1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
4 ** bug fix: parent thread now can be gc'ed even with a live
7 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
8 * new port: the Solaris operating system on x86 processors is now
9 mostly supported, though some rough edges in the environment
10 remain. (thanks to Daisuke Homma)
11 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
12 space for its heap. (thanks to Thiemo Seufer)
13 * optimization: SBCL on MIPS platforms now supports dynamic-extent
14 closures. (thanks to Thiemo Seufer)
15 * minor incompatible change: eof selects abort in the debugger.
16 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
17 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
19 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
20 deprecated, and will go away in a future revision of SBCL.
21 * minor incompatible change: GC-ON and GC-OFF are no longer
22 implemented with a counter, it does not matter how many times gc
24 * bug fix: discriminating functions for generic function classes
25 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
26 make invalid assumptions about method precedence order. (reported
28 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
29 platforms. (thanks to Thiemo Seufer)
30 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
31 * bug fix: correct pathname printing with printer escaping is on.
32 (thanks to Kevin Reid)
33 * bug fix: complex VOP definitions in "user-space" no longer trigger
34 package locks. (reported by Zach Beane)
35 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
36 for funcallable-instance functions, and is no different from
38 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
39 characters in the current readtable. (reported by Nicholas Neuss)
40 * bug fix: SUBTYPEP on various CONS types returns more a more
41 accurate acknowledgment of its certainty.
42 * optimizations: REMOVE-DUPLICATES now runs in linear time on
43 lists in some cases. This partially fixes bug 384.
44 * flush all standard streams before prompting in the REPL and the
46 * bug fix: signal handling and triggering gc do not conflict
47 directly anymore, in particular a high frequency sb-sprof does
48 not prevent gc from running
49 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
50 approximation for timezone and DST information between the
51 universal time epoch and the smallest negative 32-bit time_t.
52 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
53 year 1899 so long as the encoded time is non-negative.
54 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
55 the correct answer for non-constant rotate amounts when performing
56 an inline 32-bit rotation.
58 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
59 there is only one thread in the session
60 ** bug fix: memory leak for streams created in one thread and
62 ** bug fix: lockup when compiled with gcc4
63 ** bug fix: race that allows the gc to be triggered when gc is
65 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
66 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
68 ** bug fix: finished threads are gc'ed properly
69 * fixed some bugs revealed by Paul Dietz' test suite:
70 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
71 has arguments with a different set of applicable methods from
72 the orignal arguments.
73 ** The type error thrown by MAP now has a correct expected-type
75 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
76 name a compiled function.
77 ** Three MISC tests where a large bignum was improperly coerced to
78 a float that couldn't represent that bignum during type
79 derivation were fixed.
80 ** SUBTYPEP can now handle types involving the negation of a
81 list-form FUNCTION type.
82 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
83 as (COMPLEX (AND RATIO (NOT FIXNUM))).
84 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
86 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
87 * New feature: Experimental support for bivalent streams: streams
88 opened with :element-type :default now allow character and binary
90 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
91 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
93 * bug fix: no more highly sporadic "couldn't check whether ~S is
94 readable" when reading a stream and an interrupt hits in the middle
95 of a select system call
96 * compiler better recognizes complex arrays (reported by Eduardo
98 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
99 for variables on SBCL built with linkage-tables. (reported by Luis
101 * various error reporting improvements.
102 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
103 (Thanks to Thiemo Seufer)
104 * bug fix: nested reader invokations work correctly
105 * bug fix: it is possible to have simultaneous references to foreign
106 code and foreign data with the same name.
108 ** added x86-64 support
109 ** incompatible change: the threading api now works with thread
110 objects instead of thread ids
111 ** bug fix: threads are protected from signals and interruption when
112 starting up or going down
113 ** bug fix: a race where an exiting thread could lose its stack to gc
114 ** bug fix: don't halt on infinite error in threads if possible
115 ** fixed numerous gc deadlocks introduced in the pthread merge
116 ** bug fix: fixed thread safety issues in read and print
117 ** bug fix: debugger doesn't hang on session lock if interrupted at
118 an inappropriate moment
119 ** bug fix: run-program is now thread safe(r)
120 ** bug fix: inner with-recursive-lock no longer releases the mutex
121 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
122 * fixed some bugs revealed by Paul Dietz' test suite:
123 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
124 EXPECTED-TYPE slots filled.
125 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
127 ** MAKE-CONDITION accepts classes as type-designators.
