1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
3 * new feature: timers based on Zach Beane's excellent timer package
4 * added support for the following external formats: koi8-u,
5 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
6 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
7 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
8 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
9 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
10 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
11 * incompatible change: a threaded SBCL will no longer revert to
12 non-threaded mode on non-NPTL systems, but refuse to start entirely.
13 * bug fix: interrupts are disabled until startup is complete; no
14 more sigsegvs when receiving a signal to soon
15 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
17 * bug fix: add a workaround for the memory randomization features in
18 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
19 workaround will only be in effect on systems with the proc filesystem
21 * bug fix: printing objects of type HASH-TABLE signals a
22 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
24 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
25 have been running for over 50 days. (reported by Gilbert Baumann)
26 * bug fix: the logic for getting names of functions gets less
27 confused when confronded with alternate-metaclass
28 funcallable-instances. (reported by Cyrus Harmon)
29 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
30 consistent, even on internal alternate-metaclass objects.
31 * bug fix: SB-MOP:FUNCALLABLE-STANDARD-OBJECT is now a subclass of
32 STANDARD-OBJECT, as required by AMOP.
33 * bug fix: the classes STANDARD-CLASS and
34 SB-MOP:FUNCALLABLE-STANDARD-CLASS are now compatible in the
35 SB-MOP:VALIDATE-SUPERCLASS sense; there remains a constraint about
36 finalized classes and the FUNCTION class.
37 * bug fix: the SB-MOP:METAOBJECT class is now implemented as
39 * bug fix: flush closure information collected by physical
40 environment analysis prepass before the main pass. (bug reported
42 * bug fix: compiler pack phase does not modify a hashtable, which is
43 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
45 ** bug fix: parent thread now can be gc'ed even with a live
47 ** bug fix: binding a special with PROGV to no value is not
48 the same as that symbol not having been bound (thanks to
50 ** bug fix: binding specials is thread safe (thanks to
52 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
53 and SB-SPROF do not die with 'no handler for signal XX in
54 interrupt_handle_now(..)' anymore
55 ** bug fix: WITH-TIMEOUT works with multiple running threads
56 * fixed some bugs revealed by Paul Dietz' test suite:
57 ** ENSURE-GENERIC-FUNCTION should take a method class object for
58 the :method-class keyword argument.
60 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
61 * new port: the Solaris operating system on x86 processors is now
62 mostly supported, though some rough edges in the environment
63 remain. (thanks to Daisuke Homma)
64 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
65 space for its heap. (thanks to Thiemo Seufer)
66 * optimization: SBCL on MIPS platforms now supports dynamic-extent
67 closures. (thanks to Thiemo Seufer)
68 * minor incompatible change: eof selects abort in the debugger.
69 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
70 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
72 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
73 deprecated, and will go away in a future revision of SBCL.
74 * minor incompatible change: GC-ON and GC-OFF are no longer
75 implemented with a counter, it does not matter how many times gc
77 * bug fix: discriminating functions for generic function classes
78 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
79 make invalid assumptions about method precedence order. (reported
81 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
82 platforms. (thanks to Thiemo Seufer)
83 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
84 * bug fix: correct pathname printing with printer escaping is on.
85 (thanks to Kevin Reid)
86 * bug fix: complex VOP definitions in "user-space" no longer trigger
87 package locks. (reported by Zach Beane)
88 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
89 for funcallable-instance functions, and is no different from
91 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
92 characters in the current readtable. (reported by Nicholas Neuss)
93 * bug fix: SUBTYPEP on various CONS types returns more a more
94 accurate acknowledgment of its certainty.
95 * optimizations: REMOVE-DUPLICATES now runs in linear time on
96 lists in some cases. This partially fixes bug 384.
97 * flush all standard streams before prompting in the REPL and the
99 * bug fix: signal handling and triggering gc do not conflict
100 directly anymore, in particular a high frequency sb-sprof does
101 not prevent gc from running
102 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
103 approximation for timezone and DST information between the
104 universal time epoch and the smallest negative 32-bit time_t.
105 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
106 year 1899 so long as the encoded time is non-negative.
107 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
108 the correct answer for non-constant rotate amounts when performing
109 an inline 32-bit rotation.
111 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
112 there is only one thread in the session
113 ** bug fix: memory leak for streams created in one thread and
114 written to in another
115 ** bug fix: lockup when compiled with gcc4
116 ** bug fix: race that allows the gc to be triggered when gc is
118 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
119 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
121 ** bug fix: finished threads are gc'ed properly
122 * fixed some bugs revealed by Paul Dietz' test suite:
123 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
124 has arguments with a different set of applicable methods from
125 the orignal arguments.
126 ** The type error thrown by MAP now has a correct expected-type
128 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
129 name a compiled function.
130 ** Three MISC tests where a large bignum was improperly coerced to
131 a float that couldn't represent that bignum during type
132 derivation were fixed.
133 ** SUBTYPEP can now handle types involving the negation of a
134 list-form FUNCTION type.
135 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
136 as (COMPLEX (AND RATIO (NOT FIXNUM))).
137 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
139 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
140 * New feature: Experimental support for bivalent streams: streams
141 opened with :element-type :default now allow character and binary
142 (unsigned-byte 8) I/O
143 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
144 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
146 * bug fix: no more highly sporadic "couldn't check whether ~S is
147 readable" when reading a stream and an interrupt hits in the middle
148 of a select system call
149 * compiler better recognizes complex arrays (reported by Eduardo
151 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
152 for variables on SBCL built with linkage-tables. (reported by Luis
154 * various error reporting improvements.
155 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
156 (Thanks to Thiemo Seufer)
157 * bug fix: nested reader invokations work correctly
158 * bug fix: it is possible to have simultaneous references to foreign
159 code and foreign data with the same name.
161 ** added x86-64 support
162 ** incompatible change: the threading api now works with thread
163 objects instead of thread ids
164 ** bug fix: threads are protected from signals and interruption when
165 starting up or going down
166 ** bug fix: a race where an exiting thread could lose its stack to gc
167 ** bug fix: don't halt on infinite error in threads if possible
168 ** fixed numerous gc deadlocks introduced in the pthread merge
169 ** bug fix: fixed thread safety issues in read and print
170 ** bug fix: debugger doesn't hang on session lock if interrupted at
171 an inappropriate moment
172 ** bug fix: run-program is now thread safe(r)
173 ** bug fix: inner with-recursive-lock no longer releases the mutex
174 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
175 * fixed some bugs revealed by Paul Dietz' test suite:
176 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
177 EXPECTED-TYPE slots filled.
178 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
180 ** MAKE-CONDITION accepts classes as type-designators.
181 ** COMPILE may never return NIL.
182 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
183 range before calling Unix time functions
185 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
186 * numerous signal handling fixes to increase stability
187 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
188 * minor incompatible change: we now correctly canonize default
189 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
190 as per the MOP, rather than the historical (INITARG INITFUNCTION
191 INITFORM). (reported by Bruno Haible)
192 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
193 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
194 for more information.
195 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
196 pathname is a directory pathname.
197 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
198 * dynamic space size on PPC has been increased to 768Mb. (thanks to
200 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
201 :METACLASS argument in addition to a class name. (reported by
202 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
203 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
204 (reported by Bruno Haible; patch from Gerd Moellmann)
205 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
207 * bug fix: more cleanups to the floating point exception handling on
208 x86-64 (thanks to James Knight)
209 * bug fix: the compiler does not try to constant fold calls of
210 COERCE and BIT-* functions when they return freshly constructed
211 objects. (reported by Nikodemus Siivola and Paul Dietz)
212 * optimization: improved the MIPS versions of generic (in the
213 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
214 * optimization: direct conversion of (unsigned-byte 32) to floats on
215 the PowerPC platform.
216 * optimization: structure instances with raw slots now use less
217 memory, and probably show better memory locality. (thanks to
219 * optimization: DYNAMIC-EXTENT declarations for lists and closures
220 are treated as requests for stack allocation on the x86-64,
221 Alpha-32, PPC and SPARC platforms.
222 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
223 the wrong file descriptor; implementation of SOCKET-OPEN-P.
224 (thanks to Tony Martinez)
226 ** gcing a dead thread can no longer lead to lockups
227 ** threads block signals until they are set up properly
228 ** errno is no longer shared by threads
229 ** interrupt-thread restores the eflags register on x86
230 ** fixed some lockups due to gc/thread interaction
231 * fixed some bugs revealed by Paul Dietz' test suite:
232 ** invalid dotted lists no longer raise a read error when
234 ** signal an error if a symbol that names a declaration is used
235 as the name of a type, or vice versa
236 ** allow using the (declare (typespec var*)) abbreviation for
237 (declare (type typespec var*)) with all type specifiers
238 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
239 ** always return NIL from PPRINT-POP when OBJECT is NIL
240 ** don't signal errors when pretty-printing malformed LABELS,
241 FLET or MACROLET forms
242 ** declarations in a DOLIST body are also in scope for the
244 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
246 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
249 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
250 * fixed cross-compiler leakages that prevented building a 32-bit
251 target with a 64-bit host compiler.
252 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
253 opened with :IF-EXISTS :OVERWRITE.
254 * fixed bug 281: error for an invalid qualifier in a short-form method
255 combination method is not signalled until the faulty method is called.
256 * bug fix: iteration variable type inferrer failed to deal with open
257 intervals. (reported by Alan Shields)
258 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
259 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
260 or more runtime options were provided to the sbcl binary.
261 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
262 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
263 * optimization: there's now a fast-path for fixnum arguments in the
264 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
265 * optimization: the code generated on x86-64 is more compact thanks
266 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
267 * bug fix: floating-point exception handling now partly works on
268 x86-64. (Thanks to James Knight)
269 * improvement to the MIPS backend from Thiemo Seufer: C-style
270 64-bit long long arguments and return values to alien functions
272 * fixed some bugs revealed by Paul Dietz' test suite:
273 ** the type-error signalled from WARN has a filled-in DATUM slot.
274 ** the type-error required when a stream is not associated with
275 a file has the stream as its datum.
276 ** type-errors on single-floats on x86-64 no longer have
277 :INVALID-OBJECT as the datum
278 ** the type-errors signalled for invalid function names now have
279 a correct expected type
280 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
281 for structures defined with a :TYPE. Documentation strings for
282 typed structures are no longer immediately discarded
283 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
284 broadcast streams and synonym streams. FILE-LENGTH now also works
285 on broadcast streams.
287 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
288 * incompatible change: the --noprogrammer option, deprecated since
289 version 0.7.5, has been removed. Please use the equivalent
290 --disable-debugger option instead.
291 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
293 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
294 inoperational for a while now) has been completely removed.
295 * null lexical environments are now printed as #<NULL-LEXENV>,
296 significantly reducing the amount of clutter in typical backtraces.
297 * documentation on weak pointers, finalization, and after GC hooks
298 has been added to the manual.
299 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
300 as other specialized array types, is much faster. SUBSEQ and
301 COPY-SEQ on such arrays have also been sped up.
302 * optimization: EQL is now more efficient when at least other argument
303 is known to be of type (OR FIXNUM (NOT NUMBER)).
304 * fixed bug: the runtime is now less vulnerable to changes in the
305 size of the SBCL object on OS X, and virtual memory is reserved for
306 all spaces that need to be at a fixed address.
307 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
308 * fixed bug: finalizers and after GC hooks that cause consing are now
310 * fixed bug: compiler error messages and summaries are now printed to
311 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
312 * fixed inference of the upper bound of an iteration variable.
313 (reported by Rajat Datta).
314 * fixed bug 211e: calling local functions with duplicated constant
315 keyword argument no longer causes a bogus style warning about an
317 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
318 ftype declaration to be disregarded. (reported by Dave Roberts)
319 * fixed bug 373: caused by erronous compilation of references to alien
320 variables in the runtime on ppc/darwin.
321 * fixed bug 376: CONJUGATE type deriver.
322 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
323 (reported by Baughn on #lisp)
324 * a cross-compiler bug on non-x86 platforms has been identified and
325 fixed. (thanks to Bruno Haible)
326 * improvements to the MIPS runtime code for increased stability.
327 (thanks to Thiemo Seufer)
328 * increased the maximimum compact environment size to allow
329 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
330 * partial workaround for bug 135: don't name the function we're
331 calling for hairy cases of EVAL, so as not to accumulate one environment
332 entry per call to EVAL. (reported by Kevin Reid)
333 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
334 * optimization: type testing for non-vector arrays should be faster.
335 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
337 * bug fix: setting 31st element of a bit vector to zero did not work
339 * bug fix: redefining a class definition which failed due to a
340 previous accessor / function clash now works (but see BUGS entry
341 #380 for more problems in this area). (thanks to Zach Beane)
342 * the long form of DEFINE-METHOD-COMBINATION disables method group
343 checking when given a single method group with pattern *. (thanks
345 * on x86 compiler supports stack allocation of results of simple
346 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
347 * support for building and running on Mac OS X 10.4 "Tiger" has been added
348 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
349 * fixed some bugs related to Unicode integration:
350 ** the restarts for recovering from input and output encoding
351 errors only appear when there is in fact such an error to
353 * fixed some bugs revealed by Paul Dietz' test suite:
354 ** MISC.549 and similar: late transformation of unsafe type
355 assertions into derived types caused unexpected code
357 ** SCALE-FLOAT type deriver is less wrong.
358 ** type derivers for EXP, LOG and similar functions compute result
359 types for complex arguments better.
360 ** (MISC.563) CONJUGATE type deriver works for very restricted
362 ** out-of-line type testers for character strings are available.
363 ** EQUAL compiler transform understands specialness of objects
365 ** accessing double-floats stored on the stack now works on x86-64.
366 ** debugger internals could sometimes create invalid lispobjs,
367 resulting in GC crashes.
368 ** MISC.548: type check weakening can convert required type into
370 ** initialization forms for bindings are not in scope of free special
373 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
374 * incompatible change: thread support for non-NPTL systems has
375 been removed - locking is buggy and unreliable. A threaded
376 SBCL build will now warn at startup and refuse to create
377 new threads, unless futex support is detected in the kernel
378 * incompatible change: the top level REPL now has only an ABORT
379 restart associated with it, not TOPLEVEL and ABORT as it used to.
380 TOP and TOPLEVEL are now available as debugger commands for
381 returning to the top level.
382 * incompatible change: forms evaluated in the REPL now use the
383 global optimization policy.
384 * incompatible change: user- and system-initialization files are
385 no longer processed with LOAD, but by READ and EVAL; hence the
386 global optimization policy, startup package, readtable, etc,
388 * The .fasl file format number has been incremented because of
389 various incompatible changes.
390 * internal entry point details and argument counts no longer appear
391 in backtraces unless explicitly requested by setting
392 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
393 * built-in and standard functions no longer have names like "top
394 level local call to FOO".
395 * fixed bug 32: functions defined in non-null lexical environments
396 now have more legible printed representation
397 * fixed bug 33: functions defined in non-null lexical environemnts
398 are now more amenable to inspection by INSPECT.
399 * workaround for bug 354: XEPs no longer appear in backtraces unless
400 explicitly requested.
401 * fixed bug: receiving the signal which results from attempting to
402 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
403 SIGBUS on other *BSDs) on architectures where the C stack is also
404 the Lisp stack (x86 and x86-64) from foreign code no longer leads
405 to debugger confusion or wild instability. (reported by Cheuksan
407 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
408 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
409 (reported by Lutz Euler)
410 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
411 compiled code. (reported by Paul Dietz)
412 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
413 specialized parameter to a method to something that is not TYPEP
414 the specializer is now possible.
415 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
416 face of package deletion.
417 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
418 pathname. (reported by David Tolpin on c.l.l)
419 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
420 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
421 STRING-STREAM and FILE-STREAM, respectively.
422 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
423 than previously; in addition, its test suite is now run on build.
424 (thanks to Luke Gorrie)
425 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
426 * optimization: added a immediate representation for single-floats
428 * fixed some bugs revealed by Paul Dietz' test suite:
429 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
430 ** attempting to create a package with a colliding nickname causes
431 correctable errors to be signalled.
432 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
433 ** several x86-64 backend bugs related to sign-extension of immediate
436 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
437 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
438 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
440 * fixed loading of multiply forward-referenced layouts.
441 (thanks to Cheuksan Wang)
442 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
443 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
444 (see COMPILE-FILE documentation for details of :PRINT option)
445 * fixed bugs 19 and 317: fixed-format floating point printing is
446 more accurate. This also fixes a bug reported by Adam Warner
447 related to the ~@F format directive.
448 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
450 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
451 dispatch table if the from-char is a dispatch macro character.
452 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
453 by the wordsize no longer ignore the last word. (reported by Lutz
455 * fixed bug in type checking of dynamic-extent variables. (reported
457 * optimization: sequence traversal functions use their freedom to
458 coerce function designators to functions.
