gerd Moellmann)
1456 * bug fix in threaded builds: the system can now be suspended and
1457 resumed by shell job control with minimal disruption.
1458 * bug fixes in times and timezones >2038AD
1459 * better handling of "where is GNU make?" problem in build scripts
1460 (thanks to Nikodemus Siivola)
1461 * new optimization: inside a named function any reference to a
1462 function with the same name is considered to be a self-reference;
1463 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1464 optimization quality.
1465 * new optimization on x86: logical functions and + now have
1466 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1467 used when the result is truncated to 32 bits.
1468 * VALUES declaration is partially enabled.
1469 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1470 * fixed some bugs revealed by Paul Dietz' test suite:
1471 ** The system now obeys the constraint imposed by
1472 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1473 types form a lattice under type intersection.
1474 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1475 ** ASSOC now ignores NIL elements in an alist.
1476 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1477 and (1+ MOST-POSITIVE-FIXNUM) answers.
1478 ** The addition of a method with invalid qualifiers to a generic
1479 function does not cause an error to be signalled immediately;
1480 a warning is signalled, and the error is generated only on
1481 calling the generic function.
1482 * changed .fasl file version number, in part to add type codes for
1483 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1484 obscure ANSI requirements
1486 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1487 * fixed bug 148: failure to inline-expand a local function left
1488 garbage, confusing the compiler.
1489 * fixed bugs 3cd: structure slot readers perform type check if the
1490 slot can have an invalid value (i.e. it is either not initialized
1491 or can be written with a less specific slot writer).
1492 * bug fix: the compiler now traps array references to elements off
1493 the end of an array; previously, the bounds checking in some
1494 circumstances could go off-by-one.
1495 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1497 * type declarations for array element types now obey the description
1498 on the CLHS page "Declaration TYPE", as per discussions on
1499 sbcl-help around 2003-05-08. This means that a declaration
1500 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1501 declaration, all references to BAR will be asserted or assumed
1502 (with THE, so dependent on compiler policy) to involve objects of
1503 type FOO. Note that no such declaration is implied in
1504 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1505 * declared types of functions from the "Conditions"
1506 chapter. (reported by Paul Dietz)
1507 * bug fix: CERROR accepts a function as its first argument.
1508 * bug fix: NTH an NTHCDR accept a bignum as index
1509 arguments. (reported by Adam Warner)
1510 * optimization: character compare routines now optimize comparing
1511 against a constant character. (reported by Gilbert Baumann)
1512 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1513 argument now works properly on the MIPS platform.
1514 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1515 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1516 declared to have a functional type, does not cause an error
1518 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1519 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1520 * fixed bug: if last continuation of a deleted block has a
1521 destination, this destination should be deleted too. (reported by
1523 * fixed a bug in the bootstrap process: the host compiler's values
1524 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1525 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1526 test case from Patrik Nordebo)
1527 * improved the ability of the disassembler on the PPC platform to
1528 provide helpful disassembly notes.
1529 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1530 the class in more cases than previously.
1531 * bug fix: FILE-POSITION now understands :START and :END for
1532 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1533 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1534 8) data. (thanks to David Lichteblau)
1535 * bug fix: it is possible to add a method to a generic function
1536 without lambda list.
1537 * bug fix: reader failed to signal END-OF-FILE inside an
1538 object representation. (reported by Nikodemus Siivola)
1539 * fixed some bugs revealed by Paul Dietz' test suite:
1540 ** LAST and [N]BUTLAST should accept a bignum.
1541 ** condition slot accessors are methods.
1542 ** (VECTOR NIL) is a subtype of STRING.
1544 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1545 * minor incompatible change: some nonsensical specialized lambda
1546 lists (used in DEFMETHOD) which were previously ignored now signal
1548 * minor incompatible change: the system is now aware of the types of
1549 variables in the COMMON-LISP package, and will signal errors for
1550 most violations of these type constraints (where previously they
1551 were silently accepted).
1552 * minor incompatible change: COMPILE-FILE now uses the freedom
1553 afforded (ANSI 3.2.2.3) to use derived function types for
1554 functions defined in the same file. This also permits the system
1555 to warn on static type mismatches and function
1556 redefinition. (Currently it does not work with high DEBUG level.)
1557 * minor incompatible change: VALUES declaration is disabled.
1558 * When issuing notes, the compiler now signals a condition of type
1559 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1560 restart for use in user handlers. It is expected that the
1561 COMPILER-NOTE condition will eventually become a condition
1562 supertype to a hierarchy of note types, which will then be
1563 handleable in a similar fashion. However, at the moment, no such
1564 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1565 but it's an implementation detail, not a classification for the
1567 * Changes in type checking closed the following bugs:
1568 ** type checking of unused values (192b, 194d, 203);
1569 ** template selection based on unsafe type assertions (192c, 236);
1570 ** type checking in branches (194bc).
1571 * A short form of VALUES type specifier has ANSI meaning (it has
1572 increased the number of situations when SBCL cannot perform type
1574 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1575 keywords or constants is permissible.
1576 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1577 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1578 classes. (thanks to Antonio Martinez)
1579 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1580 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1581 argument to a pretty-print stream. Instead, it leaves any such
1582 operations to DESCRIBE-OBJECT methods.
1583 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1585 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1586 lambda lists. (thanks to David Lichteblau)
1587 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1588 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1589 (thanks to Gerd Moellmann)
1590 * fixed bug 246: increased compilation speed of long
1591 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1593 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1595 * DEFCONSTANT now throws a condition of type
1596 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1597 constant to a non-EQL value; CONTINUE and ABORT restarts
1598 respectively change and preserve the value.
1599 * fixed bug 63: The code walker, part of the implementation of CLOS,
1600 is now better at handling symbol macros.
1601 * bug fix: There is no longer an internal implementation type named
1602 CL:LENGTH. (reported by Raymond Toy)
1603 * bug fix: In macro-like defining macros/special operators the
1604 implicit block does not enclose the lambda list.
1605 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1606 enclosed in parentheses) are not suitable as type specifiers, and
1607 their use properly signals an error now.
1608 * bug fix: An argument count mismatch for a type specifier in code
1609 being compiled no longer causes an unhandled error at compile
1610 time, but signals a compile-time warning.
1611 * fixed simple vector readable printing
1612 * bug fix: DESCRIBE takes more care over whether the class
1613 precedence list slot of a class is bound before accessing it.
1614 (reported by Markus Krummenacker)
1615 * bug fix: FORMATTER can successfully compile pretty-printer format
1616 strings which use variants of the ~* directive inside.
1617 * bug fix: SEARCH now applies its TEST predicate to the elements of
1618 the arguments in the correct order. (thanks to Wolfhard Buss)
1619 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1620 in type inference. (thanks to Robert E. Brown)
1621 * bug fix: Reading in symbols with an explicit package name of ""
1622 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1623 * fixed some bugs revealed by Paul Dietz' test suite:
1624 ** NIL is now allowed as a structure slot name.
1625 ** Arbitrary numbers, not just REALs, are allowed in certain
1626 circumstances in LOOP for-as-arithmetic clauses.
1627 ** Multiple class redefinitions before slot access no longer
1628 causes a type error.
1629 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1630 association between the name and a class.
1631 ** Generic functions with non-standard method-combination and over
1632 five methods all of which return constants no longer return NIL
1633 after the first few invocations. (thanks to Gerd Moellmann)
1634 ** CALL-NEXT-METHOD with no arguments now passes the original
1635 values of the arguments, even in the presence of assignment.
1636 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1638 ** DISASSEMBLE works with closures and funcallable instances.
1639 ** ADD-METHOD now returns the generic function, not the new method.
1640 ** FIND-METHOD signals an error if the lengths of the specializers
1641 is incompatible with the generic function, even if the ERRORP
1643 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1644 which its argument is a member.
1645 ** DEFCLASS only redefines the class named by its class-name
1646 argument if that name is the proper name of the class;
1647 otherwise, it creates a new class.
1648 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1649 of the UNBOUND-SLOT condition to the name of the slot.
1650 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1651 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1652 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1653 SLOT-MAKUNBOUND in the specified fashion.
1655 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1656 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1657 cross-compilation host. As a consequence, we can now bootstrap our
1658 way up to SBCL starting with a bare gcc toolchain and human-readable
1659 source code (first the source to CLISP, then the source to SBCL).
1660 * A contributed module containing a partial implementation of the
1661 simple-streams interface has been included. (thanks to Rudi
1663 * A contributed module implementing the RFC1321 Message Digest
1664 Algorithm, known as MD5, has been included.
1665 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1666 no longer has any effect, as the code controlled by this feature
1667 has been deleted. (As far as we know, no-one has ever built using
1668 this feature, and its semantics were confused in any case).
1669 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1670 work on conditions (as required by the ANSI specification),
1671 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1672 expected behaviour on conditions. Users should note, however,
1673 that such behaviour is not required by the ANSI specification,
1674 and so use of this behaviour may render their code unportable.
1675 * fixed some bugs revealed by Paul Dietz' test suite:
1676 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1678 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1679 ** SXHASH on condition objects no longer returns NIL.
1680 ** :ALLOCATION :CLASS slots are better treated; their values are
1681 updated on class redefinition, and initforms inherited from
1682 superclasses are applied.
1683 ** REMOVE-METHOD returns its generic function argument even when
1684 no method was removed.
1685 ** SHARED-INITIALIZE now initializes the values of the requested
1686 slots, including those with :ALLOCATION :CLASS.
1687 ** ALLOCATE-INSTANCE now works on structure classes defined via
1688 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1690 ** SLOT-EXISTS-P now works on conditions, as well as structures
1692 ** MAKE-LOAD-FORM now has the required methods on
1693 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1694 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1695 keyword, and now implements the SLOT-NAMES argument.
1696 ** methods with &OPTIONAL arguments no longer allow too many
1697 arguments to be passed in the call without error.
1698 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1699 option is consistent with the required arguments of the generic
1700 function lambda list.