128 ** COMPILE may never return NIL.
129 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
130 range before calling Unix time functions
132 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
133 * numerous signal handling fixes to increase stability
134 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
135 * minor incompatible change: we now correctly canonize default
136 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
137 as per the MOP, rather than the historical (INITARG INITFUNCTION
138 INITFORM). (reported by Bruno Haible)
139 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
140 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
141 for more information.
142 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
143 pathname is a directory pathname.
144 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
145 * dynamic space size on PPC has been increased to 768Mb. (thanks to
147 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
148 :METACLASS argument in addition to a class name. (reported by
149 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
150 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
151 (reported by Bruno Haible; patch from Gerd Moellmann)
152 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
154 * bug fix: more cleanups to the floating point exception handling on
155 x86-64 (thanks to James Knight)
156 * bug fix: the compiler does not try to constant fold calls of
157 COERCE and BIT-* functions when they return freshly constructed
158 objects. (reported by Nikodemus Siivola and Paul Dietz)
159 * optimization: improved the MIPS versions of generic (in the
160 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
161 * optimization: direct conversion of (unsigned-byte 32) to floats on
162 the PowerPC platform.
163 * optimization: structure instances with raw slots now use less
164 memory, and probably show better memory locality. (thanks to
166 * optimization: DYNAMIC-EXTENT declarations for lists and closures
167 are treated as requests for stack allocation on the x86-64,
168 Alpha-32, PPC and SPARC platforms.
169 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
170 the wrong file descriptor; implementation of SOCKET-OPEN-P.
171 (thanks to Tony Martinez)
173 ** gcing a dead thread can no longer lead to lockups
174 ** threads block signals until they are set up properly
175 ** errno is no longer shared by threads
176 ** interrupt-thread restores the eflags register on x86
177 ** fixed some lockups due to gc/thread interaction
178 * fixed some bugs revealed by Paul Dietz' test suite:
179 ** invalid dotted lists no longer raise a read error when
181 ** signal an error if a symbol that names a declaration is used
182 as the name of a type, or vice versa
183 ** allow using the (declare (typespec var*)) abbreviation for
184 (declare (type typespec var*)) with all type specifiers
185 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
186 ** always return NIL from PPRINT-POP when OBJECT is NIL
187 ** don't signal errors when pretty-printing malformed LABELS,
188 FLET or MACROLET forms
189 ** declarations in a DOLIST body are also in scope for the
191 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
193 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
196 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
197 * fixed cross-compiler leakages that prevented building a 32-bit
198 target with a 64-bit host compiler.
199 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
200 opened with :IF-EXISTS :OVERWRITE.
201 * fixed bug 281: error for an invalid qualifier in a short-form method
202 combination method is not signalled until the faulty method is called.
203 * bug fix: iteration variable type inferrer failed to deal with open
204 intervals. (reported by Alan Shields)
205 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
206 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
207 or more runtime options were provided to the sbcl binary.
208 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
209 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
210 * optimization: there's now a fast-path for fixnum arguments in the
211 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
212 * optimization: the code generated on x86-64 is more compact thanks
213 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
214 * bug fix: floating-point exception handling now partly works on
215 x86-64. (Thanks to James Knight)
216 * improvement to the MIPS backend from Thiemo Seufer: C-style
217 64-bit long long arguments and return values to alien functions
219 * fixed some bugs revealed by Paul Dietz' test suite:
220 ** the type-error signalled from WARN has a filled-in DATUM slot.
221 ** the type-error required when a stream is not associated with
222 a file has the stream as its datum.
223 ** type-errors on single-floats on x86-64 no longer have
224 :INVALID-OBJECT as the datum
225 ** the type-errors signalled for invalid function names now have
226 a correct expected type
227 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
228 for structures defined with a :TYPE. Documentation strings for
229 typed structures are no longer immediately discarded
230 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
231 broadcast streams and synonym streams. FILE-LENGTH now also works
232 on broadcast streams.
234 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
235 * incompatible change: the --noprogrammer option, deprecated since
236 version 0.7.5, has been removed. Please use the equivalent
237 --disable-debugger option instead.
238 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
240 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
241 inoperational for a while now) has been completely removed.