459 * optimization: code with many calls to CLOS methods specialized on
460 CLOS classes has had redundant type checks removed.
461 * optimization: added declarations to speed up operations that access
462 the internal character database (for example STRING-UPCASE)
463 * optimization: comparison operations between floats and sufficiently small
464 fixnums no longer create extra rationals
465 * fixed some bugs related to Unicode integration:
466 ** portions of multibyte characters at the end of buffers for
467 character-based file input are correctly transferred to the
468 start of the buffer at the next read.
469 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
470 passing it through to OPEN.
471 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
472 argument given to internal calls to OPEN.
473 ** the built-in comment readers (introduced by character sequences
474 ";" and "#|") are more forgiving to encoding errors; they will
475 STYLE-WARN and then attempt to resync the stream at a character
476 boundary. (thanks to Teemu Kalvas)
477 * fixed some bugs in the x86-64 port:
478 ** Negative short int return values from c-calls are sign-extended
480 ** The stack is aligned to 16-bytes for c-calls, as required by
481 the ABI. (thanks to Cheuksan Wang)
482 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
483 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
484 for enums. (thanks to Vincent Arkesteijn)
485 ** Multiple small optimizations and bugfixes for floating point
487 * fixed some bugs revealed by Paul Dietz' test suite:
488 ** Space, Tab, Linefeed, Return and Page have the invalid
489 secondary constituent character trait.
490 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
492 ** WITH-INPUT-FROM-STRING should only update the index place on
494 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
496 ** Bit-array operations (BIT-AND and similar) worked incorrectly
497 with one-dimensional arrays with fill pointers.
498 ** TYPE-OF failed on a complex with an integer realpart and a
500 ** compiler failure during type inference for the code of form
501 (IF (EQL X (THE ...)) ...) (MISC.535).
503 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
504 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
505 platform. The port supports 61-bit fixnums, large memory spaces
506 and reloading shared object files.
507 * enhancement: saving cores with foreign code loaded is now
508 supported on ppc/Darwin in addition to the previously supported
510 * enhancement: the statistical profiler now walks deeper into the
511 call stack for more meaningful call-graphs and accrued time
512 reports (x86/x86-64 only). It also now reports time spent in
514 * enhancement: it is now possible to trace most individual methods
515 of a generic function in addition to tracing the generic function
517 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
519 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
520 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
521 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
522 produces an error. (thanks to Vincent Arkesteijn)
523 * bug fix: NAMESTRING on pathnames with :WILD components in their
524 directories works correctly. (thanks to Artem V. Andreev)
525 * fixed bug 125: compiler preserves identity of closures. (reported
527 * bug fixed: functions with &REST arguments sometimes failed with
528 "Undefined function" when compiled with (DEBUG 3). (reported by
530 * bug fix: overflow during compiling of setting element of a bit
531 vector with constant index and value. (reported by Timmy Douglas)
532 * build fix: fixed the dependence on *LOAD-PATHNAME* and
533 *COMPILE-FILE-PATHNAME* being absolute pathnames.
534 * on x86 compiler partially supports stack allocation of dynamic-extent
536 * GO and RETURN-FROM do not check the extent of their exit points
537 when compiled with SAFETY 0.
538 * fixed some bugs related to Unicode integration:
539 ** encoding and decoding errors are now much more robustly
540 handled; it should now be possible to recover even from invalid
541 input or output to the terminal. (thanks to Teemu Kalvas)
542 ** provided a first cut at implementing STRING-TO-OCTETS and
543 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
544 ** altered the SB-MD5 contributed module slightly, changing the
545 interface just enough for it to be supportable for builds where
546 lisp characters are not eight bits.
547 * fixed some bugs revealed by Paul Dietz' test suite:
548 ** the FORMATTER-generated functions for ~V[ conditionals require
549 the correct number of arguments.
550 ** READ-FROM-STRING returns the mandated second value when applied
551 to displaced strings.
552 ** the #\Rubout and #\Backspace characters are treated as invalid
553 constituent characters by the tokenizer.
555 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
556 * new feature: reloading changed shared object files with
557 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
558 * new feature: references to foreign variables and functions
559 can now be compiled and loaded before the corresponding shared
560 object file is loaded, as long as the foreign definitions are
561 available at runtime.
562 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
563 supported. (thanks to Dan Debertin)
564 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
565 just keywords. (thanks to Vincent Arkesteijn)
566 * fixed bug #331: structure-class instances corresponding to
567 DEFSTRUCT forms are now created eagerly.
568 * fixed bug #345: backtraces from calls to undefined functions work
569 on x86 as well. Related bug #61 is now also partially fixed on x86
570 (backtraces from throws to unknown catch tags.)
571 * bug fix: lambda-list parsing is now stricter vrt. order and number
572 of lambda-list keywords.
573 * bug fix: as specified by AMOP, an error is signalled if a
574 class-option appears multiple times in a DEFCLASS form. (reported
576 * bug fix: RANDOM can be compiled when the compiler derives the type
577 of its numeric argument as a disjoint set of small integers.
578 (reported by Paul Dietz)
579 * bug fix: starting a core saved with shared objects loaded when
580 those objects are not available no longer causes threaded SBCL to
581 hang. (reported by Sean Ross)
582 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
584 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
585 argument, which must always be NIL. (reported by Kalle Niemitalo)
586 * bug fix: printing 1.0d+23 no longer results in an error.
587 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
588 * bug fix: structure slot setters preserve evaluation order. (thanks
590 * fixed some bugs related to Unicode integration:
591 ** RUN-PROGRAM can allow its child to take input from a Lisp
592 stream. (reported by Stefan Scholl)
593 ** ASDF-INSTALL successfully downloads .tar.gz files in all
594 locales. (reported by Ken Causey)
595 * fixed some bugs revealed by Paul Dietz' test suite:
596 ** INCF, DECF and REMF evaluate their place form as specified in
598 ** FORMATTER expands ~{ ~} iteration directives with V or #
599 parameters correctly.
600 ** FORMATTER deals with the ~@[ ~] conditional directive where the
601 consequent uses no arguments correctly.
602 ** the system has a partial understanding of the (COMPLEX RATIO)
604 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
607 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
608 * new feature: a build-time option (controlled by the :SB-UNICODE
609 keyword feature, enabled by default) for building the system with
610 support for the entire 21-bit character space defined by the
612 * new feature: the system now has rudimentary external-format
613 support; the primary user-visible change at this time is that
614 characters with the high bit set (such as lower-case-e-acute) will
615 print correctly to a terminal in a UTF-8 environment.
616 * minor incompatible change: BASE-CHAR no longer names a class;
617 however, CHARACTER continues to do so, as required by ANSI.
618 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
619 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
620 has been moved to the SB-EXT package (temporarily re-exported from
622 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
623 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
625 * fixed bug #351: better error-handlind and reporting for malformed
627 * fixed bug #350: bignum-printing is now more memory-efficient,
628 allowing printing of very large bignums, eg. (expt 2 10000000).
629 (reported by Bruno Haible)
630 * fixed bug #302: better primitive-type selection for intersection
632 * fixed bug #308: non-graphic characters now all have names, as
633 required. (reported by Bruno Haible)
634 * bug fix: redefining a class with different superclasses now correctly
635 removes it from the direct-subclasses of its previous superclasses.
636 (reported by David Morse)
637 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
638 new value now works. (reported by Bruno Haible)
639 * bug fix: correct canonicalization of multiple non-standard slot
640 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
641 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
642 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
643 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
644 now exists, an signals an error.
645 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
646 during macroexpansion for non-iteration variables as well. (reported
647 by Bruno Haible for CMUCL)
648 * bug fix: Cyclic structures and unprintable objects in compiler
649 messages no longer cause errors. (reported by Bruno Haible)
650 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
651 and READ-FROM-STRING all now return a primary value of NIL if
652 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
653 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
654 (reported by Bruno Haible for CMUCL)
655 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
656 do not designate a condition. (reported by Bruno Haible for
658 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
659 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
660 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
661 permitting resolution in favour of any of the conflicting symbols.
662 (reported by Bruno Haible for CMUCL)
663 * bug fix: EQUAL compiler optimizations is less aggressive on
664 strings which can potentially compare true despite having distinct
665 specialized array element types.
666 * bug fix: unit enumerations can be defined without dividing by
667 zero. (thanks to Vincent Arkesteijn)
668 * FORMAT compile-time argument count checking has been enhanced.
669 (report from Bruno Haible for CMUCL)
670 * a partial workaround for the bug 262: the compiler does not try to
671 inline-expand a local function doing RETURN-FROM from a deleted
672 BLOCK. (thanks to Peter Denno for the bug report and to David
673 Wragg for the simple test case)
674 * fixed some bugs revealed by Paul Dietz' test suite:
675 ** RENAME-PACKAGE allows all package designators as new package
677 ** constraint propagation and lambda variable substitution are
678 more cautious in dealing with partially deleted code.
679 ** compiler sometimes lost reoptimization passes.
680 ** CERROR, given a condition as condition designator, passes its
681 remaining arguments to the continue format control without
683 ** Case-altering FORMAT directives work correctly on non-ASCII
685 ** The REAL type specifier handles bounds outside the floating
686 point ranges without signalling FLOATING-POINT-OVERFLOW.
687 ** Functions with IR1-transformations can create intercomponent
688 references to global functions.
689 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
691 ** FORMAT ~R treats a nil value for its first parameter correctly.
693 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
694 * enhancement: saving cores with foreign code loaded is now
695 supported on x86/NetBSD and sparc/Linux in addition to the previously
697 * bug fix: on some platforms repeated installations caused multiple
698 copies of HTML documentation to be installed -- should not happen
699 any more. (reported by Stefan Scholl)
700 * bug fix: parsing self-recursive alien record types multiple times
701 no longer causes infinite recursion. (reported by Thomas F. Burdick,
702 original patch by Helmut Eller for CMUCL)
703 * bug fix: stack-exhaustion detection works now on NetBSD as well.
704 (thanks to Richard Kreuter)
705 * bug fix: defining classes whose accessors are methods on existing
706 generic functions in other (locked) packages no longer signals
707 bogus package lock violations. (reported by François-René Rideau)
708 * bug fix: special variables as DEFMETHOD parameters no longer have
709 associated bogus type declarations. (reported by David Wragg and
711 * bug fix: read-write consistency on streams of element-type
712 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
713 * bug fix: redefiniton of the only method of a generic function with
714 no DEFGENERIC no longer emits a full WARNING. In addition,
715 redefinition of generic functions with no DEFGENERIC to an
716 incompatible lambda list now signals an error. (thanks to Zach
718 * bug fix: DEFGENERIC now works even when there's a function of the
719 same name in an enclosing lexical environment. (thanks to Zach
721 * fixed compiler failure, caused by instrumenting code during
722 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
723 * optimization: added loop analysis and improved register allocation
724 * fixed some bugs revealed by Paul Dietz' test suite:
725 ** POSITION on displaced vectors with non-zero displacement
726 returns the right answer.
727 ** (SIMPLE-STRING) is a valid type specifier for sequence
729 ** *PRINT-LEVEL* handling for slotless structures is pedantically
731 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
732 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
734 ** The justification version of the FORMAT ~< directive treats
735 non-zero minpad parameter correctly.
737 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
738 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
739 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
740 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
741 the supported interface.
742 * new feature: Single-stepping of code compiled with DEBUG 2 or
743 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
744 * new feature: saving cores with foreign code loaded is now
745 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
746 Timothy Moore's work for CMUCL)
747 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
748 parameters to * instead of NIL if no initform is supplied.
749 (reported by Johan Bockgaard)
750 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
751 correctly when FUNCALL appears as the car of the form. Note:
752 despite this FUNCALL forms are not currently subject to
753 compiler-macro expansion. (port of Raymond Toy's fix for the
754 same from CMUCL, reported by Johan Bockgaard)
755 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
756 (thanks for Teemu Kalvas)
757 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
758 case instead of terminating the iteration (thanks for Julian
759 Squires, Sean Champ and Raymond Toy)
760 * bug fix: incorrect expansion of defgeneric that caused a style
761 warning. (thanks for Zach Beane)
762 * bug fix: slot accessor effective method computation works properly
763 for classes with multiple non-standard applicable methods on
764 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
765 * on x86 compiler supports stack allocation of results of LIST and
766 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
767 CMUCL implementation by Gerd Moellmann)
768 * fixed some bugs revealed by Paul Dietz' test suite:
769 ** FORMAT strings with both the ~<~:;~> form of the justification
770 directive and pretty-printing directives cause an error.
772 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
773 * incompatible change: the internal functions
774 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
775 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
776 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
777 instead of the old functions.
778 * new feature: on platforms where "dladdr" is available foreign
779 function names now appear in backtraces. (based on Helmut Eller's
781 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
782 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
784 * documentation: ASDF manual is now distributed with SBCL.
785 * bug fix: SBCL can now load its contributed modules with REQUIRE
786 even if the system-provided entries have been removed from
787 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
789 * bug fix: backtraces involving undefined functions or assembly
790 routines are more informative. (thanks to Brian Downing)
791 * bug fix: mutually referent alien structures now work correctly.
792 (reported by Rick Taube)
793 * bug fix: structures defined by WITH-ALIEN can be referred to
794 within other definitions in the same WITH-ALIEN.
795 * bug fix: division operators (MOD, TRUNCATE and the like) with
796 constant zero divisors and integer dividends no longer generate
798 * bug fix: provide default methods for INPUT-STREAM-P and
799 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
800 * bug fix: improve the detection and resolution of MOP metacycles
801 (where effective method computation depends on the generic function
802 having its effective method computed). (reported by Bruno Haible)
803 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
804 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
805 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
806 * bug fix: pathnames with relative directory components can be
807 represented relative to default pathnames.
808 * optimization: in taking the GCD of bignums, reduce the two bignums
809 to approximately the same size (using Euclid's algorithm) before
810 applying the more sophisticated binary GCD. (thanks to Juho
812 * optimization: COUNT on bitvectors now operates word-at-a-time.
813 * optimization: ASH with a positive, but not necessarily constant,
814 (leftwards) shift, when in a modular context, is compiled to a
816 * fixed some bugs revealed by Paul Dietz' test suite:
817 ** FORMAT variable parameters ("~V<char>") are defaulted properly
818 if the corresponding argument is NIL.
819 ** FORMAT directives accepting a minpad parameter treat negative
821 ** literal commas as character parameters to FORMAT directives are
823 ** literal spaces directly after ~<Newline> directives within a
824 format-logical-block (~:< ~@:>) do not induce :FILL-style
825 conditional newlines.
826 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
827 as specified: it no longer includes conditional newlines).
828 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
830 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
831 :PER-LINE-PREFIX argument does not evaluate to a string.
833 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
834 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
835 the manual for details; for now, package locks can be disabled by
836 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
837 if no major problems are found then it is likely that they will be
838 compiled in unconditionally.
839 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
840 unsupported operators on all platforms. To load a shared library
841 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
842 object file, link it into a shared library outside of SBCL and
843 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
845 * bug workaround: ROOM T and the GC were not working together
846 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
847 reporting facility aren't preserved. That reporting has been
848 disabled, so now until and unless someone figures out how to make it
849 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
850 * minor incompatible change: as threatened around sbcl-0.8.0, the
851 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
852 The symbols are also exported from SB-PCL for backwards
853 compatibility, but more so than before SB-PCL should be treated as
854 an implementation-internal package.
855 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
857 * the SB-SPROF contrib now works on (most) non-x86 architectures.
858 It is known as of this release not to work on the Alpha, however.
859 * fixed bug #167: errors signalled due to illegal syntax in method
860 bodies are now more legible.
861 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
862 designators and can hence be used with TYPEP.
863 * fixed bug #333: CHECK-TYPE now ensures that the type error
864 signalled, if any, has the right object to be accessed by
865 TYPE-ERROR-DATUM. (reported by Tony Martinez)
866 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
867 5.1.2.3 for multiple-value place subforms. (reported by Kalle
869 * fixed bug #334: programmatic addition of slots using specialized
870 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
871 and :ALLOCATION :CLASS slots.
872 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
873 float exponents. (rereported by Peter Seibel)
874 * fixed a bug: #\Space (and other whitespace characters) are no
875 longer considered to be macro characters in standard syntax by
877 * fixed bug: initialization of condition class metaobjects no longer
878 causes an instance of the condition to be created. (reported by
880 * fixed bug: it is now possible to have slots such that
881 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
882 description is neither :INSTANCE nor :CLASS.
883 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
884 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
885 SLOT-VALUE-USING-CLASS), and no longer causes errors with
886 non-standard SLOT-DEFINITION-ALLOCATION values.
887 * fixed bugs: various race conditions handling exiting threads.
888 CL-PPCRE's thread torture test now passes.
889 * fixed arguably-a-bug: GC time proportional to number of threads in
890 system even when most of them are idle
891 * optimization: improved performance of BIT and SBIT on bit-vectors.