1701 * bug fix: REQUIRE accepts a string designator. (Thanks to
1703 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1704 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1705 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1706 * bug fix: The compiler now checks for duplicated variables in macro
1708 * bug fix: SETQ on globals returns the correct value.
1709 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1710 not a COMPILER-ERROR (followed by some other strange error on
1711 choosing the CONTINUE restart).
1712 * bug fix: make.sh and friends are now more consistent in the way that
1713 they look for GNU "make".
1715 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1716 * experimental native threads support (on x86 Linux >=2.4 only).
1717 This is not compiled in by default: you need to add :SB-THREAD to
1718 the target features. See the "Beyond ANSI" chapter of the manual
1720 * fix for longstanding nonANSIism: The old distinction between
1721 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1722 The return value from CL:FIND-CLASS is now a CLOS class, and
1723 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1724 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1725 * An interface to the MetaObject Protocol, as described in Kiczales,
1726 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1727 MIT Press, 1991, is available from the SB-MOP package.
1728 * incompatible change: the SB-PCL package should now be considered
1729 a private implementation detail, and no longer a semi-private MOP
1731 * minor incompatible change: due to rearrangement for threads, the
1732 control stack and binding stack are now allocated at arbitrary
1733 addresses instead of being hardcoded per-port. Users affected by
1734 this probably have to be doing advanced things with shared
1735 libraries, and will know who they are.
1736 * minor incompatible change: Previously, all --eval forms used were
1737 processed with READ before any of them were processed with EVAL.
1738 Now each --eval form is processed with both READ and EVAL before
1739 the next --eval form is processed. (Thus package operations like
1740 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1741 work as the user might reasonably expect.)
1742 * minor incompatible change: *STANDARD-INPUT* is now only an
1743 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1745 * minor incompatible change: Y-OR-N-P is now character-oriented,
1746 not line oriented. Also, YES-OR-NO-P now works without errors.
1747 (thanks to Antonio Martinez)
1748 * sb-aclrepl module improvements: an integrated inspector, added
1749 repl features, and a bug fix to :trace command.
1750 * Known functions, which cannot be open coded by the backend, are
1751 considered to be able to check types of their arguments. (fixing
1752 a bug report by Nathan J. Froyd)
1753 * fixed a bug in computing method discriminating functions: It is
1754 now possible to define methods specialized on classes which have
1755 forward-referenced superclasses. (thanks to Gerd Moellmann)
1756 * fixed evaluation order in optional entries (reported by Gilbert
1758 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1759 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1760 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1761 required-by-AMOP NAME argument, as well as CLASS and
1762 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1763 * fixed bug 20: DEFMETHOD can define methods using names that are
1764 not the proper names of classes to designate class specializers.
1765 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1766 Unix file descriptors, instead of blowing up. (thanks to Antonio
1768 * Garbage collection refactoring: user-visible change is that a
1769 call to the GC function during WITHOUT-GCING will not do garbage
1770 collection until the end of the WITHOUT-GCING. If you were doing
1771 this you were probably losing anyway.
1772 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1773 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1774 TYPEP the latter but not the former.
1775 * The compiler issues a full WARNING for calls to undefined functions
1776 with names from the CL package.
1777 * MAP-INTO for a vector destination is open coded. (reported by
1778 Brian Downing on c.l.l)
1779 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1780 documentation string.
1781 * fixed some bugs revealed by Paul Dietz' test suite:
1782 ** COPY-ALIST now signals an error if its argument is a dotted
1784 ** Condition slots are now accessed more correctly in the presence
1785 of multiple initargs for a given slot.
1786 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1787 exclude restarts of the same name associated with a different
1789 ** DEFCLASS of forward-referenced classes with another
1790 forward-referenced class in the superclasses list no longer
1792 ** Condition slots are now initialized once each, not multiple
1793 times. (thanks to Gerd Moellmann)
1794 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1796 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1797 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1798 arguments contain duplicated elements.
1799 ** RESTART-CASE understands local macros.
1800 ** RESTART-CASE associates exactly its own restarts with a condition.
1801 ** ENDP in safe mode checks its argument to be of type LIST.
1802 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1803 in question is unbound.
1804 ** Optimized MAKE-INSTANCE functions no longer cause internal
1805 assertion failures in the presence of duplicate initargs.
1806 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1807 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1808 propagated even in the ctor optimized implementation of
1810 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1812 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1813 * a better implementation of SXHASH on (simple) bit vectors,
1814 measured both in execution speed and in distribution of results
1815 over the positive fixnums, has been installed. Likewise, a better
1816 implementation of EQUAL for simple bit vectors is now available.
1817 * fixed CEILING optimization for a divisor of form 2^k.
1818 * fixed bug 240 (emitting extra style warnings "using the lexical
1819 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1820 by Antonio Martinez)
1821 * fixed SXHASH, giving different results for NIL depending on type
1822 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1823 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1824 variable. (found by Rolf Wester)
1825 * SBCL does not ignore type declarations for special
1826 variables. (reported by rif on c.l.l 2003-03-05)
1827 * some bug fixes in contrib/sb-aclrepl/
1828 * fixed some bugs revealed by Paul Dietz' test suite:
1829 ** a bug in the CONS type specifier, whereby the CAR and CDR
1830 types got intertwined, has been fixed;
1831 ** the type system is now able to reason about the interaction
1832 between INTEGER and RATIO types more completely;
1833 ** APPEND, [N]REVERSE and NRECONC check that those their
1834 arguments, which must be proper lists, are really so;
1835 ** An array specialized to be unable to hold elements has been
1836 implemented, as required -- yes, really -- by ANSI;
1837 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1838 on malformed property lists;
1840 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1841 * incompatible packaging change: in line with Unix convention,
1842 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1843 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1844 * REQUIRE and PROVIDE are now optionally capable of doing something
1845 useful. See the documentation string for REQUIRE.
1846 * infrastructure for a managed SBCL contrib system: contributed
1847 modules in this release include:
1848 ** the ASDF system definition facility;
1849 ** an interface to the BSD Sockets API;
1850 ** an ACL-like convenience interface to the repl;
1851 (thanks to Kevin Rosenberg)
1852 ** an implementation of ROTATE-BYTE, with efficient implementation
1854 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1855 gives a closer approximation to the right answer than previously.
1856 (thanks to Raymond Toy)
1857 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1858 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1859 optional environment arguments, as required by ANSI.
1860 * fixed bugs in other functions taking environment objects, allowing
1861 calls with an explicit NIL environment argument to be compiled
1863 * fixed bug 228: primary return values from
1864 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1865 COMPILE or FUNCTION.
1866 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1867 :TYPE will no longer signal errors on innocuous objects.
1868 * fixed bug 231b: SETQ is better at respecting type declarations in
1869 the lexical environment.
1870 * fixed a bug in DEFCLASS: classes named by symbols with no or
1871 unprintable packages can now be defined.
1872 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1873 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1874 * fixed compiler failure related to checking types of functions.
1875 (reported by Robert E. Brown)
1876 * the compiler is now much more consistent in its error-checking
1877 treatment of bounding index arguments to sequence functions: in
1878 (SAFETY 3) code, errors will be signalled in almost all cases if
1879 invalid sequence bounding indices are passed to functions defined
1880 by ANSI to operate on sequences.
1881 * fixed a bug in the build procedure: documentation of SBCL-specific
1882 packages is now preserved and available in the final Lisp image.
1883 * lifted FDEFINITION lookup out of loops in the implementation of
1884 many list operations. (thanks to Robert E. Brown)
1885 * fixed a bug in the reader: the #n# reader macro now works for
1886 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1887 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1888 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1889 E. Brown and Tony Martinez respectively)
1890 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1891 SLOT-BOUNDP: in particular, it is now able to optimize them much
1892 better, and is now not vulnerable to having packages renamed.
1893 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1894 always called when a slot is not present in an instance. (thanks
1896 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1897 longer signal an error on unreading a character following EOF on
1898 the previous constituent stream. (thanks to Tony Martinez)
1899 * fixed some bugs revealed by Paul Dietz' test suite:
1900 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1901 not just nonnegative fixnums;
1902 ** the logical bit-array operators such as BIT-AND now accept an
1903 explicit NIL for their "opt-arg" argument (to indicate a
1904 freshly-consed result bit-array);
1905 ** ELT now signals an error on an invalid sequence index in safe
1907 ** the type system is now cleverer about negations of numeric
1908 types, and consequently understands the BIGNUM and RATIO types
1910 ** the type system is now cleverer about the interaction between
1911 INTEGER and RATIO types: while bugs still remain, many more
1912 cases are accurately computed;
1913 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1914 if it is in the last clause;
1915 ** CONSTANTLY now correctly returns a side-effect-free function in
1917 ** DECLARE is no longer treated as a special-operator; in
1918 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1919 * incremented fasl file version number due to the change in the
1920 DEFSTRUCT-SLOT-DESCRIPTION structure.
1922 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1923 * minor incompatible change: code processed by the "interpreter" or
1924 EVAL now has a compilation optimization policy of (DEBUG 2)
1925 (changed from (DEBUG 1)) to improve debuggability of interactive
1926 development, and to allow the use of the debug RETURN command in
1928 * an experimental implementation of the RETURN command for the
1929 debugger has been included. (thanks to Frederik Kuivinen)
1930 * fixed bug 62: constraints were not propagated into a loop.
1931 * fixed bug in embedded calls of SORT (reported and investigated by
1933 * fixed some bugs revealed by Paul Dietz' test suite:
1934 ** printing and reading of arrays with some dimensions having
1935 length 0 (thanks to Gerd Moellmann);
1936 ** BOA constructor with &AUX argument without a default value does
1937 not cause a type error;
1938 ** CONSTANTP now returns true for all self-evaluating objects.
1940 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1941 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1942 accessors that are related by inheritance, as specified in the
1943 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1944 Valtteri Vuorikoski)
1945 * The compiler is now able to inline functions that were defined in
1946 a complex lexical environment (e.g. inside a MACROLET).
1947 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1949 * Support for the upcoming FreeBSD-5.0 release has been included.