242 * null lexical environments are now printed as #<NULL-LEXENV>,
243 significantly reducing the amount of clutter in typical backtraces.
244 * documentation on weak pointers, finalization, and after GC hooks
245 has been added to the manual.
246 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
247 as other specialized array types, is much faster. SUBSEQ and
248 COPY-SEQ on such arrays have also been sped up.
249 * optimization: EQL is now more efficient when at least other argument
250 is known to be of type (OR FIXNUM (NOT NUMBER)).
251 * fixed bug: the runtime is now less vulnerable to changes in the
252 size of the SBCL object on OS X, and virtual memory is reserved for
253 all spaces that need to be at a fixed address.
254 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
255 * fixed bug: finalizers and after GC hooks that cause consing are now
257 * fixed bug: compiler error messages and summaries are now printed to
258 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
259 * fixed inference of the upper bound of an iteration variable.
260 (reported by Rajat Datta).
261 * fixed bug 211e: calling local functions with duplicated constant
262 keyword argument no longer causes a bogus style warning about an
264 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
265 ftype declaration to be disregarded. (reported by Dave Roberts)
266 * fixed bug 373: caused by erronous compilation of references to alien
267 variables in the runtime on ppc/darwin.
268 * fixed bug 376: CONJUGATE type deriver.
269 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
270 (reported by Baughn on #lisp)
271 * a cross-compiler bug on non-x86 platforms has been identified and
272 fixed. (thanks to Bruno Haible)
273 * improvements to the MIPS runtime code for increased stability.
274 (thanks to Thiemo Seufer)
275 * increased the maximimum compact environment size to allow
276 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
277 * partial workaround for bug 135: don't name the function we're
278 calling for hairy cases of EVAL, so as not to accumulate one environment
279 entry per call to EVAL. (reported by Kevin Reid)
280 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
281 * optimization: type testing for non-vector arrays should be faster.
282 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
284 * bug fix: setting 31st element of a bit vector to zero did not work
286 * bug fix: redefining a class definition which failed due to a
287 previous accessor / function clash now works (but see BUGS entry
288 #380 for more problems in this area). (thanks to Zach Beane)
289 * the long form of DEFINE-METHOD-COMBINATION disables method group
290 checking when given a single method group with pattern *. (thanks
292 * on x86 compiler supports stack allocation of results of simple
293 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
294 * support for building and running on Mac OS X 10.4 "Tiger" has been added
295 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
296 * fixed some bugs related to Unicode integration:
297 ** the restarts for recovering from input and output encoding
298 errors only appear when there is in fact such an error to
300 * fixed some bugs revealed by Paul Dietz' test suite:
301 ** MISC.549 and similar: late transformation of unsafe type
302 assertions into derived types caused unexpected code
304 ** SCALE-FLOAT type deriver is less wrong.
305 ** type derivers for EXP, LOG and similar functions compute result
306 types for complex arguments better.
307 ** (MISC.563) CONJUGATE type deriver works for very restricted
309 ** out-of-line type testers for character strings are available.
310 ** EQUAL compiler transform understands specialness of objects
312 ** accessing double-floats stored on the stack now works on x86-64.
313 ** debugger internals could sometimes create invalid lispobjs,
314 resulting in GC crashes.
315 ** MISC.548: type check weakening can convert required type into
317 ** initialization forms for bindings are not in scope of free special
320 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
321 * incompatible change: thread support for non-NPTL systems has
322 been removed - locking is buggy and unreliable. A threaded
323 SBCL build will now warn at startup and refuse to create
324 new threads, unless futex support is detected in the kernel
325 * incompatible change: the top level REPL now has only an ABORT
326 restart associated with it, not TOPLEVEL and ABORT as it used to.
327 TOP and TOPLEVEL are now available as debugger commands for
328 returning to the top level.
329 * incompatible change: forms evaluated in the REPL now use the
330 global optimization policy.
331 * incompatible change: user- and system-initialization files are
332 no longer processed with LOAD, but by READ and EVAL; hence the
333 global optimization policy, startup package, readtable, etc,
335 * The .fasl file format number has been incremented because of
336 various incompatible changes.
337 * internal entry point details and argument counts no longer appear
338 in backtraces unless explicitly requested by setting
339 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
340 * built-in and standard functions no longer have names like "top
341 level local call to FOO".