892 * .fasl file incompatibility: The fasl file version number has
893 been incremented because of changes associated with package locks.
895 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
896 * minor incompatible change: the system no longer provides
897 optimization hints (conditions which are TYPEP
898 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
899 mode; these hints are still emitted when higher SPEED optimization
901 * new contrib module: a sampling profiler (profiling by statistical
902 sampling, rather than by instrumenting functions) is available as
903 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
904 * the behaviour of the standard function ED is now customizeable by
905 third parties through a hook variable: see ED's documentation
906 string for information on the protocol.
907 * the compiler no longer emits efficiency notes for (FUNCALL X)
908 when the type of X is uncertain under default optimization
910 * fixed bug 276: mutating a binding of a specialized parameter to a
911 method to something that is not TYPEP the specializer is now
913 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
914 exhibit the required behaviour on the x86 platform. (thanks to
915 Peter van Eynde, Eric Marsden and Bruno Haible)
916 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
917 even for difficult arguments. (reported by Peter Graves)
918 * fixed bug 141a: the backquote printer now descends quoted
920 * fixed another bug in backquote printing: no more destructive
921 modification of the form's list structure. (reported by Brian
923 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
924 move between its address being taken and the call to
925 interrupt_thread, fixing a crashing race condition.
926 * the SB-POSIX contrib implementation has been adjusted so that it
927 no longer exhibits ridiculously poor performance when constructing
928 instances corresponding to C structs.
930 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
931 * minor incompatible change: the sb-grovel contrib now treats C
932 structures as alien (in the sense of SB-ALIEN) objects rather than
933 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
934 has implications for memory management of client code
935 (sb-grovel-returned objects must now be manually managed) and for
936 type safety (alien objects now have full types).
937 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
938 used to control emission of compiler diagnostics, rather than the
939 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
940 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
941 quality should be considered deprecated.
942 * install.sh now installs the user manual as well
943 * (not quite a new documentable feature, but worth considering in
944 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
945 beginnings of a semantically meaningful condition hierarchy is
946 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
948 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
949 designator as the defaults argument.
950 * fixed bug: Displaced arrays whose displaced-to array has become
951 too small now cause ARRAY-DIMENSION to signal an error, providing
952 for safer bounds-checking. (reported by Bruno Haible)
953 * fixed bug: DEFCLASS slot definitions with identical :READER and
954 :WRITER names now signal a reasonable error. (reported by Thomas
956 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
958 * fixed bug: Invalid slot specification errors now print correctly.
959 (thanks to Zach Beane)
960 * fixed bug 320: Shared to local slot value transfers in class
961 redefinitions now happen corrently from superclasses as
962 well. (reported by Bruno Haible)
963 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
965 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
966 improper lists correctly. (reported by Bruno Haible)
967 * fixed bug 313: source-transform for <fun-name> was erroneously
968 applied to a call of a value of a variable with name <fun-name>.
969 (reported by Antonio Menezes Leitao)
970 * fixed bug 307: The obsolete instance protocol ensures that
971 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
972 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
973 remove dead unknown-values globs from the middle of the stack.
974 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
975 easier to resume long computations after using *BREAK-ON-SIGNALS*
976 to diagnose and fix failures. (thanks to Nikodemus Siivola)
977 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
978 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
979 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
980 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
981 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
983 * fixed a (fairly theoretical) bug in string printing: if
984 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
985 does not have array-element-type equal to the most general string
987 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
988 function name. (thanks to Nikodemus Siivola)
989 * fixed bug: compile-time format string checker failed on
990 non-closed ~{. (reported by Thomas F Burdick)
991 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
992 don't warn on legal loop code involving more than one aggregate
993 boolean. (thanks to Nikodemus Siivola)
994 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
995 in keyword position to MAKE-PACKAGE was not being reported as
997 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
998 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1000 * fixed bugs 280 and 312: the checking for multiple definitions in a
1001 file is less likely to become confused by uses of inline
1003 * fixed bug: the #S reader macro performs the keyword coercion
1004 specified for slot names. (reported by Kalle Niemitalo)
1005 * fixed bug: lambda lists may contain symbols whose names start with
1006 & but are not lambda-list-keywords; their occurrence triggers a
1008 * fixed bug 321: define-method-combination argument lambda lists do
1009 not cause invalid code to be generated when &optional or &aux
1010 variables have default values. (reported by Bruno Haible)
1011 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1012 have CLOS classes; this appears also to have fixed reported
1013 instability in stack exhaustion detection.
1014 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1015 MUFFLE-WARNING when no associated restart is present is now
1017 * optimization: rearranged the expansion of various defining macros
1018 so that each expands into only one top-level form in a
1019 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1021 * optimization: used a previously little-used slot in symbols to
1022 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1024 * fixed some bugs revealed by Paul Dietz' test suite:
1025 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1027 ** Fixed an optimization bug related to inheritance of initforms
1028 from local to shared slots.
1029 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1030 ** CAST optimizer forgot to flush argument derived type.
1031 ** print/read consistency on floats is now orders of magnitude
1032 more likely. (thanks also to Bruno Haible for a similar report
1034 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1035 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1036 for objects of type REAL. Make it so.
1037 ** SXHASH is specified (infelicitously) to respect similarity,
1038 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1039 it so. (thanks to Markus Ziegler)
1040 ** on the Alpha, the compiler succeeds in compiling functions
1041 returning a known number of arguments greater than 63.
1042 ** fixed handling of invalid NIL arguments in keyword position
1044 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1046 ** provided a readably-printable representation for RANDOM-STATE
1048 ** ensured that pathnames, where they have a namestring, always
1049 print using #P"..." syntax.
1051 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1052 * Support for the forthcoming 2.0 version of the NetBSD kernel
1053 running on x86 hardware has been added. (thanks to Perry
1054 E. Metzger most immediately, and others for their past work)
1055 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1056 current release version; SBCL's OpenBSD support had been broken
1057 since about the time of OpenBSD's switch to ELF binary format.)
1058 * [placeholder for DX summary]
1059 ** user code with &REST lists declared dynamic-extent, under high
1060 speed or space and low safety and debug optimization policy.
1061 * The manual has been converted to Texinfo format and the debugger
1062 chapter from the cmucl manual has been added.
1063 * A facility has been added to extract documentation strings from
1064 sbcl and store them as Texinfo-formatted snippets for inclusion in
1065 the manual (via Texinfo's @include directive)
1066 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1067 the test case to Dave Roberts)
1068 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1069 pointers. (reported by Sean Ross)
1070 * bug fix: PROFILE output is printed nicely even for large numerical
1071 values. (thanks to Zach Beane)
1072 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1073 greater than 32 handle EOF correctly.
1074 * bug fix: on X86 an immediate argument of the IMUL instruction is
1075 correctly printed in disassembly. (thanks to Lutz Euler)
1076 * bug fix: class slots in redefined classes preserve their old
1077 values. (thanks to Bruno Haible and Nikodemus Siivola)
1078 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1079 succeeds. (reported by Marco Baringer)
1080 * fixed some bugs revealed by Paul Dietz' test suite:
1081 ** READ-SEQUENCE now works on ECHO-STREAMs.
1082 ** RATIONALIZE works more according to its specification. (thanks
1085 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1086 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1087 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1088 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1089 should matter to you only if you rebind the printer control
1090 variables and then find you want different bindings in the
1091 debugger than in the ordinary execution of your program.)
1092 * The runtime build system has been tweaked to support building
1093 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1094 assembler and linker. (thanks to Nikodemus Siivola)
1095 * Unbound, undefined, undeclared variables now trigger full
1096 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1097 more often programmer error than deliberate exploitation of undefined
1099 * optimization: the hash algorithm for strings has changed to one
1100 that is less vulnerable to spurious collisions. (thanks to Juho
1102 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1103 needless bounds checking. (thanks to Juho Snellman)
1104 * optimization: implemented multiplication as a modular
1105 (UNSIGNED-BYTE 32) operation on the PPC backend.
1106 * fixed some bugs revealed by Paul Dietz' test suite:
1107 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1109 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1110 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1111 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1113 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1115 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1117 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1118 their output stream on EOF from read.
1119 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1120 have been read to end-of-file.
1121 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1123 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1124 description of determination of which consecutive characters
1126 ** Printing the "Space" character with escaping on now yields "#\\ ",
1127 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1128 ** Reading floating-point numbers with *READ-BASE* set to a number
1129 less than 10 works correctly.
1130 ** Reading floating-point numbers with *READ-BASE* set to a number
1131 more than 10 works correctly.
1132 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1133 the readtable currently in effect.
1135 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1136 * minor incompatible change: parsing of namestrings on a physical
1137 (Unix) host has changed; numbers after the final #\. in a
1138 namestring are no longer interpreted as a version field. This is
1139 intented to be largely invisible to the user, except that the
1140 meaning of the namestring "*.*.*" has changed: it now refers to a
1141 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1142 should usually be replaced by
1143 (make-pathname :name :wild :type :wild :version :wild)
1144 with the added benefit that this is more likely to be portable.
1145 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1146 option to OPEN now signals an error if the file being opened
1147 exists; this may have an impact on existing code.
1148 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1149 signals received, on the PowerPC platforms (both Linux and
1150 Darwin). (thanks to Pierre Mai for pointing out the location of
1152 * several fixes on OS X: The system now builds and runs cleanly on
1153 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1154 (thanks to Brian Mastenbrook)
1155 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1156 second-resolution: integer multiples of 1/3600 between -24 and 24.
1157 (thanks to Vincent Arkesteijn)
1158 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1159 type when called with 1 argument; PEEK-CHAR checked type of
1160 PEEK-TYPE only after having read first character from a
1161 stream. (reported by Peter Graves)
1162 * bug fix: the garbage collector now has much better locality
1163 behaviour, and in particular no longer treats all memory as being
1164 exhausted when many small objects point to each other in a deeply
1166 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1167 recognized as being TYPEP their class.
1168 * bug fix: the PUSHNEW documentation string has been corrected.
1169 (thanks to Vincent Arkesteijn)
1170 * bug fix: defaulting of the value for the last of an atypically
1171 large number of multiple values being bound was not being
1172 performed correctly on the Alpha or PPC platforms
1173 * optimization: implemented multiplication as a modular
1174 (UNSIGNED-BYTE 32) operation on the x86 backend.
1175 * optimization: SEARCH on simple-base-strings can now be open-coded.
1176 (see also contrib/compiler-extras.lisp for inspiration for
1177 teaching the compiler about the Boyer-Moore algorithm).
1178 * value, returned by MAX (and MIN) called with several EQUALP, but
1179 not EQL, arguments now does not depend on compiler settings.
1180 * fixed some bugs revealed by Paul Dietz' test suite:
1181 ** in stack analysis liveness information is propagated from
1182 non-local entry points.
1183 ** pathwise CAST removing failed when the CAST node did not start
1185 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1186 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1188 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1189 host is already defined.
1190 ** RENAME-FILE works on streams instead of signalling an internal
1192 ** PEEK-CHAR uses the current readtable when determining whether
1193 or not a character is whitespace.
1194 ** MERGE-PATHNAMES handles the case when the pathname does not
1195 specify a name while the default-pathname specifies a version.
1196 ** Pathnames now stand a better chance of respecting print/read
1198 ** Attempting to use standardized file system operators with a
1199 pathname with invalid :DIRECTORY components signals a
1201 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1203 ** DIRECTORY on logical pathnames is more correct.
1204 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1205 signal a TYPE-ERROR if their argument is not a stream.
1206 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1207 designator argument does not designate a stream.
1208 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1209 examining the synonym.
1210 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1212 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1213 element-type larger than ([UN]SIGNED-BYTE 32).
1215 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1216 * When built with the :SB-FUTEX feature, threaded builds now take
1217 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1218 for faster/more reliable mutex and condition variable support.
1219 * Incompatible change (but one you probably shouldn't have been using
1220 anyway): the interface and code for arbitrating between multiple
1221 threads in the same user session has been redesigned.
1222 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1223 error when called without an explicit environment argument.
1224 (thanks to Nikodemus Siivola)
1225 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1226 confused about their position. (thanks to Adam Warner and Gerd
1228 * bug fix: Pretty printing backquoted forms with unquotations in the
1229 argument list position of various code constructs such as LAMBDA
1230 now works correctly. (reported by Paul Dietz)
1231 * bug fix: Pretty printing unquotations no longer loses all
1232 stream position information.
1233 * optimization: performance of string output streams is now less
1234 poor for multiple small sequence writes.
1235 * optimization: performance of CSUBTYPEP in the presence of complex
1236 expressions involving CONS and NOT many times has been improved.
1237 (reported by Paul Dietz)
1238 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1240 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1241 streams for the REPL, for the debugger, and so on. (thanks to
1243 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1244 * fixed some bugs revealed by Paul Dietz' test suite:
1245 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1246 is ignored if it is too small, rather than propagating through
1247 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1248 ** extremely complex negations of CONS types were not being
1249 sufficiently canonicalized, leading to inconsistencies in
1251 ** VALUES tranformer lost derived type.
1253 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1254 * fixed a bootstrapping bug: the build process no longer assumes
1255 that the various BOOLE-related constants have the same value in
1256 host and target lisps. (noted by Paul Dietz' test suite on an
1257 SBCL binary built from CLISP)
1258 * The system can now be dynamically linked on the MIPS platform,
1259 which enables dynamic loading of foreign code from Lisp. (thanks
1260 to Ralf Baechle for discussions on the MIPS ABI)
1261 * The system now records debugging information for its own source
1262 files in a filesystem-position-independent manner, relative to
1263 the "SYS" logical host.
1264 * fixed a compiler bug: MV-LET convertion did not check references
1265 to the "max args" entry point. (reported by Brian Downing)
1266 * tweaked disassembly notes to be less confident about proclaiming
1267 some instruction as an LRA. (thanks to Brian Downing)
1268 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1269 now each have their own history, command character, and other
1270 characteristics. (thanks to David Lichteblau)
1271 * fixed some bugs revealed by Paul Dietz' test suite:
1272 ** compiler failure in compiling LOGAND expressions including a
1274 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1275 shift greater than 32.
1276 ** FLUSH-DEST did not mark blocks for type check regeneration.
1277 ** HANDLER-CASE failed to accept declarations in handler clauses
1278 in some circumstances.
1280 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1281 * New code in contrib/sb-introspect (still probably not entirely
1282 stable yet) provides some support for smart Lisp development
1283 environments like SLIME.
1284 * The conditions signalled for errors occurring when loading .fasl
1285 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1286 in a way which should help ASDF recover gracefully.
1287 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1288 stuff has been cleaned up. If you code contrib/ stuff, this might
1289 affect you, and you can look at contrib/README, contrib/STANDARDS,
1290 and/or the 0.8.4.27 diff to check.
1291 * In full calls the compiler now does not generate checks for declared
1292 argument types for all arguments.
1293 * various threading fixes
1294 ** and some experimental patches which didn't make it into
1295 the main tree for this release, but which are shipped in
1296 contrib/experimental-thread.patch as a possible fix for some
1297 failures (deadlock, spinning...) in GC-intensive multithreaded
1299 * fixed PPC build problem (source code incompatibility of different
1300 library versions): added offsetof() hackery which attempts to divine
1301 where glibc maintainers put uc_mcontext today
1302 * fixed bug 282: compiler does not trust type assertions while passing
1303 arguments to a full call.
1304 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1305 &OPTIONAL VALUES type parameter.
1306 * fix bug 214: algorithm for noting rejected templates is now more
1307 similar to that of template seletion. (also reported by rydis on
1309 * fixed bug 141b: printing backquoted information readably and prettily
1310 inserts a space where necessary.
1311 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1312 (CHARACTER 10) are now reported as errors, rather than propagated
1313 as unknown types. (reported by piso on #lisp)
1314 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1315 with values NIL and :ERROR. (thanks to Milan Zamazal)
1316 * fixed bug 191c: CLOS now does proper keyword argument checking as
1317 described in CLHS 7.6.5 and 7.6.5.1.
1318 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1319 longer raise an error; further, using a list as a for-as-arithmetic
1320 counter now raises a meaningful error.
1321 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1322 now understood by sequence creation functions such as MAKE-SEQUENCE
1324 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1325 a STRING-INPUT-STREAM.
1326 * compiler enhancement: SIGNUM is now better able to derive the type
1328 * type declarations inside WITH-SLOTS are checked. (reported by
1330 * fixed some bugs revealed by Paul Dietz' test suite:
1331 ** incorrect optimization of TRUNCATE for a positive first
1332 argument and negative second.
1333 ** compiler failure in let-convertion during flushing dead code.
1334 ** compiler failure while deriving type of TRUNCATE on an
1335 interval, containing 0.
1336 ** ASH of a negative bignum by a negative bignum count now returns
1338 ** intersection of CONS types now canonicalizes properly, fixing
1339 inconsistencies in SUBTYPEP.