1950 (thanks to Dag-Erling Smorgrav)
1951 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1952 effect when it is not in a toplevel context.
1953 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1954 stand a better chance of being correct. (thanks to Gerd
1956 * fixed bug in COERCE, which now signals an error on coercing a
1957 rational to a bounded real type which excludes the expected
1959 * The compiler is now able to derive types more accurately from the
1960 COERCE and COMPILE functions.
1961 * fixed bug 223: functional binding is considered to be constant
1962 only for symbols in the CL package.
1963 * fixed bug 231: SETQ did not check the type of a variable being set
1964 (reported by Robert E. Brown)
1965 * A new optimization for MAKE-INSTANCE has been included, fixing
1966 various bugs (including relating to :ALLOCATION :CLASS slots and
1967 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1969 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1970 ** As required by ANSI, LOOP now disallows anonymous collection
1971 clauses such as COLLECT I in conjunction with aggregate boolean
1972 clauses such as THEREIS (= I 1);
1973 ** LOOP now signals an error when any variable is reused in the
1974 same loop (including the potentially useful construct analogous
1975 to WITH A = 1 WITH A = (1+ A);
1976 ** IT is only a special loop symbol within the first clause of a
1977 conditional loop clause;
1978 ** LOOP with a typed iteration variable over a hashtable now
1979 signals a type error iff it should.
1980 * fixed some other bugs revealed by Paul Dietz' test suite:
1981 ** FILE-STREAM now names the class previously known as FD-STREAM;
1982 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1983 argument) no longer signals an error;
1984 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1985 of the slot symbol, rather than using the current package
1986 ((:CONC-NAME "") continues to intern the slot's name in the
1988 * incremented fasl file version number, because of the incompatible
1989 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1990 of the new implementation of DEFINE-COMPILER-MACRO.
1992 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1993 * Support for building SBCL for MIPS platforms running in
1994 little-endian mode has now been checked in, and basic
1995 functionality on said platforms verified.
1996 * minor incompatible change: PCL now records the pathname of a file
1997 in which methods and the like are defined, rather than its
1999 * minor incompatible change: TRUENAME now considers the truename of
2000 a file naming a directory to be the pathname with :DIRECTORY
2001 component indicating that directory.
2002 * minor incompatible change: a NAMED clause in the extended form of
2003 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2004 reason for the previous behaviour is unclear.
2005 * more systematization and improvement of CLOS and MOP conformance
2006 in PCL (thanks to Gerd Moellman and Pierre Mai):
2007 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2009 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2010 primary methods with no specializers;
2011 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2013 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2014 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2015 FORWARD-REFERENCED-CLASSes; error reporting on
2016 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2018 ** SXHASH on CLOS instances now uses a slot internal to the
2019 instance to return different numbers on distinct instances,
2020 while preserving the same return value through invocations of
2022 ** DEFMETHOD signals errors when methods with longer incongruent
2023 lambda lists are added to generic functions;
2024 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2025 CLASS, as specified in AMOP;
2026 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2027 on the order returned by the primary method for classes of
2028 class STANDARD-CLASS;
2029 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2030 * fixed some bugs shown by Paul Dietz' test suite:
2031 ** DOLIST puts its body in TAGBODY;
2032 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2034 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2035 value producing form;
2036 ** if more variables are given to PROGV than values, extra
2037 variables are bound and made to have no value;
2038 ** NSUBSTITUTE on list arguments gets the right answer with
2040 ** ELT signals an error of type TYPE-ERROR when the index argument
2041 is not a valid sequence index;
2042 ** LOOP signals (at macroexpansion time) an error of type
2043 PROGRAM-ERROR when duplicate variable names are found;
2044 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2045 ** FUNCALL of special-operators now cause an error of type
2047 ** PSETQ now works as required in the presence of side-effecting
2048 symbol-macro places;
2049 ** NCONC accepts any object as its last argument;
2050 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2052 ** loop-for-as-package does not require a package to be explicitely
2054 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2056 * fixed bug 166: compiler preserves "there is a way to go"
2057 invariant when deleting code.
2058 * fixed bug 172: macro lambda lists with required arguments after
2059 &REST arguments now cause an error to be signalled. (thanks to
2061 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2062 bug. (thanks to Matthew Danish)
2063 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2065 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2066 arguments are lexically rebound. (thanks to Gerd Moellmann and
2068 * fixed bug 194: error messages are now more informative when there
2069 is no primary method applicable in a call to a generic
2070 function. (thanks to Gerd Moellmann)
2071 * fixed bug in command line argument checking (thanks to Julian
2073 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2074 (thanks to Matthew Danish)
2075 * incremented fasl file version number, because of the
2076 SXHASH-related changes in the layout of CLOS data structures
2078 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2079 * minor incompatible change: The runtime (the Unix executable named
2080 "sbcl") is now much pickier about the .core files it will load.
2081 Essentially it now requires .core files to descend from the same
2082 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2083 as the runtime does. (The intent is to prevent the crashes which
2084 can occur, and which can even be reported as mysterious failures,
2085 when people patch the sources or change the build parameters
2086 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2087 sbcl and .core files.)
2088 * fixed bug: VALUES-LIST is no longer optimized away.
2089 * fixed bug 142: The FFI conversion of C string values to Lisp
2090 string values no longer conses excessively. (thanks to Nathan
2091 Froyd porting Raymond Toy's fix to CMU CL)
2092 * began to systematize and improve MOP conformance in PCL (thanks to
2093 Nathan Froyd, Gerd Moellman and Pierre Mai):
2094 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2096 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2097 ** FINALIZE-INHERITANCE is now called on class finalization;
2098 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2099 argument precedence order.
2100 * fixed bug 202: The compiler no longer fails on functions whose
2101 derived types contradict their declared type.
2102 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2103 so it can be non-toplevel.
2104 * The fasl file version number has changed (because of the new
2105 implementation of DEFMACRO).
2106 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2107 safe code, that any length requirement by their type-specifier
2108 argument is valid. The exceptions to this are described in bug
2110 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2111 functions now check the types of their inputs as required by ANSI.
2112 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2113 introduced symbol is DECLAREd to be SPECIAL.
2114 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2115 * fixed bug: :COUNT argument to sequence functions may be negative
2116 * fixed bug: body of DO-SYMBOLS may contain declarations
2117 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2118 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2119 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2120 symbol macro only once
2121 * fixed printing of call frame when argument list is unavailable
2122 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2123 * compiler no longer signals WARNING on unknown keyword
2126 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2127 * A beta-quality port to the mips architecture running Linux,
2128 based on the old CMUCL backend, has been made. It has been tested
2129 on a big-endian kernel, and works sufficiently well to be able to
2130 rebuild itself; it has not been tested in little-endian mode.
2131 * fixed an inconsistency between gencgc.c and purify.c which made
2132 dumping/loading .core files unreliable
2133 * fixed bug 120a: The compiler now deals correctly with IFs where
2134 the consequent is the same as the alternative, instead of
2135 misderiving the return type. (thanks to Alexey Dejneka)
2136 * fixed bug 113: Logical pathnames are now dumpable (the logical
2137 host is resolved at load-time, throwing an error if it is not
2139 * fixed bug 174: FORMAT's error message is slightly clearer when a
2140 non-printing character is used in a format directive.
2141 * fixed several bugs in compiler checking of type declarations, i.e.
2142 violations of the Python "declarations are assertions" principle
2143 (thanks to Alexey Dejneka)
2144 * fixed several bugs in PCL's error checking (thanks to Gerd
2146 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2148 * fixed bug in compilation of functions as first class values
2149 (thanks to Antonio Martinez-Shotton)
2150 * The compiler's handling TYPE-ERRORs which it can prove will
2151 inevitably happen at runtime has been cleaned up and corrected
2152 in several ways. (thanks to Alexey Dejneka)
2153 * improved argument type checking for various basic arithmetic
2154 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2155 lavished on them in the past that they can be compiled in many
2156 ways in different special cases
2157 * fixed bug 181: compiler checks validity of user supplied type
2159 * cleaned up code flushing in optimization: Function calls which
2160 should signal errors for safety purposes (e.g. which ANSI says
2161 should signal errors when their arguments are of incorrect type)
2162 are no longer optimized away.
2163 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2164 * incremented fasl file version number, because changes in the
2165 implementation of sequence functions like COERCE caused
2166 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2167 in old inline expansions) to become undefined. (Actually these
2168 changes were later undone, so we might very well be binary
2169 compatible with 0.7.7 after all, but leaving the version number
2170 incremented seemed like the simplest and most conservative
2173 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2174 * An alpha-quality port to the parisc architecture running Linux,
2175 based on the old CMUCL backend, has been made. This, even more so
2176 than the other backends, should be considered still a work in
2177 progress; known problems include that the Linux kernel in 64-bit
2178 mode does not propagate the correct sigcontext structure to
2179 userspace, and consequently SBCL on a parisc64 kernel will not
2181 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2182 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2183 suppressed.) Also now that the compiler is looking at declarations
2184 in the environment, it checks optimization declarations as well,
2185 and suppresses inlining when (> DEBUG SPEED).
2186 * More fixes have been made to treatment of floating point exception
2187 treatment and other Unix signals. In particular, floating point
2188 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2189 * The detection and handling of control stack exhaustion (infinite
2190 or very deeply nested recursion) has changed. Stack exhaustion
2191 detection is now done by write-protecting pages at the OS level
2192 and applies at all optimization settings; when found, a
2193 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2194 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2195 be caught using IGNORE-ERRORS.
2196 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2197 correctly when transforming recursive LABELS functions to LETs.
2198 (thanks to Alexey Dejneka)
2199 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2200 that are names of constants or global variables.
2201 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2202 alien routines with docstrings.
2203 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2204 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2206 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2207 (thanks to Eric Marsden)
2208 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2209 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2210 * Bug fix: The SPARC backend can now compile functions involving
2211 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2212 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2213 to a functions expecting a PATHNAME-DESIGNATOR.