342 * fixed bug 32: functions defined in non-null lexical environments
343 now have more legible printed representation
344 * fixed bug 33: functions defined in non-null lexical environemnts
345 are now more amenable to inspection by INSPECT.
346 * workaround for bug 354: XEPs no longer appear in backtraces unless
347 explicitly requested.
348 * fixed bug: receiving the signal which results from attempting to
349 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
350 SIGBUS on other *BSDs) on architectures where the C stack is also
351 the Lisp stack (x86 and x86-64) from foreign code no longer leads
352 to debugger confusion or wild instability. (reported by Cheuksan
354 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
355 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
356 (reported by Lutz Euler)
357 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
358 compiled code. (reported by Paul Dietz)
359 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
360 specialized parameter to a method to something that is not TYPEP
361 the specializer is now possible.
362 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
363 face of package deletion.
364 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
365 pathname. (reported by David Tolpin on c.l.l)
366 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
367 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
368 STRING-STREAM and FILE-STREAM, respectively.
369 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
370 than previously; in addition, its test suite is now run on build.
371 (thanks to Luke Gorrie)
372 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
373 * optimization: added a immediate representation for single-floats
375 * fixed some bugs revealed by Paul Dietz' test suite:
376 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
377 ** attempting to create a package with a colliding nickname causes
378 correctable errors to be signalled.
379 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
380 ** several x86-64 backend bugs related to sign-extension of immediate
383 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
384 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
385 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
387 * fixed loading of multiply forward-referenced layouts.
388 (thanks to Cheuksan Wang)
389 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
390 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
391 (see COMPILE-FILE documentation for details of :PRINT option)
392 * fixed bugs 19 and 317: fixed-format floating point printing is
393 more accurate. This also fixes a bug reported by Adam Warner
394 related to the ~@F format directive.
395 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
397 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
398 dispatch table if the from-char is a dispatch macro character.
399 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
400 by the wordsize no longer ignore the last word. (reported by Lutz
402 * fixed bug in type checking of dynamic-extent variables. (reported
404 * optimization: sequence traversal functions use their freedom to
405 coerce function designators to functions.
406 * optimization: code with many calls to CLOS methods specialized on
407 CLOS classes has had redundant type checks removed.
408 * optimization: added declarations to speed up operations that access
409 the internal character database (for example STRING-UPCASE)
410 * optimization: comparison operations between floats and sufficiently small
411 fixnums no longer create extra rationals
412 * fixed some bugs related to Unicode integration:
413 ** portions of multibyte characters at the end of buffers for
414 character-based file input are correctly transferred to the
415 start of the buffer at the next read.
416 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
417 passing it through to OPEN.
418 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
419 argument given to internal calls to OPEN.
420 ** the built-in comment readers (introduced by character sequences
421 ";" and "#|") are more forgiving to encoding errors; they will
422 STYLE-WARN and then attempt to resync the stream at a character
423 boundary. (thanks to Teemu Kalvas)
424 * fixed some bugs in the x86-64 port:
425 ** Negative short int return values from c-calls are sign-extended
427 ** The stack is aligned to 16-bytes for c-calls, as required by
428 the ABI. (thanks to Cheuksan Wang)
429 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
430 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
431 for enums. (thanks to Vincent Arkesteijn)
432 ** Multiple small optimizations and bugfixes for floating point
434 * fixed some bugs revealed by Paul Dietz' test suite:
435 ** Space, Tab, Linefeed, Return and Page have the invalid
436 secondary constituent character trait.
437 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
439 ** WITH-INPUT-FROM-STRING should only update the index place on
441 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
443 ** Bit-array operations (BIT-AND and similar) worked incorrectly
444 with one-dimensional arrays with fill pointers.
445 ** TYPE-OF failed on a complex with an integer realpart and a
447 ** compiler failure during type inference for the code of form
448 (IF (EQL X (THE ...)) ...) (MISC.535).
450 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
451 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
452 platform. The port supports 61-bit fixnums, large memory spaces
453 and reloading shared object files.