1341 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1342 * incompatible change: The --disable-debugger command line
1343 option now clobbers the debugger at a more fundamental
1344 level, by redefining #'INVOKE-DEBUGGER instead of by
1345 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1346 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1347 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1348 enter the debugger REPL and then suffer recursive errors
1349 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1350 while under the new system, BREAK in --disable-debugger mode
1351 terminates the system just as an unhandled error would.
1352 * fixed compiler performance when processing loops with a step >1;
1353 * bug fix: DOCUMENTATION now retrieves generic function
1354 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1355 support has been systematized, and now supports the methods
1356 specified by ANSI, along with a default method and a method for
1357 slot documentation. (reported by Nathan Froyd)
1358 * bug fix: effective methods associated with a generic function are
1359 no longer cached over a change of that generic function's method
1360 combination. (reported by Andreas Fuchs)
1361 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1362 names a non-existent directory. (thanks to Andreas Fuchs)
1363 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1364 the PPC platform, lead to stack corruption; this has been fixed.
1365 (reported by Rainer Joswig)
1366 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1367 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1368 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1369 arithmetic in general, is now much more reliable on the PPC
1371 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1372 platform now returns the right answer.
1373 * optimization: restored some effective method precomputation in
1374 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1375 precomputation is now tunable.
1376 * optimization: compiler-internal data structure use has been
1377 reviewed, and changes have been made that should improve the
1378 performance of the compiler by about 20%.
1379 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1380 simple-base-strings and simple-bit-vectors is improved.
1381 * optimization: the optimization of 32-bit logical and arithmetic
1382 functions introduced in version 0.8.3 on the x86 has been
1383 implemented on the mips, ppc and sparc platforms; an
1384 implementation of the same facility, but for 64-bit arithmetic,
1385 has been added for the alpha.
1386 * microoptimization: the compiler is better able to make use of the
1387 x86 LEA instruction for multiplication by constants.
1388 * bug fix: in some situations compiler did not report usage of
1389 generic arithmetic in (SPEED 3) policy.
1390 * bug 145b fix: compiler used wrong type specifier while converting
1391 MEMBER-types to numeric.
1392 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1394 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1395 index argument now works properly on the Alpha platform.
1396 * bug fix: floating point exception treatment on the Alpha platform
1398 * bug fix: FILE-POSITION works much better on string input and
1399 output streams. (thanks to Nikodemus Siivola)
1400 * bug fix: many threading/garbage collection symptoms sorted.
1401 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1402 might be pseudo-atomic.
1403 * internal change: Stopping for GC is now done with signals not
1404 ptrace. GC is now done in whichever thread wanted it, instead of
1406 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1408 * bug fix: result form in DO is not contained in the implicit
1410 * incompatible change: ICR structure is changed; the value part of
1411 CONTINUATION is now called LVAR; corresponding functions are
1412 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1413 * added type deriver for ISQRT (thanks to Robert E. Brown).
1414 * bug fix: better support for loading from the command line when an
1415 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1417 * fixed some bugs revealed by Paul Dietz' test suite:
1418 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1419 ** ROUND and FROUND now give the right answer when given very
1420 small float arguments.
1421 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1423 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1424 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1425 ** LCM should return a non-negative integer.
1426 ** PARSE-INTEGER returned the index of a terminator instead of the
1427 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1428 ** PARSE-INTEGER returned an incorrect index being applied to a
1430 ** LCM with two arguments of 0 returns 0 rather than signalling
1432 ** unsigned addition of a 32-bit constant with the high bit set no
1433 longer causes an internal compiler error on the x86.
1434 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1435 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1436 with negative last argument.
1437 ** byte specifiers with zero size and position no longer cause
1438 an error during type derivation.
1439 ** bignum multiplication on the Alpha platform now returns the
1441 * porting: The system now builds on SuSE AMD64, although it still
1442 generates a 32-bit binary.
1443 * .fasl file incompatibility: The fasl file version number has
1444 been incremented (because of the changes to internal compiler
1445 data structures referred to above).
1447 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1448 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1449 more accurately, on the Darwin kernel running on PowerPC hardware.
1450 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1451 * Compiler code deletion notes now signal a condition of type
1452 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1453 an associated MUFFLE-WARNING restart.
1454 * The compiler now performs limited argument count validation of
1455 constant format strings in FORMAT, and where appropriate in ERROR,
1456 CERROR and WARN. (thanks to Gerd Moellmann)
1457 * New ASDF-INSTALL contrib can be used for automatic download and
1458 installation of third-party Lisp code from CCLAN or other sites
1460 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1461 which forces another thread to execute a function supplied by the
1463 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1464 now accept and act upon their :ELEMENT-TYPE keyword argument.
1465 (reported by Martin Atzmueller, Edi Weitz)
1466 * bug fix: FILE-POSITION now accepts position designators up to
1467 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1468 is the greater. (thanks to Patrik Nordebo)
1469 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1470 Kalle Olavi Niemitalo)
1471 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1472 printable. (reported by Eric Marsden)
1473 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1474 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1475 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1476 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1477 (thanks to Stig Sandoe)
1478 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1479 UNREAD-CHAR now correctly includes the unread character in the
1480 target sequence. (thanks to Gerd Moellmann)
1481 * bug fix in threaded builds: the system can now be suspended and
1482 resumed by shell job control with minimal disruption.
1483 * bug fixes in times and timezones >2038AD
1484 * better handling of "where is GNU make?" problem in build scripts
1485 (thanks to Nikodemus Siivola)
1486 * new optimization: inside a named function any reference to a
1487 function with the same name is considered to be a self-reference;
1488 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1489 optimization quality.
1490 * new optimization on x86: logical functions and + now have
1491 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1492 used when the result is truncated to 32 bits.
1493 * VALUES declaration is partially enabled.
1494 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1495 * fixed some bugs revealed by Paul Dietz' test suite:
1496 ** The system now obeys the constraint imposed by
1497 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1498 types form a lattice under type intersection.
1499 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1500 ** ASSOC now ignores NIL elements in an alist.
1501 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1502 and (1+ MOST-POSITIVE-FIXNUM) answers.
1503 ** The addition of a method with invalid qualifiers to a generic
1504 function does not cause an error to be signalled immediately;
1505 a warning is signalled, and the error is generated only on
1506 calling the generic function.
1507 * changed .fasl file version number, in part to add type codes for
1508 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1509 obscure ANSI requirements
1511 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1512 * fixed bug 148: failure to inline-expand a local function left
1513 garbage, confusing the compiler.
1514 * fixed bugs 3cd: structure slot readers perform type check if the
1515 slot can have an invalid value (i.e. it is either not initialized
1516 or can be written with a less specific slot writer).
1517 * bug fix: the compiler now traps array references to elements off
1518 the end of an array; previously, the bounds checking in some
1519 circumstances could go off-by-one.
1520 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1522 * type declarations for array element types now obey the description
1523 on the CLHS page "Declaration TYPE", as per discussions on
1524 sbcl-help around 2003-05-08. This means that a declaration
1525 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1526 declaration, all references to BAR will be asserted or assumed
1527 (with THE, so dependent on compiler policy) to involve objects of
1528 type FOO. Note that no such declaration is implied in
1529 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1530 * declared types of functions from the "Conditions"
1531 chapter. (reported by Paul Dietz)
1532 * bug fix: CERROR accepts a function as its first argument.
1533 * bug fix: NTH an NTHCDR accept a bignum as index
1534 arguments. (reported by Adam Warner)
1535 * optimization: character compare routines now optimize comparing
1536 against a constant character. (reported by Gilbert Baumann)
1537 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1538 argument now works properly on the MIPS platform.
1539 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1540 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1541 declared to have a functional type, does not cause an error
1543 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1544 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1545 * fixed bug: if last continuation of a deleted block has a
1546 destination, this destination should be deleted too. (reported by
1548 * fixed a bug in the bootstrap process: the host compiler's values
1549 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1550 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1551 test case from Patrik Nordebo)
1552 * improved the ability of the disassembler on the PPC platform to
1553 provide helpful disassembly notes.
1554 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1555 the class in more cases than previously.
1556 * bug fix: FILE-POSITION now understands :START and :END for
1557 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1558 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1559 8) data. (thanks to David Lichteblau)
1560 * bug fix: it is possible to add a method to a generic function
1561 without lambda list.
1562 * bug fix: reader failed to signal END-OF-FILE inside an
1563 object representation. (reported by Nikodemus Siivola)
1564 * fixed some bugs revealed by Paul Dietz' test suite:
1565 ** LAST and [N]BUTLAST should accept a bignum.
1566 ** condition slot accessors are methods.
1567 ** (VECTOR NIL) is a subtype of STRING.
1569 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1570 * minor incompatible change: some nonsensical specialized lambda
1571 lists (used in DEFMETHOD) which were previously ignored now signal
1573 * minor incompatible change: the system is now aware of the types of
1574 variables in the COMMON-LISP package, and will signal errors for
1575 most violations of these type constraints (where previously they
1576 were silently accepted).
1577 * minor incompatible change: COMPILE-FILE now uses the freedom
1578 afforded (ANSI 3.2.2.3) to use derived function types for
1579 functions defined in the same file. This also permits the system
1580 to warn on static type mismatches and function
1581 redefinition. (Currently it does not work with high DEBUG level.)
1582 * minor incompatible change: VALUES declaration is disabled.
1583 * When issuing notes, the compiler now signals a condition of type
1584 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1585 restart for use in user handlers. It is expected that the
1586 COMPILER-NOTE condition will eventually become a condition
1587 supertype to a hierarchy of note types, which will then be
1588 handleable in a similar fashion. However, at the moment, no such
1589 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1590 but it's an implementation detail, not a classification for the
1592 * Changes in type checking closed the following bugs:
1593 ** type checking of unused values (192b, 194d, 203);
1594 ** template selection based on unsafe type assertions (192c, 236);
1595 ** type checking in branches (194bc).
1596 * A short form of VALUES type specifier has ANSI meaning (it has
1597 increased the number of situations when SBCL cannot perform type
1599 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1600 keywords or constants is permissible.
1601 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1602 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1603 classes. (thanks to Antonio Martinez)
1604 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1605 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1606 argument to a pretty-print stream. Instead, it leaves any such
1607 operations to DESCRIBE-OBJECT methods.
1608 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1610 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1611 lambda lists. (thanks to David Lichteblau)
1612 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1613 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1614 (thanks to Gerd Moellmann)
1615 * fixed bug 246: increased compilation speed of long
1616 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1618 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1620 * DEFCONSTANT now throws a condition of type
1621 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1622 constant to a non-EQL value; CONTINUE and ABORT restarts
1623 respectively change and preserve the value.
1624 * fixed bug 63: The code walker, part of the implementation of CLOS,
1625 is now better at handling symbol macros.
1626 * bug fix: There is no longer an internal implementation type named
1627 CL:LENGTH. (reported by Raymond Toy)
1628 * bug fix: In macro-like defining macros/special operators the
1629 implicit block does not enclose the lambda list.
1630 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1631 enclosed in parentheses) are not suitable as type specifiers, and
1632 their use properly signals an error now.
1633 * bug fix: An argument count mismatch for a type specifier in code
1634 being compiled no longer causes an unhandled error at compile
1635 time, but signals a compile-time warning.
1636 * fixed simple vector readable printing
1637 * bug fix: DESCRIBE takes more care over whether the class
1638 precedence list slot of a class is bound before accessing it.
1639 (reported by Markus Krummenacker)
1640 * bug fix: FORMATTER can successfully compile pretty-printer format
1641 strings which use variants of the ~* directive inside.
1642 * bug fix: SEARCH now applies its TEST predicate to the elements of
1643 the arguments in the correct order. (thanks to Wolfhard Buss)
1644 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1645 in type inference. (thanks to Robert E. Brown)
1646 * bug fix: Reading in symbols with an explicit package name of ""
1647 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1648 * fixed some bugs revealed by Paul Dietz' test suite:
1649 ** NIL is now allowed as a structure slot name.
1650 ** Arbitrary numbers, not just REALs, are allowed in certain
1651 circumstances in LOOP for-as-arithmetic clauses.
1652 ** Multiple class redefinitions before slot access no longer
1653 causes a type error.
1654 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1655 association between the name and a class.
1656 ** Generic functions with non-standard method-combination and over
1657 five methods all of which return constants no longer return NIL
1658 after the first few invocations. (thanks to Gerd Moellmann)
1659 ** CALL-NEXT-METHOD with no arguments now passes the original
1660 values of the arguments, even in the presence of assignment.
1661 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1663 ** DISASSEMBLE works with closures and funcallable instances.
1664 ** ADD-METHOD now returns the generic function, not the new method.
1665 ** FIND-METHOD signals an error if the lengths of the specializers
1666 is incompatible with the generic function, even if the ERRORP
1668 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1669 which its argument is a member.
1670 ** DEFCLASS only redefines the class named by its class-name
1671 argument if that name is the proper name of the class;
1672 otherwise, it creates a new class.
1673 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1674 of the UNBOUND-SLOT condition to the name of the slot.
1675 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1676 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1677 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1678 SLOT-MAKUNBOUND in the specified fashion.
1680 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1681 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1682 cross-compilation host. As a consequence, we can now bootstrap our
1683 way up to SBCL starting with a bare gcc toolchain and human-readable
1684 source code (first the source to CLISP, then the source to SBCL).
1685 * A contributed module containing a partial implementation of the
1686 simple-streams interface has been included. (thanks to Rudi
1688 * A contributed module implementing the RFC1321 Message Digest
1689 Algorithm, known as MD5, has been included.
1690 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1691 no longer has any effect, as the code controlled by this feature
1692 has been deleted. (As far as we know, no-one has ever built using
1693 this feature, and its semantics were confused in any case).
1694 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1695 work on conditions (as required by the ANSI specification),
1696 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1697 expected behaviour on conditions. Users should note, however,
1698 that such behaviour is not required by the ANSI specification,
1699 and so use of this behaviour may render their code unportable.
1700 * fixed some bugs revealed by Paul Dietz' test suite:
1701 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1703 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1704 ** SXHASH on condition objects no longer returns NIL.
1705 ** :ALLOCATION :CLASS slots are better treated; their values are
1706 updated on class redefinition, and initforms inherited from
1707 superclasses are applied.
1708 ** REMOVE-METHOD returns its generic function argument even when
1709 no method was removed.
1710 ** SHARED-INITIALIZE now initializes the values of the requested
1711 slots, including those with :ALLOCATION :CLASS.
1712 ** ALLOCATE-INSTANCE now works on structure classes defined via
1713 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1715 ** SLOT-EXISTS-P now works on conditions, as well as structures
1717 ** MAKE-LOAD-FORM now has the required methods on
1718 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1719 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1720 keyword, and now implements the SLOT-NAMES argument.
1721 ** methods with &OPTIONAL arguments no longer allow too many
1722 arguments to be passed in the call without error.
1723 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1724 option is consistent with the required arguments of the generic
1725 function lambda list.
1726 * bug fix: REQUIRE accepts a string designator. (Thanks to
1728 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1729 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1730 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1731 * bug fix: The compiler now checks for duplicated variables in macro
1733 * bug fix: SETQ on globals returns the correct value.
1734 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1735 not a COMPILER-ERROR (followed by some other strange error on
1736 choosing the CONTINUE restart).
1737 * bug fix: make.sh and friends are now more consistent in the way that
1738 they look for GNU "make".
1740 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1741 * experimental native threads support (on x86 Linux >=2.4 only).
1742 This is not compiled in by default: you need to add :SB-THREAD to
1743 the target features. See the "Beyond ANSI" chapter of the manual
1745 * fix for longstanding nonANSIism: The old distinction between
1746 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1747 The return value from CL:FIND-CLASS is now a CLOS class, and
1748 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1749 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1750 * An interface to the MetaObject Protocol, as described in Kiczales,
1751 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1752 MIT Press, 1991, is available from the SB-MOP package.
1753 * incompatible change: the SB-PCL package should now be considered
1754 a private implementation detail, and no longer a semi-private MOP
1756 * minor incompatible change: due to rearrangement for threads, the
1757 control stack and binding stack are now allocated at arbitrary
1758 addresses instead of being hardcoded per-port. Users affected by
1759 this probably have to be doing advanced things with shared
1760 libraries, and will know who they are.
1761 * minor incompatible change: Previously, all --eval forms used were
1762 processed with READ before any of them were processed with EVAL.
1763 Now each --eval form is processed with both READ and EVAL before
1764 the next --eval form is processed. (Thus package operations like
1765 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1766 work as the user might reasonably expect.)
1767 * minor incompatible change: *STANDARD-INPUT* is now only an
1768 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1770 * minor incompatible change: Y-OR-N-P is now character-oriented,
1771 not line oriented. Also, YES-OR-NO-P now works without errors.