2214 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2215 lambda lists. (thanks to Alexey Dejneka)
2216 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2217 functions so that EXPORTing the name of the class doesn't cause
2218 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2219 to Antonio Martinez for reporting this)
2220 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2221 constant-folded without failing an assertion. (thanks to Einar
2222 Floystad Dorum for reporting this)
2223 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2224 types are now dealt with more correctly by the compiler.
2225 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2226 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2227 behaviour analogous to RENAME-FILE. This puts its behaviour more
2228 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2230 * The fasl file version number has changed again. (because of the
2231 bug fix involving the names of PCL MAKE-INSTANCE functions)
2233 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2234 * bug fix: Floating point exceptions are treated much more
2235 consistently on the x86/Linux and PPC/Linux platforms.
2236 * Array initialization with :INITIAL-ELEMENT is now much faster for
2237 cases when the compiler cannot open code the array creation, but
2238 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2239 array accesses have also seen a speed increase.
2240 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2241 specified. (This was at the root of some bad interactions between
2242 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2244 * bug fix: Internal error arguments for undefined functions are now
2245 computed correctly on the PPC/Linux platform.
2246 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2247 Raymond Toy's patch for CMU CL)
2248 * Support for the Solaris 9 operating environment has been included
2249 (thanks to Daniel Merritt)
2250 * A very ugly but hopefully complete draft of the missing FFI chapter
2251 of the manual has been created by reformatting the corresponding
2252 CMU CL manual chapter into (currently very ugly and incoherent)
2253 DocBook and bringing it up to date for SBCL behavior. Thus, the
2254 manual is now essentially complete, at least by my extreme
2255 once-and-only-once standards, whereby it's acceptable to refer to
2256 the doc strings of SB-EXT functions as the primary documentation.
2257 * The fasl file version number has changed again, due to cleanup of
2258 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2259 is no longer a static symbol.)
2261 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2262 * SBCL now builds with OpenMCL (version 0.12) as the
2263 cross-compilation host; also, more progress has been made toward
2264 bootstrapping under CLISP.
2265 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2267 * bug 158 fixed: The compiler can now deal with integer loop
2268 increments different from 1; fixing this turned out also to fix
2270 * bug 169 fixed: no more bogus warnings about using lexical bindings
2271 despite the presence of perfectly good SPECIAL declarations (thanks
2272 to David Lichteblau)
2273 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2274 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2275 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2277 * bug 180 fixed: Method combination specifications no longer ignore
2278 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2279 * bug fix: Structure type predicate functions now check their argument
2280 count as they should.
2281 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2282 correctly. (thanks to Pierre Mai)
2283 * minor incompatible change: The --noprogrammer option is deprecated
2284 in favor of the new --disable-debugger option, which is very similar.
2285 (The major difference is that it takes effect at a slightly different
2286 time at startup, causing handling of errors in --sysinit and
2287 --userinit files will be affected differently.) The
2288 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2289 been added to allow this functionality to be controlled from ordinary
2290 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2291 maintainers, who might want to run non-interactive scripts to
2292 build SBCL cores which will later be used interactively.)
2293 * minor incompatible change: The LOAD function no longer, when given
2294 a wild pathname to load, loads all files matching that pathname.
2295 Instead, an error of type FILE-ERROR is signalled.
2297 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2298 * bug 147 fixed: The compiler preserves its block link/count
2299 invariants more correctly now so that it doesn't crash. (thanks
2301 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2303 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2304 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2305 no-value-for-this-character as NIL (as specified by ANSI).
2306 * HOST-NAMESTRING on physical pathnames now returns a string that is
2307 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2308 (thanks to Christophe Rhodes)
2309 * The Alpha port handles icache flushing more correctly. (thanks to
2311 * More progress has been made toward bootstrapping under CLISP. (thanks
2312 to Christophe Rhodes)
2313 * The fasl file format has changed again, because dynamic loading
2314 on OpenBSD (which has non-ELF object files) motivated some cleanups
2315 in the way that foreign symbols are transformed and passed around.
2316 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2317 is no longer treated as whitespace by the reader, but instead as
2318 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2319 |A
\7fB|, instead of A as it used to.
2321 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2322 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2323 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2324 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2325 this as of 0.7.1.45, but was left out of the previous news section
2326 (thanks to Dan Barlow)
2327 * SBCL now runs on the Solaris operating system on SPARC architectures
2328 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2329 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2330 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2331 * SPARC backend cleanups, allowing builds of cores optimized for V8
2332 and V9 SPARCS, and also emission of code targeted to a particular
2333 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2335 * SBCL is closer to bootstrapping under CLISP, thanks to various
2336 fixes by Christophe Rhodes.
2337 * The fasl file format has changed again, to allow the compiler's
2338 INFO database to support symbol macros.
2339 * The user manual (in doc/) is formatted into HTML more nicely.
2340 (thanks to coreythomas)
2341 * The system is smarter about SUBTYPEP relationships, especially
2342 those involving NOT types (including types such as ATOM which are
2343 represented internally using NOT types). Thus SUBTYPEP is less
2344 likely to return (VALUES NIL NIL) in general, and in particular
2345 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2347 * The fasl file format has changed again, because the internal
2348 representation of types now includes a new slot to support the new
2349 SUBTYPEP-of-NOT-types logic.
2350 * (not a change in the main branch of SBCL, but a related prototype
2351 which can hopefully be merged into the main branch of SBCL in the
2352 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2353 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2354 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2355 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2356 dispatch macro character combinations. (thanks to Alexey Dejneka)
2357 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2358 that superficially look like logical namestrings correctly.
2359 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2361 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2362 default supplied-p) for &optional and &key arguments. (thanks to
2365 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2366 * incompatible change: The compiler is now less aggressive about
2367 tail call optimization, doing it only when (> SPACE DEBUG) or
2368 (> SPEED DEBUG). (This is an incompatible change because there are
2369 programs which relied on the old CMU-CL-style behavior to optimize
2370 away their unbounded recursion which will now die of stack overflow.)
2371 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2372 for non-GENCGC systems has been increased to 20M (since that
2373 seems much closer to the likely performance optimum for modern
2374 systems than the old 4M value was)
2375 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2376 and *DEBUG-PRINT-LEVEL*
2377 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2378 of CMU CL's support for SPARC, and various endianness and other
2379 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2380 * new syntactic sugar for the Unix command line: --load foo.bar is now
2381 an alternate notation for --eval '(load "foo.bar")'.
2383 ** The system now detects stack overflow and handles it gracefully,
2384 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2385 optimization settings. (This is a good thing in general, and
2386 its introduction in this version should be particularly timely
2387 for anyone whose code fails because of suppression of tail
2389 ** The system now hunts for the C variable "environ" in a more
2390 devious way, to avoid segfaults when the C library version differs
2391 between compile time and run time. (thanks to Christophe Rhodes)
2392 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2393 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2394 ** The compiler no longer issues bogus style warnings for undefined
2395 classes in the same source file as the DEFCLASSes which defined
2396 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2398 ** fixes in CONDITION class precedence list for undefined function
2399 errors (thanks to Alexei Dejneka)
2400 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2401 correctly. (thanks to Dan Barlow)
2402 ** portability fixes aiming at bootstrapping under CLISP (thanks
2403 to Dave McDonald and Christophe Rhodes)
2404 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2405 ** fixes in type translation and and type inference (thanks to
2407 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2408 ** various fixes in the new ports (thanks to Dan Barlow)
2409 * several changes related to debugging:
2410 ** suppression of tail recursion, as noted above
2411 ** stack overflow detection, as noted above
2412 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2413 is now the default. (For some time encapsulation has been more
2414 reliable than the breakpoint-based :ENCAPSULATE NIL
2415 implementation, at least on X86 systems; and I just noticed that
2416 encapsulation also seems closer to the spirit of the ANSI
2419 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2421 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2422 up properly again. (There was a packaging bug in 0.7.0 which
2423 left their definitions in SB-SYS::LOAD-FOREIGN and
2424 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2425 vital for most things which interface to C-level interfaces,
2426 like extensions working with sockets or databases or
2427 Perl-compatible regexes or whatever, and the need to fix
2428 this bug is the main reason that 0.7.1 was released so
2430 ** DEFGENERIC is now choosier about the methods it redefines, so that
2431 reLOADing a previously-LOADed file containing DEFGENERICs does
2432 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2433 cycle now works as it should. (thanks to Alexey Dejneka)
2434 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2435 fixed by Christophe Rhodes. (He actually submitted this patch
2436 months ago, and I delayed until after 0.7.0.)
2437 ** Bug 111 (internal compiler confusion about runtime checks on
2438 FUNCTION types) was fixed by Alexey Dejneka.
2439 * Some internal cleanups (getting rid of variables which aren't
2440 needed now that the byte interpreter is gone) caused the fasl
2441 file format number to change again.
2443 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2444 * major incompatible change: The default fasl file extension, i.e. the
2445 default extension for files produced by COMPILE-FILE, has changed
2446 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2448 ** There are many changes in the implementation of the compiler.
2449 SBCL is now essentially a compiler-only implementation of ANSI
2450 Common Lisp. EVAL still "interprets" a few special cases, but
2451 almost all the interesting cases are handled by creating
2452 a LAMBDA expression, calling COMPILE on it, then calling
2453 FUNCALL on the result.
2454 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2455 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2456 Since the code is newer, there might still be some new bugs
2457 (though not as many as before Martin Atzmueller's fixes:-). But
2458 the new code is substantially simpler and clearer, and hopefully
2459 any remaining bugs will be simpler, less fundamental, and more
2460 fixable then the bugs in the old code.
2461 ** The revised compiler is still a little unsteady on its feet.
2463 *** The debugging information it produces (particularly the names
2464 of FUNCTION objects) is sometimes much less useful than what
2465 the old compiler produced.
2466 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2467 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2468 within a MACROLET) has been temporarily weakened.