454 * enhancement: saving cores with foreign code loaded is now
455 supported on ppc/Darwin in addition to the previously supported
457 * enhancement: the statistical profiler now walks deeper into the
458 call stack for more meaningful call-graphs and accrued time
459 reports (x86/x86-64 only). It also now reports time spent in
461 * enhancement: it is now possible to trace most individual methods
462 of a generic function in addition to tracing the generic function
464 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
466 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
467 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
468 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
469 produces an error. (thanks to Vincent Arkesteijn)
470 * bug fix: NAMESTRING on pathnames with :WILD components in their
471 directories works correctly. (thanks to Artem V. Andreev)
472 * fixed bug 125: compiler preserves identity of closures. (reported
474 * bug fixed: functions with &REST arguments sometimes failed with
475 "Undefined function" when compiled with (DEBUG 3). (reported by
477 * bug fix: overflow during compiling of setting element of a bit
478 vector with constant index and value. (reported by Timmy Douglas)
479 * build fix: fixed the dependence on *LOAD-PATHNAME* and
480 *COMPILE-FILE-PATHNAME* being absolute pathnames.
481 * on x86 compiler partially supports stack allocation of dynamic-extent
483 * GO and RETURN-FROM do not check the extent of their exit points
484 when compiled with SAFETY 0.
485 * fixed some bugs related to Unicode integration:
486 ** encoding and decoding errors are now much more robustly
487 handled; it should now be possible to recover even from invalid
488 input or output to the terminal. (thanks to Teemu Kalvas)
489 ** provided a first cut at implementing STRING-TO-OCTETS and
490 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
491 ** altered the SB-MD5 contributed module slightly, changing the
492 interface just enough for it to be supportable for builds where
493 lisp characters are not eight bits.
494 * fixed some bugs revealed by Paul Dietz' test suite:
495 ** the FORMATTER-generated functions for ~V[ conditionals require
496 the correct number of arguments.
497 ** READ-FROM-STRING returns the mandated second value when applied
498 to displaced strings.
499 ** the #\Rubout and #\Backspace characters are treated as invalid
500 constituent characters by the tokenizer.
502 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
503 * new feature: reloading changed shared object files with
504 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
505 * new feature: references to foreign variables and functions
506 can now be compiled and loaded before the corresponding shared
507 object file is loaded, as long as the foreign definitions are
508 available at runtime.
509 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
510 supported. (thanks to Dan Debertin)
511 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
512 just keywords. (thanks to Vincent Arkesteijn)
513 * fixed bug #331: structure-class instances corresponding to
514 DEFSTRUCT forms are now created eagerly.
515 * fixed bug #345: backtraces from calls to undefined functions work
516 on x86 as well. Related bug #61 is now also partially fixed on x86
517 (backtraces from throws to unknown catch tags.)
518 * bug fix: lambda-list parsing is now stricter vrt. order and number
519 of lambda-list keywords.
520 * bug fix: as specified by AMOP, an error is signalled if a
521 class-option appears multiple times in a DEFCLASS form. (reported
523 * bug fix: RANDOM can be compiled when the compiler derives the type
524 of its numeric argument as a disjoint set of small integers.
525 (reported by Paul Dietz)
526 * bug fix: starting a core saved with shared objects loaded when
527 those objects are not available no longer causes threaded SBCL to
528 hang. (reported by Sean Ross)
529 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
531 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
532 argument, which must always be NIL. (reported by Kalle Niemitalo)
533 * bug fix: printing 1.0d+23 no longer results in an error.
534 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
535 * bug fix: structure slot setters preserve evaluation order. (thanks
537 * fixed some bugs related to Unicode integration:
538 ** RUN-PROGRAM can allow its child to take input from a Lisp
539 stream. (reported by Stefan Scholl)
540 ** ASDF-INSTALL successfully downloads .tar.gz files in all
541 locales. (reported by Ken Causey)
542 * fixed some bugs revealed by Paul Dietz' test suite:
543 ** INCF, DECF and REMF evaluate their place form as specified in
545 ** FORMATTER expands ~{ ~} iteration directives with V or #
546 parameters correctly.
547 ** FORMATTER deals with the ~@[ ~] conditional directive where the
548 consequent uses no arguments correctly.
549 ** the system has a partial understanding of the (COMPLEX RATIO)
551 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
554 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
555 * new feature: a build-time option (controlled by the :SB-UNICODE
556 keyword feature, enabled by default) for building the system with
557 support for the entire 21-bit character space defined by the
559 * new feature: the system now has rudimentary external-format
560 support; the primary user-visible change at this time is that
561 characters with the high bit set (such as lower-case-e-acute) will
562 print correctly to a terminal in a UTF-8 environment.