1772 (thanks to Antonio Martinez)
1773 * sb-aclrepl module improvements: an integrated inspector, added
1774 repl features, and a bug fix to :trace command.
1775 * Known functions, which cannot be open coded by the backend, are
1776 considered to be able to check types of their arguments. (fixing
1777 a bug report by Nathan J. Froyd)
1778 * fixed a bug in computing method discriminating functions: It is
1779 now possible to define methods specialized on classes which have
1780 forward-referenced superclasses. (thanks to Gerd Moellmann)
1781 * fixed evaluation order in optional entries (reported by Gilbert
1783 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1784 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1785 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1786 required-by-AMOP NAME argument, as well as CLASS and
1787 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1788 * fixed bug 20: DEFMETHOD can define methods using names that are
1789 not the proper names of classes to designate class specializers.
1790 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1791 Unix file descriptors, instead of blowing up. (thanks to Antonio
1793 * Garbage collection refactoring: user-visible change is that a
1794 call to the GC function during WITHOUT-GCING will not do garbage
1795 collection until the end of the WITHOUT-GCING. If you were doing
1796 this you were probably losing anyway.
1797 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1798 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1799 TYPEP the latter but not the former.
1800 * The compiler issues a full WARNING for calls to undefined functions
1801 with names from the CL package.
1802 * MAP-INTO for a vector destination is open coded. (reported by
1803 Brian Downing on c.l.l)
1804 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1805 documentation string.
1806 * fixed some bugs revealed by Paul Dietz' test suite:
1807 ** COPY-ALIST now signals an error if its argument is a dotted
1809 ** Condition slots are now accessed more correctly in the presence
1810 of multiple initargs for a given slot.
1811 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1812 exclude restarts of the same name associated with a different
1814 ** DEFCLASS of forward-referenced classes with another
1815 forward-referenced class in the superclasses list no longer
1817 ** Condition slots are now initialized once each, not multiple
1818 times. (thanks to Gerd Moellmann)
1819 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1821 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1822 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1823 arguments contain duplicated elements.
1824 ** RESTART-CASE understands local macros.
1825 ** RESTART-CASE associates exactly its own restarts with a condition.
1826 ** ENDP in safe mode checks its argument to be of type LIST.
1827 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1828 in question is unbound.
1829 ** Optimized MAKE-INSTANCE functions no longer cause internal
1830 assertion failures in the presence of duplicate initargs.
1831 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1832 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1833 propagated even in the ctor optimized implementation of
1835 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1837 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1838 * a better implementation of SXHASH on (simple) bit vectors,
1839 measured both in execution speed and in distribution of results
1840 over the positive fixnums, has been installed. Likewise, a better
1841 implementation of EQUAL for simple bit vectors is now available.
1842 * fixed CEILING optimization for a divisor of form 2^k.
1843 * fixed bug 240 (emitting extra style warnings "using the lexical
1844 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1845 by Antonio Martinez)
1846 * fixed SXHASH, giving different results for NIL depending on type
1847 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1848 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1849 variable. (found by Rolf Wester)
1850 * SBCL does not ignore type declarations for special
1851 variables. (reported by rif on c.l.l 2003-03-05)
1852 * some bug fixes in contrib/sb-aclrepl/
1853 * fixed some bugs revealed by Paul Dietz' test suite:
1854 ** a bug in the CONS type specifier, whereby the CAR and CDR
1855 types got intertwined, has been fixed;
1856 ** the type system is now able to reason about the interaction
1857 between INTEGER and RATIO types more completely;
1858 ** APPEND, [N]REVERSE and NRECONC check that those their
1859 arguments, which must be proper lists, are really so;
1860 ** An array specialized to be unable to hold elements has been
1861 implemented, as required -- yes, really -- by ANSI;
1862 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1863 on malformed property lists;
1865 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1866 * incompatible packaging change: in line with Unix convention,
1867 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1868 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1869 * REQUIRE and PROVIDE are now optionally capable of doing something
1870 useful. See the documentation string for REQUIRE.
1871 * infrastructure for a managed SBCL contrib system: contributed
1872 modules in this release include:
1873 ** the ASDF system definition facility;
1874 ** an interface to the BSD Sockets API;
1875 ** an ACL-like convenience interface to the repl;
1876 (thanks to Kevin Rosenberg)
1877 ** an implementation of ROTATE-BYTE, with efficient implementation
1879 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1880 gives a closer approximation to the right answer than previously.
1881 (thanks to Raymond Toy)
1882 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1883 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1884 optional environment arguments, as required by ANSI.
1885 * fixed bugs in other functions taking environment objects, allowing
1886 calls with an explicit NIL environment argument to be compiled
1888 * fixed bug 228: primary return values from
1889 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1890 COMPILE or FUNCTION.
1891 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1892 :TYPE will no longer signal errors on innocuous objects.
1893 * fixed bug 231b: SETQ is better at respecting type declarations in
1894 the lexical environment.
1895 * fixed a bug in DEFCLASS: classes named by symbols with no or
1896 unprintable packages can now be defined.
1897 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1898 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1899 * fixed compiler failure related to checking types of functions.
1900 (reported by Robert E. Brown)
1901 * the compiler is now much more consistent in its error-checking
1902 treatment of bounding index arguments to sequence functions: in
1903 (SAFETY 3) code, errors will be signalled in almost all cases if
1904 invalid sequence bounding indices are passed to functions defined
1905 by ANSI to operate on sequences.
1906 * fixed a bug in the build procedure: documentation of SBCL-specific
1907 packages is now preserved and available in the final Lisp image.
1908 * lifted FDEFINITION lookup out of loops in the implementation of
1909 many list operations. (thanks to Robert E. Brown)
1910 * fixed a bug in the reader: the #n# reader macro now works for
1911 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1912 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1913 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1914 E. Brown and Tony Martinez respectively)
1915 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1916 SLOT-BOUNDP: in particular, it is now able to optimize them much
1917 better, and is now not vulnerable to having packages renamed.
1918 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1919 always called when a slot is not present in an instance. (thanks
1921 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1922 longer signal an error on unreading a character following EOF on
1923 the previous constituent stream. (thanks to Tony Martinez)
1924 * fixed some bugs revealed by Paul Dietz' test suite:
1925 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1926 not just nonnegative fixnums;
1927 ** the logical bit-array operators such as BIT-AND now accept an
1928 explicit NIL for their "opt-arg" argument (to indicate a
1929 freshly-consed result bit-array);
1930 ** ELT now signals an error on an invalid sequence index in safe
1932 ** the type system is now cleverer about negations of numeric
1933 types, and consequently understands the BIGNUM and RATIO types
1935 ** the type system is now cleverer about the interaction between
1936 INTEGER and RATIO types: while bugs still remain, many more
1937 cases are accurately computed;
1938 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1939 if it is in the last clause;
1940 ** CONSTANTLY now correctly returns a side-effect-free function in
1942 ** DECLARE is no longer treated as a special-operator; in
1943 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1944 * incremented fasl file version number due to the change in the
1945 DEFSTRUCT-SLOT-DESCRIPTION structure.
1947 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1948 * minor incompatible change: code processed by the "interpreter" or
1949 EVAL now has a compilation optimization policy of (DEBUG 2)
1950 (changed from (DEBUG 1)) to improve debuggability of interactive
1951 development, and to allow the use of the debug RETURN command in
1953 * an experimental implementation of the RETURN command for the
1954 debugger has been included. (thanks to Frederik Kuivinen)
1955 * fixed bug 62: constraints were not propagated into a loop.
1956 * fixed bug in embedded calls of SORT (reported and investigated by
1958 * fixed some bugs revealed by Paul Dietz' test suite:
1959 ** printing and reading of arrays with some dimensions having
1960 length 0 (thanks to Gerd Moellmann);
1961 ** BOA constructor with &AUX argument without a default value does
1962 not cause a type error;
1963 ** CONSTANTP now returns true for all self-evaluating objects.
1965 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1966 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1967 accessors that are related by inheritance, as specified in the
1968 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1969 Valtteri Vuorikoski)
1970 * The compiler is now able to inline functions that were defined in
1971 a complex lexical environment (e.g. inside a MACROLET).
1972 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1974 * Support for the upcoming FreeBSD-5.0 release has been included.
1975 (thanks to Dag-Erling Smorgrav)
1976 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1977 effect when it is not in a toplevel context.
1978 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1979 stand a better chance of being correct. (thanks to Gerd
1981 * fixed bug in COERCE, which now signals an error on coercing a
1982 rational to a bounded real type which excludes the expected
1984 * The compiler is now able to derive types more accurately from the
1985 COERCE and COMPILE functions.
1986 * fixed bug 223: functional binding is considered to be constant
1987 only for symbols in the CL package.
1988 * fixed bug 231: SETQ did not check the type of a variable being set
1989 (reported by Robert E. Brown)
1990 * A new optimization for MAKE-INSTANCE has been included, fixing
1991 various bugs (including relating to :ALLOCATION :CLASS slots and
1992 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1994 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1995 ** As required by ANSI, LOOP now disallows anonymous collection
1996 clauses such as COLLECT I in conjunction with aggregate boolean
1997 clauses such as THEREIS (= I 1);
1998 ** LOOP now signals an error when any variable is reused in the
1999 same loop (including the potentially useful construct analogous
2000 to WITH A = 1 WITH A = (1+ A);
2001 ** IT is only a special loop symbol within the first clause of a
2002 conditional loop clause;
2003 ** LOOP with a typed iteration variable over a hashtable now
2004 signals a type error iff it should.
2005 * fixed some other bugs revealed by Paul Dietz' test suite:
2006 ** FILE-STREAM now names the class previously known as FD-STREAM;
2007 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2008 argument) no longer signals an error;
2009 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2010 of the slot symbol, rather than using the current package
2011 ((:CONC-NAME "") continues to intern the slot's name in the
2013 * incremented fasl file version number, because of the incompatible
2014 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2015 of the new implementation of DEFINE-COMPILER-MACRO.
2017 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2018 * Support for building SBCL for MIPS platforms running in
2019 little-endian mode has now been checked in, and basic
2020 functionality on said platforms verified.
2021 * minor incompatible change: PCL now records the pathname of a file
2022 in which methods and the like are defined, rather than its
2024 * minor incompatible change: TRUENAME now considers the truename of
2025 a file naming a directory to be the pathname with :DIRECTORY
2026 component indicating that directory.
2027 * minor incompatible change: a NAMED clause in the extended form of
2028 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2029 reason for the previous behaviour is unclear.
2030 * more systematization and improvement of CLOS and MOP conformance
2031 in PCL (thanks to Gerd Moellman and Pierre Mai):
2032 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2034 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2035 primary methods with no specializers;
2036 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2038 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2039 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2040 FORWARD-REFERENCED-CLASSes; error reporting on
2041 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2043 ** SXHASH on CLOS instances now uses a slot internal to the
2044 instance to return different numbers on distinct instances,
2045 while preserving the same return value through invocations of
2047 ** DEFMETHOD signals errors when methods with longer incongruent
2048 lambda lists are added to generic functions;
2049 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2050 CLASS, as specified in AMOP;
2051 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2052 on the order returned by the primary method for classes of
2053 class STANDARD-CLASS;
2054 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2055 * fixed some bugs shown by Paul Dietz' test suite:
2056 ** DOLIST puts its body in TAGBODY;
2057 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2059 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2060 value producing form;
2061 ** if more variables are given to PROGV than values, extra
2062 variables are bound and made to have no value;
2063 ** NSUBSTITUTE on list arguments gets the right answer with
2065 ** ELT signals an error of type TYPE-ERROR when the index argument
2066 is not a valid sequence index;
2067 ** LOOP signals (at macroexpansion time) an error of type
2068 PROGRAM-ERROR when duplicate variable names are found;
2069 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2070 ** FUNCALL of special-operators now cause an error of type
2072 ** PSETQ now works as required in the presence of side-effecting
2073 symbol-macro places;
2074 ** NCONC accepts any object as its last argument;
2075 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2077 ** loop-for-as-package does not require a package to be explicitely
2079 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2081 * fixed bug 166: compiler preserves "there is a way to go"
2082 invariant when deleting code.
2083 * fixed bug 172: macro lambda lists with required arguments after
2084 &REST arguments now cause an error to be signalled. (thanks to
2086 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2087 bug. (thanks to Matthew Danish)
2088 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2090 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2091 arguments are lexically rebound. (thanks to Gerd Moellmann and
2093 * fixed bug 194: error messages are now more informative when there
2094 is no primary method applicable in a call to a generic
2095 function. (thanks to Gerd Moellmann)
2096 * fixed bug in command line argument checking (thanks to Julian
2098 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2099 (thanks to Matthew Danish)
2100 * incremented fasl file version number, because of the
2101 SXHASH-related changes in the layout of CLOS data structures
2103 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2104 * minor incompatible change: The runtime (the Unix executable named
2105 "sbcl") is now much pickier about the .core files it will load.
2106 Essentially it now requires .core files to descend from the same
2107 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2108 as the runtime does. (The intent is to prevent the crashes which
2109 can occur, and which can even be reported as mysterious failures,
2110 when people patch the sources or change the build parameters
2111 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2112 sbcl and .core files.)
2113 * fixed bug: VALUES-LIST is no longer optimized away.
2114 * fixed bug 142: The FFI conversion of C string values to Lisp
2115 string values no longer conses excessively. (thanks to Nathan
2116 Froyd porting Raymond Toy's fix to CMU CL)
2117 * began to systematize and improve MOP conformance in PCL (thanks to
2118 Nathan Froyd, Gerd Moellman and Pierre Mai):
2119 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2121 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2122 ** FINALIZE-INHERITANCE is now called on class finalization;
2123 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2124 argument precedence order.
2125 * fixed bug 202: The compiler no longer fails on functions whose
2126 derived types contradict their declared type.
2127 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2128 so it can be non-toplevel.
2129 * The fasl file version number has changed (because of the new
2130 implementation of DEFMACRO).
2131 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2132 safe code, that any length requirement by their type-specifier
2133 argument is valid. The exceptions to this are described in bug
2135 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2136 functions now check the types of their inputs as required by ANSI.
2137 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2138 introduced symbol is DECLAREd to be SPECIAL.
2139 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2140 * fixed bug: :COUNT argument to sequence functions may be negative
2141 * fixed bug: body of DO-SYMBOLS may contain declarations
2142 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2143 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2144 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2145 symbol macro only once
2146 * fixed printing of call frame when argument list is unavailable
2147 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2148 * compiler no longer signals WARNING on unknown keyword
2151 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2152 * A beta-quality port to the mips architecture running Linux,
2153 based on the old CMUCL backend, has been made. It has been tested
2154 on a big-endian kernel, and works sufficiently well to be able to
2155 rebuild itself; it has not been tested in little-endian mode.
2156 * fixed an inconsistency between gencgc.c and purify.c which made
2157 dumping/loading .core files unreliable
2158 * fixed bug 120a: The compiler now deals correctly with IFs where
2159 the consequent is the same as the alternative, instead of
2160 misderiving the return type. (thanks to Alexey Dejneka)
2161 * fixed bug 113: Logical pathnames are now dumpable (the logical
2162 host is resolved at load-time, throwing an error if it is not
2164 * fixed bug 174: FORMAT's error message is slightly clearer when a
2165 non-printing character is used in a format directive.
2166 * fixed several bugs in compiler checking of type declarations, i.e.
2167 violations of the Python "declarations are assertions" principle
2168 (thanks to Alexey Dejneka)
2169 * fixed several bugs in PCL's error checking (thanks to Gerd
2171 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2173 * fixed bug in compilation of functions as first class values
2174 (thanks to Antonio Martinez-Shotton)
2175 * The compiler's handling TYPE-ERRORs which it can prove will
2176 inevitably happen at runtime has been cleaned up and corrected
2177 in several ways. (thanks to Alexey Dejneka)
2178 * improved argument type checking for various basic arithmetic
2179 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2180 lavished on them in the past that they can be compiled in many
2181 ways in different special cases
2182 * fixed bug 181: compiler checks validity of user supplied type
2184 * cleaned up code flushing in optimization: Function calls which
2185 should signal errors for safety purposes (e.g. which ANSI says
2186 should signal errors when their arguments are of incorrect type)
2187 are no longer optimized away.
2188 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2189 * incremented fasl file version number, because changes in the
2190 implementation of sequence functions like COERCE caused
2191 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2192 in old inline expansions) to become undefined. (Actually these
2193 changes were later undone, so we might very well be binary
2194 compatible with 0.7.7 after all, but leaving the version number
2195 incremented seemed like the simplest and most conservative
2198 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2199 * An alpha-quality port to the parisc architecture running Linux,
2200 based on the old CMUCL backend, has been made. This, even more so
2201 than the other backends, should be considered still a work in
2202 progress; known problems include that the Linux kernel in 64-bit
2203 mode does not propagate the correct sigcontext structure to
2204 userspace, and consequently SBCL on a parisc64 kernel will not
2206 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2207 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2208 suppressed.) Also now that the compiler is looking at declarations
2209 in the environment, it checks optimization declarations as well,
2210 and suppresses inlining when (> DEBUG SPEED).