2469 ** There are new compiler optimizations for various functions:
2470 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2471 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2472 *** the math functions TRUNCATE, FLOOR, and CEILING
2473 *** the function-of-all-trades COERCE
2474 Mostly these should be transparent, but there's one
2475 potentially-annoying problem (bug 117): when the compiler
2476 inline-expands a function and does type analysis on the result,
2477 it can create control paths which have type mismatches, and
2478 when it can't prove that those control paths aren't taken,
2479 it will issue WARNINGs about the type mismatches. This is
2480 a particular problem in practice for the new sequence functions.
2481 It's not clear how this should be fixed, and for now, a
2482 workaround is given in the entry for 117 in the BUGS file.
2483 ** (Because of the interaction between the two previous items --
2484 occasional inlining problems and new inline expansions -- some
2485 of the new sequence function optimizations won't really kick in
2486 completely until debugging information, and then inlining, are
2487 straightened out in some future version.)
2488 * minor incompatible changes:
2489 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2490 DIRECTORY no longer implicitly promotes NIL slots of its
2491 pathname argument to :WILD. In particular, when you ask for the
2492 contents of a directory (which you used to be able to do without
2493 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2494 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2495 ** changes in behavior that ANSI explicitly defines to be
2496 implementation dependent:
2497 *** The new compiler-only implementation still conforms with ANSI,
2498 but acts a little different than before. Besides the obvious
2499 changes in performance tradeoffs (that the cost per form passed
2500 to EVAL has gone up, and the cost per form executed by EVAL
2501 has gone down), the behavior of the system changes a little
2502 because there are no longer any interpreted function objects.
2503 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2504 e.g. doing COMPILE on the output of interactive DEFUN is
2506 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2508 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2509 has changed from (:CL) to NIL.
2510 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2511 e.g. #\Newline and #\Tab, don't have names specified in the
2512 ANSI Common Lisp standard, is now based on their ASCII symbolic
2513 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2514 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2515 are no longer used for output.
2516 ** changes in internal implementation constants:
2517 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2518 4 million. (If your application spends a lot of time GCing and
2519 you have a lot of RAM, you might want to experiment with
2520 increasing it even more.)
2521 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2522 However, almost all old code should still continue to work without
2523 immediate update, as SB-C-CALL is now a (deprecated) nickname
2525 ** Old operator names in the style DEF-FOO are now deprecated in
2526 favor of new corresponding names DEFINE-FOO, for consistency with
2527 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2528 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2529 internal symbols, but a few supported extensions like
2530 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2531 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2532 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2533 etc. as you get deeper into recursive calls to the debugger
2534 command loop, instead of the old "5]", "5]]", "5]]]"
2535 sequence. (I was motivated to do this when squabbles between
2536 ILISP and SBCL left me very deeply nested in the debugger. In the
2537 short term, this change will probably provoke more ILISP/SBCL
2538 squabbles, but hopefully it will be an improvement in the long run.)
2539 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2540 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2541 function, since declaiming return types involving aliens is
2542 (1) annoyingly messy to do by hand and (2) vital to efficient
2543 compilation of code which calls such functions.
2544 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2545 longer reexported by the SB-EXT package. They're solely useful
2546 for alien code, so it seems more logical that you should get
2547 them from the SB-ALIEN package, not in SB-EXT.
2548 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2549 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2550 features. Instead, the code that they used to control is always
2551 built into the system.
2552 * many other bug fixes
2553 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2554 advantage of the new EVAL-WHEN stuff and to clean them up in
2555 general, and they are now more ANSI-compliant in a number of
2556 ways. Martin Atzmueller is responsible for a lot of this.
2557 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2559 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2560 *** correct ERROR type for various file operations
2561 *** some fixes for Lisp streams
2562 *** DEFMETHOD syntax checking
2563 *** changing old weird representation of debug information as
2564 strings (which, among their other deficiencies, don't transform
2565 correctly when you rename packages, and don't change their
2566 print representation when you change things like *PACKAGE*
2567 and *PRINT-LENGTH*) to symbols and lists of symbols
2568 He also made several improvements and fixed several bugs in DESCRIBE.
2569 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2571 *** misbehavior of WRITE-STRING/WRITE-LINE
2572 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2573 and several other LOOP problems as well
2574 *** DIRECTORY when similar filenames are present
2575 *** DEFGENERIC with :METHOD options
2576 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2577 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2578 *** argument ordering in FIND with :TEST option
2579 *** mishandled package designator argument in APROPOS-LIST
2580 *** various problems in the backquote readmacro
2581 *** a bug in APROPOS
2582 *** probably some others that I'm not describing very well here,
2583 since the CVS log documents them by reference to sbcl-devel
2584 messages, and the SourceForge archives aren't working well.:-(
2585 ** Dan Barlow improved the Alpha port (and is making progress on the
2586 PPC port, for those of you who think different).
2587 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2588 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2589 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2590 continues to work on the SPARC port (for those of you in a position
2591 to look down upon our little PC-compatible boxes from a great height).
2592 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2593 and uses that copy, rather than the current dynamic value, when
2594 it's trying to decide whether to truncate output. Thus e.g.
2595 (let ((*print-lines* 50))
2596 (pprint-logical-block (stream nil)
2598 (let ((*print-lines* 8))
2599 (print (aref possiblybigthings i) stream)))))
2600 should now truncate the logical block only at 50 lines, instead of
2601 often truncating it at 8 lines, as it did before.
2602 * The doc/cmucl/ directory, containing old CMU CL documentation
2603 from the time of the fork, is no longer part of the base system.
2604 SourceForge has shut down its anonymous FTP service, and with it
2605 my original plan for distributing the old CMU CL documentation
2606 there. For now, if you need these files you can download an old
2607 SBCL source release and extract them from it.
2608 * The fasl file version number changed again, for dozens of reasons,
2609 some of which are apparent above.
2611 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2612 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2613 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2614 patch, so that DISASSEMBLE output is much nicer.
2615 * The code in the SB-PROFILE package now seems reasonably stable.
2616 I still haven't decided what the final interface should look like
2617 (I'd like PROFILE to interact cleanly with TRACE, since both
2618 facilities use function encapsulation) but if you have a need
2619 for profiling now, you can probably use it successfully with
2620 the current CMU-CL-style interface.
2621 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2622 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2623 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2624 link now returns the dangling link itself, and for similar
2625 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2626 link itself. (In these cases the old code signalled an error and
2627 looped endlessly, respectively.) Thus, DIRECTORY now works even
2628 in the presence of dangling and cyclic symbolic links.
2629 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2630 is now a supported extension again, since the consensus on
2631 sbcl-devel was that it can be useful for ordinary development
2632 work, not just for debugging SBCL itself.
2633 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2634 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2635 that currently-defined functions might be redefined later with
2636 different return types.
2637 * Hash tables can be printed readably, as inspired by CMU CL code
2638 of Eric Marsden and SBCL code of Martin Atzmueller.
2639 * better error handling in CLOS method combination, thanks to
2640 Martin Atzmueller porting Pierre Mai's CMU CL patches
2641 * more overflow fixes for >16Mbyte I/O buffers
2642 * A bug in READ has been fixed, so that now a single Ctrl-D
2643 character suffices to cause end-of-file on character streams.
2644 In particular, now you only need one Ctrl-D at the command
2645 line (not two) to exit SBCL.
2646 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2648 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2649 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2650 does the right thing.
2651 * The implementation of some type tests, especially for CONDITION
2652 types, is now tidier and maybe faster, due to CMU CL code
2653 originally by Douglas Crosher, ported by Martin Atzmueller.
2654 * Some math functions have been fixed, and there are new
2655 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2656 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2657 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2658 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2659 likely to be built into sbcl-0.7.0.)
2660 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2661 CL, ported to SBCL by Martin Atzmueller)
2662 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2663 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2664 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2665 * There's a new slam.sh hack to shorten the edit/compile/debug
2666 cycle for low-level changes to SBCL itself, and a new
2667 :SB-AFTER-XC-CORE target feature to control the generation of
2668 the after-xc.core file needed by slam.sh.
2669 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2670 COMPILE-FILE is no longer supported, so that now every function
2671 gets an entry point, so that block compilation looks a little
2672 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2673 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2674 returns the number of bytes consed since the system started,
2675 rather than the number consed since the first time the function
2676 was called. (The new definition parallels ANSI functions like
2677 CL:GET-INTERNAL-RUN-TIME.)
2678 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2679 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2680 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2681 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2682 * Fasl file version numbers are now independent of the target CPU,
2683 since historically most system changes which required version
2684 number changes have affected all CPUs equally. Similarly,
2685 the byte fasl file version is now equal to the ordinary
2688 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2689 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2690 is no longer recognized. I apologize for this, because it was
2691 listed in SB-EXT as a supported extension, but I found that
2692 its existing behavior was poorly specified, as well as incorrectly
2693 specified, and it looked like too much of a mess to straighten it
2694 out. I have enough on my hands trying to get ANSI stuff to work..
2695 * many patches ported from CMU CL by Martin Atzmueller, with
2696 half a dozen bug fixes in pretty-printing and the debugger, and
2697 half a dozen others elsewhere
2698 * fixed bug 13: Floating point infinities are now supported again.
2699 They might still be a little bit flaky, but thanks to bug reports
2700 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2701 as flaky as they were.
2702 * The --noprogrammer command line option is now supported. (Its
2703 behavior is slightly different in detail from what the old man
2704 page claimed it would do, but it's still appropriate under the
2705 same circumstances that the man page talks about.)
2706 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2707 are now supported, and enabled by default. Thus, the compiler can
2708 handle many floating point and complex operations much less
2709 inefficiently. (Thus e.g. you can implement a complex FFT
2711 * The compiler now detects type mismatches between DECLAIM FTYPE
2712 and DEFUN better, and implements CHECK-TYPE more correctly, and
2713 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2714 thanks to patches from Martin Atzmueller.
2715 * various fixes to make the cross-compiler more portable to
2716 ANSI-conforming-but-different cross-compilation hosts (notably
2717 Lispworks for Windows, following bug reports from Arthur Lemmens)
2718 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2719 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2720 fixed thanks to Pierre Mai's CMU CL patches.