2211 * More fixes have been made to treatment of floating point exception
2212 treatment and other Unix signals. In particular, floating point
2213 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2214 * The detection and handling of control stack exhaustion (infinite
2215 or very deeply nested recursion) has changed. Stack exhaustion
2216 detection is now done by write-protecting pages at the OS level
2217 and applies at all optimization settings; when found, a
2218 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2219 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2220 be caught using IGNORE-ERRORS.
2221 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2222 correctly when transforming recursive LABELS functions to LETs.
2223 (thanks to Alexey Dejneka)
2224 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2225 that are names of constants or global variables.
2226 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2227 alien routines with docstrings.
2228 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2229 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2231 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2232 (thanks to Eric Marsden)
2233 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2234 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2235 * Bug fix: The SPARC backend can now compile functions involving
2236 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2237 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2238 to a functions expecting a PATHNAME-DESIGNATOR.
2239 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2240 lambda lists. (thanks to Alexey Dejneka)
2241 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2242 functions so that EXPORTing the name of the class doesn't cause
2243 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2244 to Antonio Martinez for reporting this)
2245 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2246 constant-folded without failing an assertion. (thanks to Einar
2247 Floystad Dorum for reporting this)
2248 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2249 types are now dealt with more correctly by the compiler.
2250 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2251 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2252 behaviour analogous to RENAME-FILE. This puts its behaviour more
2253 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2255 * The fasl file version number has changed again. (because of the
2256 bug fix involving the names of PCL MAKE-INSTANCE functions)
2258 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2259 * bug fix: Floating point exceptions are treated much more
2260 consistently on the x86/Linux and PPC/Linux platforms.
2261 * Array initialization with :INITIAL-ELEMENT is now much faster for
2262 cases when the compiler cannot open code the array creation, but
2263 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2264 array accesses have also seen a speed increase.
2265 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2266 specified. (This was at the root of some bad interactions between
2267 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2269 * bug fix: Internal error arguments for undefined functions are now
2270 computed correctly on the PPC/Linux platform.
2271 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2272 Raymond Toy's patch for CMU CL)
2273 * Support for the Solaris 9 operating environment has been included
2274 (thanks to Daniel Merritt)
2275 * A very ugly but hopefully complete draft of the missing FFI chapter
2276 of the manual has been created by reformatting the corresponding
2277 CMU CL manual chapter into (currently very ugly and incoherent)
2278 DocBook and bringing it up to date for SBCL behavior. Thus, the
2279 manual is now essentially complete, at least by my extreme
2280 once-and-only-once standards, whereby it's acceptable to refer to
2281 the doc strings of SB-EXT functions as the primary documentation.
2282 * The fasl file version number has changed again, due to cleanup of
2283 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2284 is no longer a static symbol.)
2286 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2287 * SBCL now builds with OpenMCL (version 0.12) as the
2288 cross-compilation host; also, more progress has been made toward
2289 bootstrapping under CLISP.
2290 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2292 * bug 158 fixed: The compiler can now deal with integer loop
2293 increments different from 1; fixing this turned out also to fix
2295 * bug 169 fixed: no more bogus warnings about using lexical bindings
2296 despite the presence of perfectly good SPECIAL declarations (thanks
2297 to David Lichteblau)
2298 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2299 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2300 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2302 * bug 180 fixed: Method combination specifications no longer ignore
2303 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2304 * bug fix: Structure type predicate functions now check their argument
2305 count as they should.
2306 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2307 correctly. (thanks to Pierre Mai)
2308 * minor incompatible change: The --noprogrammer option is deprecated
2309 in favor of the new --disable-debugger option, which is very similar.
2310 (The major difference is that it takes effect at a slightly different
2311 time at startup, causing handling of errors in --sysinit and
2312 --userinit files will be affected differently.) The
2313 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2314 been added to allow this functionality to be controlled from ordinary
2315 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2316 maintainers, who might want to run non-interactive scripts to
2317 build SBCL cores which will later be used interactively.)
2318 * minor incompatible change: The LOAD function no longer, when given
2319 a wild pathname to load, loads all files matching that pathname.
2320 Instead, an error of type FILE-ERROR is signalled.
2322 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2323 * bug 147 fixed: The compiler preserves its block link/count
2324 invariants more correctly now so that it doesn't crash. (thanks
2326 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2328 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2329 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2330 no-value-for-this-character as NIL (as specified by ANSI).
2331 * HOST-NAMESTRING on physical pathnames now returns a string that is
2332 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2333 (thanks to Christophe Rhodes)
2334 * The Alpha port handles icache flushing more correctly. (thanks to
2336 * More progress has been made toward bootstrapping under CLISP. (thanks
2337 to Christophe Rhodes)
2338 * The fasl file format has changed again, because dynamic loading
2339 on OpenBSD (which has non-ELF object files) motivated some cleanups
2340 in the way that foreign symbols are transformed and passed around.
2341 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2342 is no longer treated as whitespace by the reader, but instead as
2343 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
2344 |A
\x7fB|, instead of A as it used to.
2346 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2347 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2348 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2349 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2350 this as of 0.7.1.45, but was left out of the previous news section
2351 (thanks to Dan Barlow)
2352 * SBCL now runs on the Solaris operating system on SPARC architectures
2353 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2354 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2355 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2356 * SPARC backend cleanups, allowing builds of cores optimized for V8
2357 and V9 SPARCS, and also emission of code targeted to a particular
2358 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2360 * SBCL is closer to bootstrapping under CLISP, thanks to various
2361 fixes by Christophe Rhodes.
2362 * The fasl file format has changed again, to allow the compiler's
2363 INFO database to support symbol macros.
2364 * The user manual (in doc/) is formatted into HTML more nicely.
2365 (thanks to coreythomas)
2366 * The system is smarter about SUBTYPEP relationships, especially
2367 those involving NOT types (including types such as ATOM which are
2368 represented internally using NOT types). Thus SUBTYPEP is less
2369 likely to return (VALUES NIL NIL) in general, and in particular
2370 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2372 * The fasl file format has changed again, because the internal
2373 representation of types now includes a new slot to support the new
2374 SUBTYPEP-of-NOT-types logic.
2375 * (not a change in the main branch of SBCL, but a related prototype
2376 which can hopefully be merged into the main branch of SBCL in the
2377 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2378 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2379 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2380 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2381 dispatch macro character combinations. (thanks to Alexey Dejneka)
2382 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2383 that superficially look like logical namestrings correctly.
2384 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2386 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2387 default supplied-p) for &optional and &key arguments. (thanks to
2390 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2391 * incompatible change: The compiler is now less aggressive about
2392 tail call optimization, doing it only when (> SPACE DEBUG) or
2393 (> SPEED DEBUG). (This is an incompatible change because there are
2394 programs which relied on the old CMU-CL-style behavior to optimize
2395 away their unbounded recursion which will now die of stack overflow.)
2396 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2397 for non-GENCGC systems has been increased to 20M (since that
2398 seems much closer to the likely performance optimum for modern
2399 systems than the old 4M value was)
2400 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2401 and *DEBUG-PRINT-LEVEL*
2402 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2403 of CMU CL's support for SPARC, and various endianness and other
2404 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2405 * new syntactic sugar for the Unix command line: --load foo.bar is now
2406 an alternate notation for --eval '(load "foo.bar")'.
2408 ** The system now detects stack overflow and handles it gracefully,
2409 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2410 optimization settings. (This is a good thing in general, and
2411 its introduction in this version should be particularly timely
2412 for anyone whose code fails because of suppression of tail
2414 ** The system now hunts for the C variable "environ" in a more
2415 devious way, to avoid segfaults when the C library version differs
2416 between compile time and run time. (thanks to Christophe Rhodes)
2417 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2418 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2419 ** The compiler no longer issues bogus style warnings for undefined
2420 classes in the same source file as the DEFCLASSes which defined
2421 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2423 ** fixes in CONDITION class precedence list for undefined function
2424 errors (thanks to Alexei Dejneka)
2425 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2426 correctly. (thanks to Dan Barlow)
2427 ** portability fixes aiming at bootstrapping under CLISP (thanks
2428 to Dave McDonald and Christophe Rhodes)
2429 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2430 ** fixes in type translation and and type inference (thanks to
2432 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2433 ** various fixes in the new ports (thanks to Dan Barlow)
2434 * several changes related to debugging:
2435 ** suppression of tail recursion, as noted above
2436 ** stack overflow detection, as noted above
2437 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2438 is now the default. (For some time encapsulation has been more
2439 reliable than the breakpoint-based :ENCAPSULATE NIL
2440 implementation, at least on X86 systems; and I just noticed that
2441 encapsulation also seems closer to the spirit of the ANSI
2444 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2446 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2447 up properly again. (There was a packaging bug in 0.7.0 which
2448 left their definitions in SB-SYS::LOAD-FOREIGN and
2449 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2450 vital for most things which interface to C-level interfaces,
2451 like extensions working with sockets or databases or
2452 Perl-compatible regexes or whatever, and the need to fix
2453 this bug is the main reason that 0.7.1 was released so
2455 ** DEFGENERIC is now choosier about the methods it redefines, so that
2456 reLOADing a previously-LOADed file containing DEFGENERICs does
2457 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2458 cycle now works as it should. (thanks to Alexey Dejneka)
2459 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2460 fixed by Christophe Rhodes. (He actually submitted this patch
2461 months ago, and I delayed until after 0.7.0.)
2462 ** Bug 111 (internal compiler confusion about runtime checks on
2463 FUNCTION types) was fixed by Alexey Dejneka.
2464 * Some internal cleanups (getting rid of variables which aren't
2465 needed now that the byte interpreter is gone) caused the fasl
2466 file format number to change again.
2468 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2469 * major incompatible change: The default fasl file extension, i.e. the
2470 default extension for files produced by COMPILE-FILE, has changed
2471 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2473 ** There are many changes in the implementation of the compiler.
2474 SBCL is now essentially a compiler-only implementation of ANSI
2475 Common Lisp. EVAL still "interprets" a few special cases, but
2476 almost all the interesting cases are handled by creating
2477 a LAMBDA expression, calling COMPILE on it, then calling
2478 FUNCALL on the result.
2479 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2480 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2481 Since the code is newer, there might still be some new bugs
2482 (though not as many as before Martin Atzmueller's fixes:-). But
2483 the new code is substantially simpler and clearer, and hopefully
2484 any remaining bugs will be simpler, less fundamental, and more
2485 fixable then the bugs in the old code.
2486 ** The revised compiler is still a little unsteady on its feet.
2488 *** The debugging information it produces (particularly the names
2489 of FUNCTION objects) is sometimes much less useful than what
2490 the old compiler produced.
2491 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2492 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2493 within a MACROLET) has been temporarily weakened.
2494 ** There are new compiler optimizations for various functions:
2495 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2496 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2497 *** the math functions TRUNCATE, FLOOR, and CEILING
2498 *** the function-of-all-trades COERCE
2499 Mostly these should be transparent, but there's one
2500 potentially-annoying problem (bug 117): when the compiler
2501 inline-expands a function and does type analysis on the result,
2502 it can create control paths which have type mismatches, and
2503 when it can't prove that those control paths aren't taken,
2504 it will issue WARNINGs about the type mismatches. This is
2505 a particular problem in practice for the new sequence functions.
2506 It's not clear how this should be fixed, and for now, a
2507 workaround is given in the entry for 117 in the BUGS file.
2508 ** (Because of the interaction between the two previous items --
2509 occasional inlining problems and new inline expansions -- some
2510 of the new sequence function optimizations won't really kick in
2511 completely until debugging information, and then inlining, are
2512 straightened out in some future version.)
2513 * minor incompatible changes:
2514 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2515 DIRECTORY no longer implicitly promotes NIL slots of its
2516 pathname argument to :WILD. In particular, when you ask for the
2517 contents of a directory (which you used to be able to do without
2518 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2519 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2520 ** changes in behavior that ANSI explicitly defines to be
2521 implementation dependent:
2522 *** The new compiler-only implementation still conforms with ANSI,
2523 but acts a little different than before. Besides the obvious
2524 changes in performance tradeoffs (that the cost per form passed
2525 to EVAL has gone up, and the cost per form executed by EVAL
2526 has gone down), the behavior of the system changes a little
2527 because there are no longer any interpreted function objects.
2528 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2529 e.g. doing COMPILE on the output of interactive DEFUN is
2531 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2533 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2534 has changed from (:CL) to NIL.
2535 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2536 e.g. #\Newline and #\Tab, don't have names specified in the
2537 ANSI Common Lisp standard, is now based on their ASCII symbolic
2538 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2539 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2540 are no longer used for output.
2541 ** changes in internal implementation constants:
2542 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2543 4 million. (If your application spends a lot of time GCing and
2544 you have a lot of RAM, you might want to experiment with
2545 increasing it even more.)
2546 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2547 However, almost all old code should still continue to work without
2548 immediate update, as SB-C-CALL is now a (deprecated) nickname
2550 ** Old operator names in the style DEF-FOO are now deprecated in
2551 favor of new corresponding names DEFINE-FOO, for consistency with
2552 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2553 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2554 internal symbols, but a few supported extensions like
2555 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2556 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2557 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2558 etc. as you get deeper into recursive calls to the debugger
2559 command loop, instead of the old "5]", "5]]", "5]]]"
2560 sequence. (I was motivated to do this when squabbles between
2561 ILISP and SBCL left me very deeply nested in the debugger. In the
2562 short term, this change will probably provoke more ILISP/SBCL
2563 squabbles, but hopefully it will be an improvement in the long run.)
2564 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2565 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2566 function, since declaiming return types involving aliens is
2567 (1) annoyingly messy to do by hand and (2) vital to efficient
2568 compilation of code which calls such functions.
2569 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2570 longer reexported by the SB-EXT package. They're solely useful
2571 for alien code, so it seems more logical that you should get
2572 them from the SB-ALIEN package, not in SB-EXT.
2573 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2574 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2575 features. Instead, the code that they used to control is always
2576 built into the system.
2577 * many other bug fixes
2578 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2579 advantage of the new EVAL-WHEN stuff and to clean them up in
2580 general, and they are now more ANSI-compliant in a number of
2581 ways. Martin Atzmueller is responsible for a lot of this.
2582 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2584 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2585 *** correct ERROR type for various file operations
2586 *** some fixes for Lisp streams
2587 *** DEFMETHOD syntax checking
2588 *** changing old weird representation of debug information as
2589 strings (which, among their other deficiencies, don't transform
2590 correctly when you rename packages, and don't change their
2591 print representation when you change things like *PACKAGE*
2592 and *PRINT-LENGTH*) to symbols and lists of symbols
2593 He also made several improvements and fixed several bugs in DESCRIBE.
2594 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2596 *** misbehavior of WRITE-STRING/WRITE-LINE
2597 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2598 and several other LOOP problems as well
2599 *** DIRECTORY when similar filenames are present
2600 *** DEFGENERIC with :METHOD options
2601 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2602 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2603 *** argument ordering in FIND with :TEST option
2604 *** mishandled package designator argument in APROPOS-LIST
2605 *** various problems in the backquote readmacro
2606 *** a bug in APROPOS
2607 *** probably some others that I'm not describing very well here,
2608 since the CVS log documents them by reference to sbcl-devel
2609 messages, and the SourceForge archives aren't working well.:-(
2610 ** Dan Barlow improved the Alpha port (and is making progress on the
2611 PPC port, for those of you who think different).
2612 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2613 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2614 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2615 continues to work on the SPARC port (for those of you in a position
2616 to look down upon our little PC-compatible boxes from a great height).
2617 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2618 and uses that copy, rather than the current dynamic value, when
2619 it's trying to decide whether to truncate output. Thus e.g.
2620 (let ((*print-lines* 50))
2621 (pprint-logical-block (stream nil)
2623 (let ((*print-lines* 8))
2624 (print (aref possiblybigthings i) stream)))))
2625 should now truncate the logical block only at 50 lines, instead of
2626 often truncating it at 8 lines, as it did before.
2627 * The doc/cmucl/ directory, containing old CMU CL documentation
2628 from the time of the fork, is no longer part of the base system.
2629 SourceForge has shut down its anonymous FTP service, and with it
2630 my original plan for distributing the old CMU CL documentation
2631 there. For now, if you need these files you can download an old
2632 SBCL source release and extract them from it.
2633 * The fasl file version number changed again, for dozens of reasons,
2634 some of which are apparent above.
2636 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2637 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2638 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2639 patch, so that DISASSEMBLE output is much nicer.
2640 * The code in the SB-PROFILE package now seems reasonably stable.
2641 I still haven't decided what the final interface should look like
2642 (I'd like PROFILE to interact cleanly with TRACE, since both
2643 facilities use function encapsulation) but if you have a need
2644 for profiling now, you can probably use it successfully with
2645 the current CMU-CL-style interface.
2646 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2647 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2648 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2649 link now returns the dangling link itself, and for similar
2650 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2651 link itself. (In these cases the old code signalled an error and
2652 looped endlessly, respectively.) Thus, DIRECTORY now works even
2653 in the presence of dangling and cyclic symbolic links.
2654 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2655 is now a supported extension again, since the consensus on
2656 sbcl-devel was that it can be useful for ordinary development
2657 work, not just for debugging SBCL itself.
2658 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2659 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2660 that currently-defined functions might be redefined later with
2661 different return types.
2662 * Hash tables can be printed readably, as inspired by CMU CL code
2663 of Eric Marsden and SBCL code of Martin Atzmueller.
2664 * better error handling in CLOS method combination, thanks to
2665 Martin Atzmueller porting Pierre Mai's CMU CL patches
2666 * more overflow fixes for >16Mbyte I/O buffers
2667 * A bug in READ has been fixed, so that now a single Ctrl-D
2668 character suffices to cause end-of-file on character streams.
2669 In particular, now you only need one Ctrl-D at the command
2670 line (not two) to exit SBCL.
2671 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2673 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2674 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2675 does the right thing.
2676 * The implementation of some type tests, especially for CONDITION
2677 types, is now tidier and maybe faster, due to CMU CL code
2678 originally by Douglas Crosher, ported by Martin Atzmueller.
2679 * Some math functions have been fixed, and there are new
2680 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2681 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2682 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2683 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2684 likely to be built into sbcl-0.7.0.)
2685 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2686 CL, ported to SBCL by Martin Atzmueller)
2687 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2688 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2689 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2690 * There's a new slam.sh hack to shorten the edit/compile/debug
2691 cycle for low-level changes to SBCL itself, and a new
2692 :SB-AFTER-XC-CORE target feature to control the generation of
2693 the after-xc.core file needed by slam.sh.
2694 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2695 COMPILE-FILE is no longer supported, so that now every function
2696 gets an entry point, so that block compilation looks a little
2697 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2698 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2699 returns the number of bytes consed since the system started,
2700 rather than the number consed since the first time the function
2701 was called. (The new definition parallels ANSI functions like
2702 CL:GET-INTERNAL-RUN-TIME.)
2703 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2704 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2705 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2706 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2707 * Fasl file version numbers are now independent of the target CPU,
2708 since historically most system changes which required version
2709 number changes have affected all CPUs equally. Similarly,
2710 the byte fasl file version is now equal to the ordinary
2713 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2714 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2715 is no longer recognized. I apologize for this, because it was
2716 listed in SB-EXT as a supported extension, but I found that
2717 its existing behavior was poorly specified, as well as incorrectly
2718 specified, and it looked like too much of a mess to straighten it
2719 out. I have enough on my hands trying to get ANSI stuff to work..
2720 * many patches ported from CMU CL by Martin Atzmueller, with
2721 half a dozen bug fixes in pretty-printing and the debugger, and
2722 half a dozen others elsewhere
2723 * fixed bug 13: Floating point infinities are now supported again.
2724 They might still be a little bit flaky, but thanks to bug reports
2725 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2726 as flaky as they were.
2727 * The --noprogrammer command line option is now supported. (Its
2728 behavior is slightly different in detail from what the old man
2729 page claimed it would do, but it's still appropriate under the
2730 same circumstances that the man page talks about.)
2731 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2732 are now supported, and enabled by default. Thus, the compiler can
2733 handle many floating point and complex operations much less
2734 inefficiently. (Thus e.g. you can implement a complex FFT
2736 * The compiler now detects type mismatches between DECLAIM FTYPE
2737 and DEFUN better, and implements CHECK-TYPE more correctly, and
2738 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2739 thanks to patches from Martin Atzmueller.
2740 * various fixes to make the cross-compiler more portable to
2741 ANSI-conforming-but-different cross-compilation hosts (notably
2742 Lispworks for Windows, following bug reports from Arthur Lemmens)
2743 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2744 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2745 fixed thanks to Pierre Mai's CMU CL patches.
2746 * fixes to keep the system from overflowing internal counters when
2747 it tries to use i/o buffers larger than 16M bytes
2748 * fixed bug 45a: Various internal functions required to support
2749 complex special functions have been merged from CMU CL sources.
2750 (When I was first setting up SBCL, I misunderstood a compile-time
2751 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2752 * improved support for type intersection and union, fixing bug 12
2753 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2754 more obscure bugs as well
2755 * some steps toward byte-compiling non-performance-critical
2756 parts of the system, courtesy of patches from Martin Atzmueller
2757 * Christophe Rhodes has made some debian packages of sbcl at
2758 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2759 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2760 stable, but are nonetheless usable. When he's ready, I'd be happy
2761 to add them to the SourceForge "File Releases" section. (And if
2762 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2763 * new fasl file format version number (because of changes in
2764 internal representation of (OR ..) types to accommodate the new
2765 support for (AND ..) types, among other things)
2767 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2768 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2770 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2772 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2773 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2774 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2775 * DESCRIBE now works on structure objects again.
2776 * Most function call argument type mismatches are now handled as
2777 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2778 know whether the function will be redefined before the call is
2779 executed. (The compiler could flag local calls with full WARNINGs,
2780 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2781 it doesn't keep track of enough information to know whether calls
2782 are local in this sense.)
2783 * Compiler output is now more verbose, with messages truncated
2784 later than before. (There should be some supported way for users
2785 to override the default verbosity, but I haven't decided how to
2786 provide it yet, so this behavior is still controlled by the internal
2787 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2788 src/compiler/ir1util.lisp.)
2789 * Fasl file format version numbers have increased again, because
2790 support for the Gray streams extension changes the layout of the
2791 system's STREAM objects.
2792 * The Gray subclassable streams extension now works, thanks to a
2793 patch from Martin Atzmueller.
2794 * The full LOAD-FOREIGN extension (not just the primitive
2795 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2796 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2797 but like most other programs, it defaults to copying the Unix
2798 environment from the original process instead of starting the
2799 new process in an empty environment.
2800 * Extensions which manipulate the Unix environment now support
2801 an :ENVIRONMENT keyword option which doesn't smash case or
2802 do other bad things. The CMU-CL-style :ENV option is retained
2803 for porting convenience.
2804 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2805 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2807 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2809 * A patch from Martin Atzmueller seems to have solved the SIGINT
2810 problem, and as far as we know, signal-handling now works cleanly.
2811 (If you find any new bugs, please report them!)
2812 * The system no longer defaults Lisp source file names to types
2813 ".l", ".cl", or ".lsp", but only to ".lisp".
2814 * The compiler no longer uses special default file extensions for
2815 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2816 seems to expect a single default extension for all compiled code,
2817 and there's no compelling reason to try to stretch the standard
2818 to allow two different extensions.) Instead, byte-compiled files
2819 default to the same extension as native-compiled files.
2820 * Fasl file format version numbers have increased again, because
2821 a rearrangement of internal implementation packages made some
2822 dumped symbols in old fasl files unreadable in new cores.
2823 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2824 many fewer weird special cases.
2825 * Bug #17 (differing COMPILE-FILE behavior between logical and
2826 physical pathnames) has been fixed, and some related misbehavior too,
2827 thanks to a patch from Martin Atzmueller.
2828 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2829 by Tim Moore, ported to SBCL by Martin Atzmueller.
2830 * Martin Atzmueller fixed several filesystem-related problems,
2831 including bug #36, in part by porting CMU CL patches, which were
2832 written in part by Paul Werkowski.
2833 * More compiler warnings in src/runtime/ are gone, thanks to
2834 more patches from Martin Atzmueller.
2835 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2838 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2840 * DESCRIBE now works on CONDITION objects.
2841 * The debugger now handles errors which arise when trying to print
2842 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2844 * The build system now uses an additional file, customize-target-features.lisp,
2845 to allow local modifications to the target *FEATURES* list. (The point of
2846 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2847 debugging features enabled, without having to worry about propagating it
2848 into everyone's system when I do a "cvs update".) When no
2849 customize-target-features.lisp file exists, the target *FEATURES* list
2850 should be constructed the same way as before.
2851 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2852 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2853 warn about behavior which is undefined under the ANSI spec. Note: This
2854 is specified by ANSI, but it's not very popular with programmers.
2855 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2856 macro in the SBCL sources for an example of a workaround which you
2857 might use to make portable ANSI-standard code which does what you want.
2858 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2859 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2860 when it's used as a non-toplevel form. (This is required in order
2861 to implement the DEFCONSTANT-EQX macro.)
2862 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2863 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2864 funny things, due to bugs in the implementation of EVAL-WHEN
2865 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2866 non-toplevel DEFCONSTANTs are uncommon.)
2867 * The core file version number and fasl file version number have been
2868 incremented, because the old noncompliant DEFCONSTANT behavior involved
2869 calling functions which no longer exist, and because I also took the
2870 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2871 * fixed bug 1 (error handling before read-eval-print loop starts), and
2872 redid debugger restarts and related debugger commands somewhat while
2874 ** The QUIT debugger command is gone, since it did something
2875 rather different than the SB-EXT:QUIT command, and since it never
2876 worked properly outside the main toplevel read/eval/print loop.
2877 Invoking the new TOPLEVEL restart provides the same functionality.
2878 ** The GO debugger command is also gone, since you can just invoke
2879 the CONTINUE restart directly instead.
2880 ** The TOP debugger command is also gone, since it's redundant with the
2881 FRAME 0 command, and since it interfered with abbreviations for the
2883 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2885 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2886 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2887 some porting work by Martin Atzmueller)
2888 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2889 with a boatload of patches which he ported to SBCL. Now that those
2891 ** The system tries to make sure that its low-priority messages
2892 are prefixed by semicolons, to help people who like to use
2893 syntax highlighting in their ILISP buffer. (This patch
2894 was originally due to Raymond Toy.)
2895 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2896 patches originally written by Raymond Toy.
2897 ** The compiler understands coercion between single-value and
2898 multiple-VALUES type expressions better, getting rid of some very
2899 weird behavior, thanks to patches originally by Robert MacLachlan
2900 and Douglas Crosher.
2901 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2902 lambda lists, thanks to a patch originally by Pierre Mai.
2903 ** The system no longer bogusly warns about "abbreviated type
2905 ** The compiler gets less confused by inlining and RETURN-FROM,
2906 thanks to some patches originally by Tim Moore.
2907 ** The system no longer hangs when dumping circular lists to fasl
2908 files, thanks to a patch originally from Douglas Crosher.
2909 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2910 undefined function error.
2911 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2912 in structure slot initforms) for now, documented workaround instead:-|
2913 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2914 * fixed bug 5: added stubs for various Gray stream functions called
2915 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2916 installed, at least appropriate type errors are generated
2917 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2918 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2919 consistently in DEFMETHOD forms.
2920 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2921 it doesn't seem to affect SBCL after all
2922 * The C runtime system now builds with better optimization and many
2923 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2925 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2927 * The system is now under CVS at SourceForge (instead of the
2928 CVS repository on my home machine).
2929 * The new signal handling code has been tweaked to treat register
2930 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2931 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2932 downstream, e.g. in debug-int.lisp, has implicit dependencies
2933 on the unsignedness of integer representation of machine words,
2934 and that caused the system to bomb out with infinite regress
2935 when trying to recover from type errors involving signed values,
2936 e.g. (BUTLAST '(1 2 3) -1).)
2937 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2938 (This was one of the bugs Peter Van Eynde reported back in July.)
2939 * The system now uses code inspired by Colin Walters' O(N)
2940 implementation of MAP (from the cmucl-imp@cons.org mailing
2941 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2942 inline the MAP operation, and there is more than one
2943 sequence argument to the MAP call (so that it can't just
2944 do ETYPECASE once and for all based on the type of the
2945 single sequence argument). (The old non-inline implementation
2946 of the general M-argument sequence-of-length-N case required
2947 O(M*N*N) time when any of the sequence arguments were LISTs.)
2948 * The QUIT :UNIX-CODE keyword argument has been renamed to
2949 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2950 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2952 * Raymond Wiker's patches to port dynamic loading from Linux to
2953 FreeBSD have been added.
2954 * The BUGS file is now more nearly up to date, thanks in large part
2955 to Martin Atzmueller's review of it.
2956 * The debugger now flushes standard output streams before it begins
2957 its output ("debugger invoked" and so forth).
2958 * The core version number and fasl file version number have both
2959 been incremented, because of incompatible changes in the layout
2961 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2962 used to not be called for a saved Lisp image.)
2963 * Martin Atzmueller's version of a patch to fix a compiler crash,
2964 as posted on sbcl-devel 13 September 2000, has been installed.
2965 * Instead of installing Martin Atzmueller's patch for the
2966 compiler transform for SUBSEQ, I deleted the compiler transform,
2967 and transforms for some similar consing operations.
2968 * A bug in signal handling which kept TRACE from working on OpenBSD
2970 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2971 away by constant folding
2972 * The system now defines its address space constants in one place
2973 (in the Lisp sources), and propagates them automatically elsewhere
2974 (through GENESIS and the sbcl.h file). Therefore, patching the
2975 address map is less unnecessarily tedious and error-prone. The
2976 Lisp names of address space constants have also been systematized.
2977 * CVS tags like dollar-Header-dollar have been removed from
2978 the sources, because they have never saved me trouble and
2979 they've been source of trouble working with patches and other
2980 diff-related operations.
2981 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2982 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2984 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2986 * The system has been ported to OpenBSD.
2987 * The system now compiles with a simple "sh make.sh" on the systems
2988 that it's supported on. I.e., now you no longer need to tweak
2989 text in the target-features.lisp-expr and symlinks in src/runtime/
2990 by hand, the make.sh takes care of it for you.
2991 * The system is no longer so grossly inefficient when compiling code
2992 involving vectors implemented as general (not simple) vectors (VECTOR T),
2993 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2994 which dares to use the various sequence functions on non-simple
2995 vectors, takes less of a performance hit.
2996 * There is now a primitive type predicate VECTOR-T-P
2997 to test for the (VECTOR T) type, so that e.g.
2998 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2999 can now be compiled with some semblance of efficiency. (The old code
3000 turned the type declaration into a full call to %TYPEP at runtime!)
3001 * AREF on (VECTOR T) is still not fast, since it's still compiled
3002 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3003 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3004 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3005 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3006 without making full calls to SUBTYPEP at runtime.
3007 (Some analogous efficiency issues for non-simple vectors specialized to
3008 element types other than T, or for non-simple multidimensional arrays,
3009 have not been addressed. They could almost certainly be handled the
3010 same way if anyone is motivated to do so.)
3011 * The changes in array handling break binary compatibility, so
3012 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3013 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3014 instead of (VALUES T T).
3015 * By following the instructions that Dan Barlow posted to sbcl-devel
3016 on 2 July 2000, I was able to enable primitive dynamic object
3017 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3018 functionality is not implemented (since it calls ld to resolve
3019 library references automatically, requiring RUN-PROGRAM for its
3020 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3021 try to resolve library references) is now supported.
3022 * The system now flushes the standard output streams when it terminates,
3023 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3024 them at several other probably-convenient times, e.g. in each pass of
3025 the toplevel read-eval-print loop, and after evaluating a form given
3026 as an "--eval" command-line option. (These changes were motivated by a
3027 discussion of stream flushing issues on cmucl-imp in August 2000.)
3028 * The source transform for TYPEP of array types no longer assumes
3029 that an array whose element type is a not-yet-defined type
3030 is implemented as an array of T, but instead punts, so that the
3031 type will be interpreted at runtime.
3032 * There is now some support for cross-compiling in make.sh: each of
3033 the phases of make.sh has its own script. (This should be transparent
3034 to people doing ordinary, non-cross-compile builds.)
3035 * Since my laptop doesn't have hundreds of megabytes of memory like
3036 my desktop machine, I became more motivated to do some items on
3037 my to-do list in order to reduce the size of the system a little:
3038 ** Arrange for various needed-only-at-cold-init things to be
3039 uninterned after cold init. To support this, those things have
3040 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3041 symbols with such names are now uninterned after cold init).
3042 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3043 fasl files for cold load.
3044 ** Remove the old compiler structure pooling code (which used to
3045 be conditional on the target feature :SB-ALLOC) completely.
3046 ** Redo the representation of some data in cold init to be more compact.
3047 (I also looked into supporting byte compiled code at bootstrap time,
3048 which would probably reduce the size of the system a lot, but that
3049 looked too complicated, so I punted for now.)
3050 * The maximum signal nesting depth in the src/runtime/ support code has
3051 been reduced from 4096 to 256. (I don't know any reason for the very
3052 large old value. If the new smaller value turns out to break something,
3053 I'll probably just bump it back up.)
3054 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3056 * Many, many bugs reported by Peter Van Eynde have been added to
3057 the BUGS list; some have even been fixed.