2721 * fixes to keep the system from overflowing internal counters when
2722 it tries to use i/o buffers larger than 16M bytes
2723 * fixed bug 45a: Various internal functions required to support
2724 complex special functions have been merged from CMU CL sources.
2725 (When I was first setting up SBCL, I misunderstood a compile-time
2726 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2727 * improved support for type intersection and union, fixing bug 12
2728 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2729 more obscure bugs as well
2730 * some steps toward byte-compiling non-performance-critical
2731 parts of the system, courtesy of patches from Martin Atzmueller
2732 * Christophe Rhodes has made some debian packages of sbcl at
2733 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2734 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2735 stable, but are nonetheless usable. When he's ready, I'd be happy
2736 to add them to the SourceForge "File Releases" section. (And if
2737 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2738 * new fasl file format version number (because of changes in
2739 internal representation of (OR ..) types to accommodate the new
2740 support for (AND ..) types, among other things)
2742 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2743 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2745 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2747 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2748 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2749 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2750 * DESCRIBE now works on structure objects again.
2751 * Most function call argument type mismatches are now handled as
2752 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2753 know whether the function will be redefined before the call is
2754 executed. (The compiler could flag local calls with full WARNINGs,
2755 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2756 it doesn't keep track of enough information to know whether calls
2757 are local in this sense.)
2758 * Compiler output is now more verbose, with messages truncated
2759 later than before. (There should be some supported way for users
2760 to override the default verbosity, but I haven't decided how to
2761 provide it yet, so this behavior is still controlled by the internal
2762 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2763 src/compiler/ir1util.lisp.)
2764 * Fasl file format version numbers have increased again, because
2765 support for the Gray streams extension changes the layout of the
2766 system's STREAM objects.
2767 * The Gray subclassable streams extension now works, thanks to a
2768 patch from Martin Atzmueller.
2769 * The full LOAD-FOREIGN extension (not just the primitive
2770 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2771 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2772 but like most other programs, it defaults to copying the Unix
2773 environment from the original process instead of starting the
2774 new process in an empty environment.
2775 * Extensions which manipulate the Unix environment now support
2776 an :ENVIRONMENT keyword option which doesn't smash case or
2777 do other bad things. The CMU-CL-style :ENV option is retained
2778 for porting convenience.
2779 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2780 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2782 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2784 * A patch from Martin Atzmueller seems to have solved the SIGINT
2785 problem, and as far as we know, signal-handling now works cleanly.
2786 (If you find any new bugs, please report them!)
2787 * The system no longer defaults Lisp source file names to types
2788 ".l", ".cl", or ".lsp", but only to ".lisp".
2789 * The compiler no longer uses special default file extensions for
2790 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2791 seems to expect a single default extension for all compiled code,
2792 and there's no compelling reason to try to stretch the standard
2793 to allow two different extensions.) Instead, byte-compiled files
2794 default to the same extension as native-compiled files.
2795 * Fasl file format version numbers have increased again, because
2796 a rearrangement of internal implementation packages made some
2797 dumped symbols in old fasl files unreadable in new cores.
2798 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2799 many fewer weird special cases.
2800 * Bug #17 (differing COMPILE-FILE behavior between logical and
2801 physical pathnames) has been fixed, and some related misbehavior too,
2802 thanks to a patch from Martin Atzmueller.
2803 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2804 by Tim Moore, ported to SBCL by Martin Atzmueller.
2805 * Martin Atzmueller fixed several filesystem-related problems,
2806 including bug #36, in part by porting CMU CL patches, which were
2807 written in part by Paul Werkowski.
2808 * More compiler warnings in src/runtime/ are gone, thanks to
2809 more patches from Martin Atzmueller.
2810 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2813 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2815 * DESCRIBE now works on CONDITION objects.
2816 * The debugger now handles errors which arise when trying to print
2817 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2819 * The build system now uses an additional file, customize-target-features.lisp,
2820 to allow local modifications to the target *FEATURES* list. (The point of
2821 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2822 debugging features enabled, without having to worry about propagating it
2823 into everyone's system when I do a "cvs update".) When no
2824 customize-target-features.lisp file exists, the target *FEATURES* list
2825 should be constructed the same way as before.
2826 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2827 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2828 warn about behavior which is undefined under the ANSI spec. Note: This
2829 is specified by ANSI, but it's not very popular with programmers.
2830 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2831 macro in the SBCL sources for an example of a workaround which you
2832 might use to make portable ANSI-standard code which does what you want.
2833 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2834 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2835 when it's used as a non-toplevel form. (This is required in order
2836 to implement the DEFCONSTANT-EQX macro.)
2837 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2838 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2839 funny things, due to bugs in the implementation of EVAL-WHEN
2840 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2841 non-toplevel DEFCONSTANTs are uncommon.)
2842 * The core file version number and fasl file version number have been
2843 incremented, because the old noncompliant DEFCONSTANT behavior involved
2844 calling functions which no longer exist, and because I also took the
2845 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2846 * fixed bug 1 (error handling before read-eval-print loop starts), and
2847 redid debugger restarts and related debugger commands somewhat while
2849 ** The QUIT debugger command is gone, since it did something
2850 rather different than the SB-EXT:QUIT command, and since it never
2851 worked properly outside the main toplevel read/eval/print loop.
2852 Invoking the new TOPLEVEL restart provides the same functionality.
2853 ** The GO debugger command is also gone, since you can just invoke
2854 the CONTINUE restart directly instead.
2855 ** The TOP debugger command is also gone, since it's redundant with the
2856 FRAME 0 command, and since it interfered with abbreviations for the
2858 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2860 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2861 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2862 some porting work by Martin Atzmueller)
2863 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2864 with a boatload of patches which he ported to SBCL. Now that those
2866 ** The system tries to make sure that its low-priority messages
2867 are prefixed by semicolons, to help people who like to use
2868 syntax highlighting in their ILISP buffer. (This patch
2869 was originally due to Raymond Toy.)
2870 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2871 patches originally written by Raymond Toy.
2872 ** The compiler understands coercion between single-value and
2873 multiple-VALUES type expressions better, getting rid of some very
2874 weird behavior, thanks to patches originally by Robert MacLachlan
2875 and Douglas Crosher.
2876 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2877 lambda lists, thanks to a patch originally by Pierre Mai.
2878 ** The system no longer bogusly warns about "abbreviated type
2880 ** The compiler gets less confused by inlining and RETURN-FROM,
2881 thanks to some patches originally by Tim Moore.
2882 ** The system no longer hangs when dumping circular lists to fasl
2883 files, thanks to a patch originally from Douglas Crosher.
2884 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2885 undefined function error.
2886 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2887 in structure slot initforms) for now, documented workaround instead:-|
2888 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2889 * fixed bug 5: added stubs for various Gray stream functions called
2890 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2891 installed, at least appropriate type errors are generated
2892 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2893 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2894 consistently in DEFMETHOD forms.
2895 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2896 it doesn't seem to affect SBCL after all
2897 * The C runtime system now builds with better optimization and many
2898 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2900 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2902 * The system is now under CVS at SourceForge (instead of the
2903 CVS repository on my home machine).
2904 * The new signal handling code has been tweaked to treat register
2905 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2906 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2907 downstream, e.g. in debug-int.lisp, has implicit dependencies
2908 on the unsignedness of integer representation of machine words,
2909 and that caused the system to bomb out with infinite regress
2910 when trying to recover from type errors involving signed values,
2911 e.g. (BUTLAST '(1 2 3) -1).)
2912 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2913 (This was one of the bugs Peter Van Eynde reported back in July.)
2914 * The system now uses code inspired by Colin Walters' O(N)
2915 implementation of MAP (from the cmucl-imp@cons.org mailing
2916 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2917 inline the MAP operation, and there is more than one
2918 sequence argument to the MAP call (so that it can't just
2919 do ETYPECASE once and for all based on the type of the
2920 single sequence argument). (The old non-inline implementation
2921 of the general M-argument sequence-of-length-N case required
2922 O(M*N*N) time when any of the sequence arguments were LISTs.)
2923 * The QUIT :UNIX-CODE keyword argument has been renamed to
2924 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2925 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2927 * Raymond Wiker's patches to port dynamic loading from Linux to
2928 FreeBSD have been added.
2929 * The BUGS file is now more nearly up to date, thanks in large part
2930 to Martin Atzmueller's review of it.
2931 * The debugger now flushes standard output streams before it begins
2932 its output ("debugger invoked" and so forth).
2933 * The core version number and fasl file version number have both
2934 been incremented, because of incompatible changes in the layout
2936 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2937 used to not be called for a saved Lisp image.)
2938 * Martin Atzmueller's version of a patch to fix a compiler crash,
2939 as posted on sbcl-devel 13 September 2000, has been installed.
2940 * Instead of installing Martin Atzmueller's patch for the
2941 compiler transform for SUBSEQ, I deleted the compiler transform,
2942 and transforms for some similar consing operations.
2943 * A bug in signal handling which kept TRACE from working on OpenBSD
2945 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2946 away by constant folding
2947 * The system now defines its address space constants in one place
2948 (in the Lisp sources), and propagates them automatically elsewhere
2949 (through GENESIS and the sbcl.h file). Therefore, patching the
2950 address map is less unnecessarily tedious and error-prone. The
2951 Lisp names of address space constants have also been systematized.
2952 * CVS tags like dollar-Header-dollar have been removed from
2953 the sources, because they have never saved me trouble and
2954 they've been source of trouble working with patches and other
2955 diff-related operations.
2956 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2957 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2959 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2961 * The system has been ported to OpenBSD.
2962 * The system now compiles with a simple "sh make.sh" on the systems
2963 that it's supported on. I.e., now you no longer need to tweak
2964 text in the target-features.lisp-expr and symlinks in src/runtime/
2965 by hand, the make.sh takes care of it for you.