3058 * While enabling dynamic object file loading, I tried to make the
3059 code easier to understand, renaming various functions and variables
3060 with less ambiguous names, and changing some function calling
3061 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3062 * While trying to figure out how to do the OpenBSD port, I tried to
3063 clean up some of the code in src/runtime/. In particular, I dropped
3064 support for non-POSIX signal handling, added various comments,
3065 tweaked the code to reduce the number of compilation warnings, and
3066 renamed some files to increase consistency.
3067 * To support the new automatic configuration functionality in make.sh,
3068 the source file target-features.lisp-expr has been replaced with the
3069 source file base-target-features.lisp-expr and the machine-generated
3070 file local-target-features.lisp-expr.
3071 * fixed a stupid quoting error in make.sh so that using CMU CL
3072 "lisp -batch" as cross-compilation host works again
3074 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3076 * DESCRIBE no longer tries to call itself recursively to describe
3077 bound/fbound values, so that it no longer fails on symbols which are
3078 bound to themselves (like keywords, T, and NIL).
3079 * DESCRIBE now works on generic functions.
3080 * The printer now prints less-screwed-up representations of closures
3081 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3082 * A private symbol is used instead of the :EMPTY keyword previously
3083 used to mark empty slots in hash tables. Thus
3084 (DEFVAR *HT* (MAKE-HASH-TABLE))
3085 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3086 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3087 now does what ANSI says that it should. (You can still get
3088 similar noncompliant behavior if bang on the hash table
3089 implementation with all the symbols you get back from
3090 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3091 This breaks binary compatibility, since tests for equality to
3092 :EMPTY are wired into things like the macroexpansion of
3093 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3095 * There's now a minimal placeholder implementation for CL:STEP,
3096 as required by ANSI.
3097 * An obscure bug in the interaction of the normal compiler, the byte
3098 compiler, inlining, and structure predicates has been patched
3099 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3100 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3102 * Missing ordinary arguments in a macro call are now detected even
3103 when the macro lambda list contains &KEY or &REST.
3104 * The debugger no longer complains about encountering the top of the
3105 stack when you type "FRAME 0" to explicitly instruct it to go to
3106 the top of the stack. And it now prints the frame you request even
3107 if it's the current frame (instead of saying "You are here.").
3108 * As specified by ANSI, the system now always prints keywords
3109 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3111 * The default initial SIZE of HASH-TABLEs is now smaller.
3112 * Type information from CLOS class dispatch is now propagated
3113 into DEFMETHOD bodies, so that e.g.
3114 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3116 is now basically equivalent to
3117 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3118 (DECLARE (TYPE SINGLE-FLOAT X))
3120 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3121 operation, without having to do run-time type dispatch.
3122 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3123 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3124 * Since I don't seem to be making big file reorganizations very often
3125 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3126 is overflowing my ability to conveniently back them up), I've finally
3127 checked the system into CVS. (The CVS repository is on my home system,
3128 not at SourceForge -- putting it on SourceForge might come later.)
3129 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3130 high-level GC-NOTIFY-FOO functions send their output. (There's
3131 still very little control of where low-level verbose GC functions
3132 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3133 less than it used to -- the GC-NOTIFY-FOO functions are now under
3134 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3135 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3136 in only one place in the source code, and propagates it automatically
3137 everywhere that it's needed. Thus e.g. when I bump the version from
3138 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3139 * The C source files now include boilerplate legalese and documentation
3140 at the head of each file (just as the Lisp source files already did).
3141 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3142 to his page will be replaced with a link to his new CLiki service.
3144 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3146 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3147 * The build process now looks for GNU make under the default name "gmake",
3148 instead of "make" as it used to. If GNU make is not available as "gmake"
3149 on your system, you can change this default behavior by setting the
3150 GNUMAKE environment variable.
3151 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3152 can build without error under CMU CL.
3154 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3156 * There is now a partial SBCL user manual (with some new text and some
3157 text cribbed from the CMU CL manual).
3158 * The beginnings of a profiler have been added (starting with the
3159 CMU CL profiler and simplifying and cleaning up). Eventually the
3160 main interface should be through the TRACE macro, but for now,
3161 it's still accessed through vaguely CMU-CL-style functions and macros
3162 exported from the package SB-PROFILE.
3163 * Some problems left over from porting CMU CL to the new
3164 cross-compilation bootstrap process have been cleaned up:
3165 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3166 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3167 part when getting rid of the compiler *BACKEND* stuff.)
3168 ** The value of *NULL-TYPE* was screwed up, because it was
3169 being initialized before the type system knew the final
3170 definition of the 'NULL type. This screwed up several key
3171 optimizations in the compiler, causing inefficiency in all sorts
3172 of places. (I found it because I wanted to understand why
3173 GET-INTERNAL-RUN-TIME was consing.)
3174 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3175 PROCLAIM FTYPE information. Unfortunately this broke binary
3176 compatibility again, since now the forms output by DEFGENERIC
3177 to refer to functions which didn't exist in 0.6.3.
3178 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3179 can use the new (as of 0.6.3) transform for SOME into MAP into
3181 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3182 values to (INTEGER 0 1000000), so that the time code will no longer
3183 occasionally get blown up by Linux returning 1000000 microseconds
3184 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3185 its output conform to the ANSI spec. (Alas, this makes its output
3186 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3187 * A full call to MAP NIL with a single sequence argument no longer conses.
3188 * fixes to problems pointed out by Martin Atzmueller:
3189 * The manual page no longer talks about multiprocessing as though
3190 it were currently supported.
3191 * The ILISP support patches have been removed from the distribution,
3192 because as of version 5.10.1, ILISP now supports SBCL without us
3193 having to maintain patches.
3194 * added a modified version of Raymond Toy's recent CMU CL patch for
3195 EQUALP comparison of HASH-TABLE
3197 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3199 * The system still can't cross-compile itself with
3200 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3201 entails), but at least it can compile more of itself that way
3202 than it used to be able to, and various buglets which were uncovered
3203 by trying to cross-compile itself that way have now been fixed.
3204 * This release breaks binary compatibility again. This time
3205 at least I've incremented the FASL file format version to 2, so that the
3206 problem can be detected reliably instead of just causing weird errors.
3207 * various new style warnings:
3208 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3209 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3210 ** using the lexical binding of a variable named in the *FOO* style
3211 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3212 as specified by ANSI.
3213 * *RANDOM-STATE* is no longer automatically initialized from
3214 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3215 default behavior of the system is to repeat its behavior every time
3216 it's run. If you'd like to change this behavior, you can always
3217 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3218 old convention there was no comparably easy way to get the system to
3219 repeat its behavior every time it was run.
3220 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3221 FTYPE declarations has been removed, in favor of their ANSI
3222 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3223 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3224 the types of their sequence arguments can be determined at compile time.
3225 This is done through a new open code expansion for MAP which eliminates
3226 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3227 argument types can be determined at compile time.
3228 * The optimizer now transforms COERCE into an identity operation when it
3229 can prove that the coerced object is already of the correct type. (This
3230 can be a win for machine generated code, including the output of other
3231 optimization transforms, such as the MAP transform above.)
3232 * Credit information has been moved from source file headers into CREDITS.
3233 * Source file headers have been made more standard.
3234 * The CASE macro now compiles without complaining even when it has
3237 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3239 * (Note that the way that the PCL macroexpansions were rewritten
3240 to accommodate the change in DEFGENERIC below breaks binary
3241 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3242 not run under sbcl-0.6.2. Once we get out of alpha releases,
3243 i.e. hit release 1.0.0, we'll probably try to maintain binary
3244 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3245 and sbcl-1.4.4. Until then, however, it might be fairly common
3246 for maintenance releases to break binary compatibility.)
3247 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3249 * The reporting of unhandled signals has been changed to print some
3250 explanatory text as well as the report form. (Previously only
3251 the report form was printed.)
3252 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3253 it defines, so that the compiler no longer issues undefined function
3254 warnings for compiled-but-not-yet-loaded generic functions.
3255 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3256 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3257 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3258 section of the ANSI spec are supported.
3259 * The "" nickname for the "KEYWORD" package has been removed.
3260 The reader still handles symbol tokens which begin with a package marker
3261 as keywords, but it doesn't expose its mechanism for doing so in the
3262 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3263 * The system now issues STYLE-WARNINGs for contradictory TYPE
3264 proclamations. (Warnings for contradictory FTYPE proclamations would
3265 be nice too, but those can't be done usefully unless the type system
3266 is made smarter about FUNCTION types.)
3267 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3268 systematized, so that "*target-*.lisp is supposed to exist only on the
3269 target and imply that there's a related file which exists on the
3270 host, and *host-*.lisp is supposed to exist only on the host and imply
3271 that there's a related file which exists on the target. This involves a
3272 lot of renaming. Hopefully the acute confusion caused by the renaming
3273 will be justified by the reduction in chronic confusion..
3274 ** runtime-type.lisp -> early-target-type.lisp
3275 ** target-type.lisp -> late-target-type.lisp
3276 ** early-host-format.lisp -> early-format.lisp
3277 ** late-host-format.lisp -> late-format.lisp
3278 ** host-error.lisp -> misc-error.lisp
3279 ** early-error.lisp -> early-target-error.lisp
3280 ** late-error.lisp -> late-target-error.lisp
3281 ** host-defboot.lisp -> early-defboot.lisp
3282 ** code/misc.lisp -> code/target-misc.lisp
3283 ** code/host-misc.lisp -> code/misc.lisp
3284 ** code/numbers.lisp -> code/target-numbers.lisp
3285 ** code/early-numbers.lisp -> numbers.lisp
3286 ** early-host-type.lisp -> early-type.lisp
3287 ** late-host-type.lisp -> late-type.lisp
3288 ** host-typep.lisp -> typep.lisp
3289 ** load.lisp -> target-load.lisp
3290 ** host-load.lisp -> load.lisp
3291 ** host-disassem.lisp -> disassem.lisp
3292 ** host-insts.lisp -> insts.lisp
3293 ** byte-comp.lisp -> target-byte-comp.lisp
3294 ** host-byte-comp.lisp -> byte-comp.lisp
3295 ** host-signal.lisp -> signal.lisp
3296 ** host-defstruct.lisp -> defstruct.lisp
3297 ** late-target-type.lisp -> deftypes-for-target.lisp
3298 Furthermore, several other previously target-only files foo.lisp (e.g.
3299 hash-table.lisp and random.lisp) have been split into a target-and-host
3300 foo.lisp file and a target-only target-foo.lisp file, with their key type
3301 definitions in the target-and-host part, so that the cross-compiler will
3302 know more about target types.
3303 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3304 gone away. In their place are various *BACKEND-FOO* variables
3305 corresponding to the slots of the old structure.
3306 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3307 into the target SBCL has been fixed.
3308 * The chill.lisp system for loading cold code into a running SBCL
3310 * Support for the CMU CL "scavenger hook" extension has been removed.
3311 (It was undocumented and unused in the CMU CL sources that SBCL was
3312 derived from, and stale in sbcl-0.6.1.)
3313 * Various errors in the cross-compiler type system were detected
3314 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3315 (enabling various consistency checks). Many of them were fixed,
3316 but some hard problems remain, so the compiler is back to
3317 running without *TYPE-SYSTEM-INITIALIZED* for now.
3318 * As part of the cross-compiler type system cleanup, I implemented
3319 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3320 * I have started adding UNCROSS calls throughout the type system
3321 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3322 on cross-compiler input files will be able to go away, and various
3324 * CONSTANTP now returns true for quoted forms (as explicitly required
3327 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3329 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3330 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3331 (MAKE-LIST -1). In the longer term, it ought to become true
3332 that declarations are assertions even at SAFETY 1. For now, it's not
3333 quite true even at SAFETY 3, but it's at least more nearly true..
3334 (Note that this change seems to increases the size of the system by
3335 O(5%) and to decrease the speed of the compiler by 20% or more.)
3336 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3337 for the problem of printing dozens of lines of distracting information
3338 about low-level system machinery as part of the top stack frame
3339 on entry to the debugger when an undefined function was called.
3340 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3341 is not reset to COMMON-LISP-USER.
3342 * Compilation of stuff related to dyncount.lisp has been made conditional
3343 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3344 smaller. The various dyncount-related symbols have been moved into
3345 a new "SB-DYNCOUNT" package.
3346 * tty-inspect.lisp has been renamed to inspect.lisp.
3347 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3348 feature has gone away. (When we eventually port to other flavors of
3349 libc and/or Unix, we'll try to make the differences between flavors
3350 invisible at the user level.)
3351 * Various other *FEATURES* tags, and/or their associated conditionals,
3352 have been removed if obsolescent, or given better documentation, or
3353 sometimes given more-mnemonic names.
3355 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3357 * tidied up "make.sh" script
3358 * tidied up system directory structure
3359 * better "clean.sh" behavior
3360 * added doc/FOR-CMUCL-DEVELOPERS
3361 * many many small tweaks to output format, e.g. removing possibly-confusing
3362 trailing #\. character in DESCRIBE-INSTANCE
3363 * (EQUALP #\A 'A) no longer signals an error.
3364 * new hashing code, including EQUALP hashing
3365 * tidied up Lisp initialization and toplevel
3366 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3367 * command line argument processing
3368 * added POSIX-GETENV function to deal with Unix-ish environment variables
3369 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3370 terminating SBCL on EOF
3371 * non-verbose GC by default
3372 * There is no more "sbcl" shell script; the sbcl file is now the C
3373 runtime executable (just like CMU CL).
3374 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3376 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3377 symbol and package information as Lisp native symbol and package objects
3378 instead of strings naming symbols and strings naming packages. This way,
3379 whenever packages are renamed (as in warm init), debug information is
3380 transformed along with everything else.
3381 * tweaked the optimization policy declarations which control the building
3382 of SBCL itself. Now, among other things, the system no longer saves
3383 source location debugging information. (This helps two problems at once
3384 by reducing SBCL size and by keeping SBCL from trying to look for its
3385 sources -- which may not exist -- when reporting errors.)
3386 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3387 debugging and testing purposes
3388 * cleaned up printing, making the printer call PRINT-OBJECT for
3389 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3390 methods, giving nearly-ANSI behavior
3391 * converted almost all special variables to use *FOO* naming convention
3392 * deleted PARSE-TIME functionality, since it can be done portably
3393 * moved some files out of cold init into warm init
3394 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3396 * regularized formatting of source files
3397 * added an install.sh script
3398 * fixed ridiculous memory usage of cross-compiler by making
3399 compiler/alloc.lisp not try to do pooling unless it can hook
3400 itself into the GC of the cross-compilation host. Now the system
3401 builds nicely on my old laptop.
3402 * added :SB-ALLOC in target-features.lisp-expr
3403 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3404 was not implemented)
3405 * re-did condition handling and note reporting in the compiler. Notes
3406 are no longer handled by signalling conditions. Style warnings
3407 and warnings are handled more correctly and reported in such a way
3408 that it's easy to find one or the other in your output (so that you
3409 can e.g. figure out which of many problems caused COMPILE-FILE to
3411 * changed the severity of several compiler warnings from full WARNING
3412 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3413 compiler note reporting so that it doesn't use the condition system
3414 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3415 COMPILE-FILE command)
3416 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3417 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3418 no longer supported, since their implementation was deeply intertwingled
3419 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3420 * removed bogus "support" for compiler macros named (SETF FOO), and
3421 removed the compiler macro for SETF INFO (but only after making a fool
3422 of myself on the cmucl-imp mailing list by posting a bogus patch for
3423 DEFINE-COMPILER-MACRO..)
3424 * Compiled files containing forms which have side effects on the Lisp
3425 reader (such as DEFPACKAGE forms) are now handled more correctly.
3426 (Compiler queuing of top level lambdas has been suppressed by setting
3427 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3428 * deleted various currently-unused source files, e.g. gengc.lisp. They
3429 may be added back at some point e.g. when porting to other architectures,
3430 but until they are it's distracting to distribute them and to try to
3432 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3433 there were so many of them they're just distractions, and UNCROSS is
3434 known to be able to handle the current sources
3435 * moved PROFILE functionality into TRACE, so that it will be clear
3436 how the wrapping and unwrapping of functions when you profile them
3437 interacts with the wrapping and unwrapping of functions when you
3438 trace them. (Actually, the functionality isn't there yet, but at least
3439 the interface specification is there. Hopefully, the functionality will
3440 arrive with some future maintenance release.)
3441 * removed host-oops.lisp
3442 * changed signature of QUIT function to allow UNIX-CODE argument
3443 * fixed READ-SEQUENCE bug
3444 * tweaked verbose GC output so that it looks more like the progress
3445 output that ANSI specifies for functions like LOAD
3446 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3447 * added <http://sbcl.sourceforge.com> to the banner information printed by