2966 * The system is no longer so grossly inefficient when compiling code
2967 involving vectors implemented as general (not simple) vectors (VECTOR T),
2968 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2969 which dares to use the various sequence functions on non-simple
2970 vectors, takes less of a performance hit.
2971 * There is now a primitive type predicate VECTOR-T-P
2972 to test for the (VECTOR T) type, so that e.g.
2973 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2974 can now be compiled with some semblance of efficiency. (The old code
2975 turned the type declaration into a full call to %TYPEP at runtime!)
2976 * AREF on (VECTOR T) is still not fast, since it's still compiled
2977 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2978 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2979 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2980 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2981 without making full calls to SUBTYPEP at runtime.
2982 (Some analogous efficiency issues for non-simple vectors specialized to
2983 element types other than T, or for non-simple multidimensional arrays,
2984 have not been addressed. They could almost certainly be handled the
2985 same way if anyone is motivated to do so.)
2986 * The changes in array handling break binary compatibility, so
2987 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2988 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2989 instead of (VALUES T T).
2990 * By following the instructions that Dan Barlow posted to sbcl-devel
2991 on 2 July 2000, I was able to enable primitive dynamic object
2992 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2993 functionality is not implemented (since it calls ld to resolve
2994 library references automatically, requiring RUN-PROGRAM for its
2995 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2996 try to resolve library references) is now supported.
2997 * The system now flushes the standard output streams when it terminates,
2998 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2999 them at several other probably-convenient times, e.g. in each pass of
3000 the toplevel read-eval-print loop, and after evaluating a form given
3001 as an "--eval" command-line option. (These changes were motivated by a
3002 discussion of stream flushing issues on cmucl-imp in August 2000.)
3003 * The source transform for TYPEP of array types no longer assumes
3004 that an array whose element type is a not-yet-defined type
3005 is implemented as an array of T, but instead punts, so that the
3006 type will be interpreted at runtime.
3007 * There is now some support for cross-compiling in make.sh: each of
3008 the phases of make.sh has its own script. (This should be transparent
3009 to people doing ordinary, non-cross-compile builds.)
3010 * Since my laptop doesn't have hundreds of megabytes of memory like
3011 my desktop machine, I became more motivated to do some items on
3012 my to-do list in order to reduce the size of the system a little:
3013 ** Arrange for various needed-only-at-cold-init things to be
3014 uninterned after cold init. To support this, those things have
3015 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3016 symbols with such names are now uninterned after cold init).
3017 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3018 fasl files for cold load.
3019 ** Remove the old compiler structure pooling code (which used to
3020 be conditional on the target feature :SB-ALLOC) completely.
3021 ** Redo the representation of some data in cold init to be more compact.
3022 (I also looked into supporting byte compiled code at bootstrap time,
3023 which would probably reduce the size of the system a lot, but that
3024 looked too complicated, so I punted for now.)
3025 * The maximum signal nesting depth in the src/runtime/ support code has
3026 been reduced from 4096 to 256. (I don't know any reason for the very
3027 large old value. If the new smaller value turns out to break something,
3028 I'll probably just bump it back up.)
3029 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3031 * Many, many bugs reported by Peter Van Eynde have been added to
3032 the BUGS list; some have even been fixed.
3033 * While enabling dynamic object file loading, I tried to make the
3034 code easier to understand, renaming various functions and variables
3035 with less ambiguous names, and changing some function calling
3036 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3037 * While trying to figure out how to do the OpenBSD port, I tried to
3038 clean up some of the code in src/runtime/. In particular, I dropped
3039 support for non-POSIX signal handling, added various comments,
3040 tweaked the code to reduce the number of compilation warnings, and
3041 renamed some files to increase consistency.
3042 * To support the new automatic configuration functionality in make.sh,
3043 the source file target-features.lisp-expr has been replaced with the
3044 source file base-target-features.lisp-expr and the machine-generated
3045 file local-target-features.lisp-expr.
3046 * fixed a stupid quoting error in make.sh so that using CMU CL
3047 "lisp -batch" as cross-compilation host works again
3049 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3051 * DESCRIBE no longer tries to call itself recursively to describe
3052 bound/fbound values, so that it no longer fails on symbols which are
3053 bound to themselves (like keywords, T, and NIL).
3054 * DESCRIBE now works on generic functions.
3055 * The printer now prints less-screwed-up representations of closures
3056 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3057 * A private symbol is used instead of the :EMPTY keyword previously
3058 used to mark empty slots in hash tables. Thus
3059 (DEFVAR *HT* (MAKE-HASH-TABLE))
3060 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3061 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3062 now does what ANSI says that it should. (You can still get
3063 similar noncompliant behavior if bang on the hash table
3064 implementation with all the symbols you get back from
3065 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3066 This breaks binary compatibility, since tests for equality to
3067 :EMPTY are wired into things like the macroexpansion of
3068 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3070 * There's now a minimal placeholder implementation for CL:STEP,
3071 as required by ANSI.
3072 * An obscure bug in the interaction of the normal compiler, the byte
3073 compiler, inlining, and structure predicates has been patched
3074 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3075 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3077 * Missing ordinary arguments in a macro call are now detected even
3078 when the macro lambda list contains &KEY or &REST.
3079 * The debugger no longer complains about encountering the top of the
3080 stack when you type "FRAME 0" to explicitly instruct it to go to
3081 the top of the stack. And it now prints the frame you request even
3082 if it's the current frame (instead of saying "You are here.").
3083 * As specified by ANSI, the system now always prints keywords
3084 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3086 * The default initial SIZE of HASH-TABLEs is now smaller.
3087 * Type information from CLOS class dispatch is now propagated
3088 into DEFMETHOD bodies, so that e.g.
3089 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3091 is now basically equivalent to
3092 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3093 (DECLARE (TYPE SINGLE-FLOAT X))
3095 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3096 operation, without having to do run-time type dispatch.
3097 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3098 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3099 * Since I don't seem to be making big file reorganizations very often
3100 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3101 is overflowing my ability to conveniently back them up), I've finally
3102 checked the system into CVS. (The CVS repository is on my home system,
3103 not at SourceForge -- putting it on SourceForge might come later.)
3104 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3105 high-level GC-NOTIFY-FOO functions send their output. (There's
3106 still very little control of where low-level verbose GC functions
3107 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3108 less than it used to -- the GC-NOTIFY-FOO functions are now under
3109 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3110 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3111 in only one place in the source code, and propagates it automatically
3112 everywhere that it's needed. Thus e.g. when I bump the version from
3113 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3114 * The C source files now include boilerplate legalese and documentation
3115 at the head of each file (just as the Lisp source files already did).
3116 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3117 to his page will be replaced with a link to his new CLiki service.
3119 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3121 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3122 * The build process now looks for GNU make under the default name "gmake",
3123 instead of "make" as it used to. If GNU make is not available as "gmake"
3124 on your system, you can change this default behavior by setting the
3125 GNUMAKE environment variable.
3126 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3127 can build without error under CMU CL.
3129 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3131 * There is now a partial SBCL user manual (with some new text and some
3132 text cribbed from the CMU CL manual).
3133 * The beginnings of a profiler have been added (starting with the
3134 CMU CL profiler and simplifying and cleaning up). Eventually the
3135 main interface should be through the TRACE macro, but for now,
3136 it's still accessed through vaguely CMU-CL-style functions and macros
3137 exported from the package SB-PROFILE.
3138 * Some problems left over from porting CMU CL to the new
3139 cross-compilation bootstrap process have been cleaned up:
3140 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3141 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3142 part when getting rid of the compiler *BACKEND* stuff.)
3143 ** The value of *NULL-TYPE* was screwed up, because it was
3144 being initialized before the type system knew the final
3145 definition of the 'NULL type. This screwed up several key
3146 optimizations in the compiler, causing inefficiency in all sorts
3147 of places. (I found it because I wanted to understand why
3148 GET-INTERNAL-RUN-TIME was consing.)
3149 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3150 PROCLAIM FTYPE information. Unfortunately this broke binary
3151 compatibility again, since now the forms output by DEFGENERIC
3152 to refer to functions which didn't exist in 0.6.3.
3153 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3154 can use the new (as of 0.6.3) transform for SOME into MAP into
3156 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3157 values to (INTEGER 0 1000000), so that the time code will no longer
3158 occasionally get blown up by Linux returning 1000000 microseconds
3159 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3160 its output conform to the ANSI spec. (Alas, this makes its output
3161 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3162 * A full call to MAP NIL with a single sequence argument no longer conses.
3163 * fixes to problems pointed out by Martin Atzmueller:
3164 * The manual page no longer talks about multiprocessing as though
3165 it were currently supported.
3166 * The ILISP support patches have been removed from the distribution,
3167 because as of version 5.10.1, ILISP now supports SBCL without us
3168 having to maintain patches.
3169 * added a modified version of Raymond Toy's recent CMU CL patch for
3170 EQUALP comparison of HASH-TABLE
3172 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3174 * The system still can't cross-compile itself with
3175 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3176 entails), but at least it can compile more of itself that way
3177 than it used to be able to, and various buglets which were uncovered
3178 by trying to cross-compile itself that way have now been fixed.
3179 * This release breaks binary compatibility again. This time
3180 at least I've incremented the FASL file format version to 2, so that the
3181 problem can be detected reliably instead of just causing weird errors.
3182 * various new style warnings:
3183 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3184 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3185 ** using the lexical binding of a variable named in the *FOO* style
3186 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3187 as specified by ANSI.
3188 * *RANDOM-STATE* is no longer automatically initialized from
3189 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3190 default behavior of the system is to repeat its behavior every time
3191 it's run. If you'd like to change this behavior, you can always
3192 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3193 old convention there was no comparably easy way to get the system to
3194 repeat its behavior every time it was run.
3195 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3196 FTYPE declarations has been removed, in favor of their ANSI
3197 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3198 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3199 the types of their sequence arguments can be determined at compile time.
3200 This is done through a new open code expansion for MAP which eliminates
3201 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3202 argument types can be determined at compile time.
3203 * The optimizer now transforms COERCE into an identity operation when it
3204 can prove that the coerced object is already of the correct type. (This
3205 can be a win for machine generated code, including the output of other
3206 optimization transforms, such as the MAP transform above.)
3207 * Credit information has been moved from source file headers into CREDITS.
3208 * Source file headers have been made more standard.
3209 * The CASE macro now compiles without complaining even when it has
3212 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3214 * (Note that the way that the PCL macroexpansions were rewritten
3215 to accommodate the change in DEFGENERIC below breaks binary
3216 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3217 not run under sbcl-0.6.2. Once we get out of alpha releases,
3218 i.e. hit release 1.0.0, we'll probably try to maintain binary
3219 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3220 and sbcl-1.4.4. Until then, however, it might be fairly common
3221 for maintenance releases to break binary compatibility.)
3222 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3224 * The reporting of unhandled signals has been changed to print some
3225 explanatory text as well as the report form. (Previously only
3226 the report form was printed.)
3227 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3228 it defines, so that the compiler no longer issues undefined function
3229 warnings for compiled-but-not-yet-loaded generic functions.
3230 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3231 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3232 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3233 section of the ANSI spec are supported.
3234 * The "" nickname for the "KEYWORD" package has been removed.
3235 The reader still handles symbol tokens which begin with a package marker
3236 as keywords, but it doesn't expose its mechanism for doing so in the
3237 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3238 * The system now issues STYLE-WARNINGs for contradictory TYPE
3239 proclamations. (Warnings for contradictory FTYPE proclamations would
3240 be nice too, but those can't be done usefully unless the type system
3241 is made smarter about FUNCTION types.)
3242 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3243 systematized, so that "*target-*.lisp is supposed to exist only on the
3244 target and imply that there's a related file which exists on the
3245 host, and *host-*.lisp is supposed to exist only on the host and imply
3246 that there's a related file which exists on the target. This involves a
3247 lot of renaming. Hopefully the acute confusion caused by the renaming
3248 will be justified by the reduction in chronic confusion..
3249 ** runtime-type.lisp -> early-target-type.lisp
3250 ** target-type.lisp -> late-target-type.lisp
3251 ** early-host-format.lisp -> early-format.lisp
3252 ** late-host-format.lisp -> late-format.lisp
3253 ** host-error.lisp -> misc-error.lisp
3254 ** early-error.lisp -> early-target-error.lisp
3255 ** late-error.lisp -> late-target-error.lisp
3256 ** host-defboot.lisp -> early-defboot.lisp
3257 ** code/misc.lisp -> code/target-misc.lisp
3258 ** code/host-misc.lisp -> code/misc.lisp
3259 ** code/numbers.lisp -> code/target-numbers.lisp
3260 ** code/early-numbers.lisp -> numbers.lisp
3261 ** early-host-type.lisp -> early-type.lisp
3262 ** late-host-type.lisp -> late-type.lisp
3263 ** host-typep.lisp -> typep.lisp
3264 ** load.lisp -> target-load.lisp
3265 ** host-load.lisp -> load.lisp
3266 ** host-disassem.lisp -> disassem.lisp
3267 ** host-insts.lisp -> insts.lisp
3268 ** byte-comp.lisp -> target-byte-comp.lisp
3269 ** host-byte-comp.lisp -> byte-comp.lisp
3270 ** host-signal.lisp -> signal.lisp
3271 ** host-defstruct.lisp -> defstruct.lisp
3272 ** late-target-type.lisp -> deftypes-for-target.lisp
3273 Furthermore, several other previously target-only files foo.lisp (e.g.
3274 hash-table.lisp and random.lisp) have been split into a target-and-host
3275 foo.lisp file and a target-only target-foo.lisp file, with their key type
3276 definitions in the target-and-host part, so that the cross-compiler will
3277 know more about target types.
3278 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3279 gone away. In their place are various *BACKEND-FOO* variables
3280 corresponding to the slots of the old structure.
3281 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3282 into the target SBCL has been fixed.
3283 * The chill.lisp system for loading cold code into a running SBCL
3285 * Support for the CMU CL "scavenger hook" extension has been removed.
3286 (It was undocumented and unused in the CMU CL sources that SBCL was
3287 derived from, and stale in sbcl-0.6.1.)
3288 * Various errors in the cross-compiler type system were detected
3289 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3290 (enabling various consistency checks). Many of them were fixed,
3291 but some hard problems remain, so the compiler is back to
3292 running without *TYPE-SYSTEM-INITIALIZED* for now.
3293 * As part of the cross-compiler type system cleanup, I implemented
3294 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3295 * I have started adding UNCROSS calls throughout the type system
3296 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3297 on cross-compiler input files will be able to go away, and various
3299 * CONSTANTP now returns true for quoted forms (as explicitly required
3302 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3304 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3305 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3306 (MAKE-LIST -1). In the longer term, it ought to become true
3307 that declarations are assertions even at SAFETY 1. For now, it's not
3308 quite true even at SAFETY 3, but it's at least more nearly true..
3309 (Note that this change seems to increases the size of the system by
3310 O(5%) and to decrease the speed of the compiler by 20% or more.)
3311 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3312 for the problem of printing dozens of lines of distracting information
3313 about low-level system machinery as part of the top stack frame
3314 on entry to the debugger when an undefined function was called.
3315 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3316 is not reset to COMMON-LISP-USER.
3317 * Compilation of stuff related to dyncount.lisp has been made conditional
3318 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3319 smaller. The various dyncount-related symbols have been moved into
3320 a new "SB-DYNCOUNT" package.
3321 * tty-inspect.lisp has been renamed to inspect.lisp.
3322 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3323 feature has gone away. (When we eventually port to other flavors of
3324 libc and/or Unix, we'll try to make the differences between flavors
3325 invisible at the user level.)
3326 * Various other *FEATURES* tags, and/or their associated conditionals,
3327 have been removed if obsolescent, or given better documentation, or
3328 sometimes given more-mnemonic names.
3330 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3332 * tidied up "make.sh" script
3333 * tidied up system directory structure
3334 * better "clean.sh" behavior
3335 * added doc/FOR-CMUCL-DEVELOPERS
3336 * many many small tweaks to output format, e.g. removing possibly-confusing
3337 trailing #\. character in DESCRIBE-INSTANCE
3338 * (EQUALP #\A 'A) no longer signals an error.
3339 * new hashing code, including EQUALP hashing
3340 * tidied up Lisp initialization and toplevel
3341 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3342 * command line argument processing
3343 * added POSIX-GETENV function to deal with Unix-ish environment variables
3344 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3345 terminating SBCL on EOF
3346 * non-verbose GC by default
3347 * There is no more "sbcl" shell script; the sbcl file is now the C
3348 runtime executable (just like CMU CL).
3349 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3351 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3352 symbol and package information as Lisp native symbol and package objects
3353 instead of strings naming symbols and strings naming packages. This way,
3354 whenever packages are renamed (as in warm init), debug information is
3355 transformed along with everything else.
3356 * tweaked the optimization policy declarations which control the building
3357 of SBCL itself. Now, among other things, the system no longer saves
3358 source location debugging information. (This helps two problems at once
3359 by reducing SBCL size and by keeping SBCL from trying to look for its
3360 sources -- which may not exist -- when reporting errors.)
3361 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3362 debugging and testing purposes
3363 * cleaned up printing, making the printer call PRINT-OBJECT for
3364 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3365 methods, giving nearly-ANSI behavior
3366 * converted almost all special variables to use *FOO* naming convention
3367 * deleted PARSE-TIME functionality, since it can be done portably
3368 * moved some files out of cold init into warm init
3369 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3371 * regularized formatting of source files
3372 * added an install.sh script
3373 * fixed ridiculous memory usage of cross-compiler by making
3374 compiler/alloc.lisp not try to do pooling unless it can hook
3375 itself into the GC of the cross-compilation host. Now the system
3376 builds nicely on my old laptop.
3377 * added :SB-ALLOC in target-features.lisp-expr
3378 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3379 was not implemented)
3380 * re-did condition handling and note reporting in the compiler. Notes
3381 are no longer handled by signalling conditions. Style warnings
3382 and warnings are handled more correctly and reported in such a way
3383 that it's easy to find one or the other in your output (so that you
3384 can e.g. figure out which of many problems caused COMPILE-FILE to
3386 * changed the severity of several compiler warnings from full WARNING
3387 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3388 compiler note reporting so that it doesn't use the condition system
3389 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3390 COMPILE-FILE command)
3391 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3392 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3393 no longer supported, since their implementation was deeply intertwingled
3394 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3395 * removed bogus "support" for compiler macros named (SETF FOO), and
3396 removed the compiler macro for SETF INFO (but only after making a fool
3397 of myself on the cmucl-imp mailing list by posting a bogus patch for
3398 DEFINE-COMPILER-MACRO..)
3399 * Compiled files containing forms which have side effects on the Lisp
3400 reader (such as DEFPACKAGE forms) are now handled more correctly.
3401 (Compiler queuing of top level lambdas has been suppressed by setting
3402 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3403 * deleted various currently-unused source files, e.g. gengc.lisp. They
3404 may be added back at some point e.g. when porting to other architectures,
3405 but until they are it's distracting to distribute them and to try to
3407 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3408 there were so many of them they're just distractions, and UNCROSS is
3409 known to be able to handle the current sources
3410 * moved PROFILE functionality into TRACE, so that it will be clear
3411 how the wrapping and unwrapping of functions when you profile them
3412 interacts with the wrapping and unwrapping of functions when you
3413 trace them. (Actually, the functionality isn't there yet, but at least
3414 the interface specification is there. Hopefully, the functionality will
3415 arrive with some future maintenance release.)
3416 * removed host-oops.lisp
3417 * changed signature of QUIT function to allow UNIX-CODE argument
3418 * fixed READ-SEQUENCE bug
3419 * tweaked verbose GC output so that it looks more like the progress
3420 output that ANSI specifies for functions like LOAD
3421 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3422 * added <http://sbcl.sourceforge.com> to the banner information printed by