563 * minor incompatible change: BASE-CHAR no longer names a class;
564 however, CHARACTER continues to do so, as required by ANSI.
565 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
566 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
567 has been moved to the SB-EXT package (temporarily re-exported from
569 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
570 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
572 * fixed bug #351: better error-handlind and reporting for malformed
574 * fixed bug #350: bignum-printing is now more memory-efficient,
575 allowing printing of very large bignums, eg. (expt 2 10000000).
576 (reported by Bruno Haible)
577 * fixed bug #302: better primitive-type selection for intersection
579 * fixed bug #308: non-graphic characters now all have names, as
580 required. (reported by Bruno Haible)
581 * bug fix: redefining a class with different superclasses now correctly
582 removes it from the direct-subclasses of its previous superclasses.
583 (reported by David Morse)
584 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
585 new value now works. (reported by Bruno Haible)
586 * bug fix: correct canonicalization of multiple non-standard slot
587 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
588 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
589 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
590 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
591 now exists, an signals an error.
592 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
593 during macroexpansion for non-iteration variables as well. (reported
594 by Bruno Haible for CMUCL)
595 * bug fix: Cyclic structures and unprintable objects in compiler
596 messages no longer cause errors. (reported by Bruno Haible)
597 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
598 and READ-FROM-STRING all now return a primary value of NIL if
599 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
600 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
601 (reported by Bruno Haible for CMUCL)
602 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
603 do not designate a condition. (reported by Bruno Haible for
605 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
606 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
607 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
608 permitting resolution in favour of any of the conflicting symbols.
609 (reported by Bruno Haible for CMUCL)
610 * bug fix: EQUAL compiler optimizations is less aggressive on
611 strings which can potentially compare true despite having distinct
612 specialized array element types.
613 * bug fix: unit enumerations can be defined without dividing by
614 zero. (thanks to Vincent Arkesteijn)
615 * FORMAT compile-time argument count checking has been enhanced.
616 (report from Bruno Haible for CMUCL)
617 * a partial workaround for the bug 262: the compiler does not try to
618 inline-expand a local function doing RETURN-FROM from a deleted
619 BLOCK. (thanks to Peter Denno for the bug report and to David
620 Wragg for the simple test case)
621 * fixed some bugs revealed by Paul Dietz' test suite:
622 ** RENAME-PACKAGE allows all package designators as new package
624 ** constraint propagation and lambda variable substitution are
625 more cautious in dealing with partially deleted code.
626 ** compiler sometimes lost reoptimization passes.
627 ** CERROR, given a condition as condition designator, passes its
628 remaining arguments to the continue format control without
630 ** Case-altering FORMAT directives work correctly on non-ASCII
632 ** The REAL type specifier handles bounds outside the floating
633 point ranges without signalling FLOATING-POINT-OVERFLOW.
634 ** Functions with IR1-transformations can create intercomponent
635 references to global functions.
636 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
638 ** FORMAT ~R treats a nil value for its first parameter correctly.
640 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
641 * enhancement: saving cores with foreign code loaded is now
642 supported on x86/NetBSD and sparc/Linux in addition to the previously
644 * bug fix: on some platforms repeated installations caused multiple
645 copies of HTML documentation to be installed -- should not happen
646 any more. (reported by Stefan Scholl)
647 * bug fix: parsing self-recursive alien record types multiple times
648 no longer causes infinite recursion. (reported by Thomas F. Burdick,
649 original patch by Helmut Eller for CMUCL)
650 * bug fix: stack-exhaustion detection works now on NetBSD as well.
651 (thanks to Richard Kreuter)
652 * bug fix: defining classes whose accessors are methods on existing
653 generic functions in other (locked) packages no longer signals
654 bogus package lock violations. (reported by François-René Rideau)
655 * bug fix: special variables as DEFMETHOD parameters no longer have
656 associated bogus type declarations. (reported by David Wragg and
658 * bug fix: read-write consistency on streams of element-type
659 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
660 * bug fix: redefiniton of the only method of a generic function with
661 no DEFGENERIC no longer emits a full WARNING. In addition,
662 redefinition of generic functions with no DEFGENERIC to an
663 incompatible lambda list now signals an error. (thanks to Zach
665 * bug fix: DEFGENERIC now works even when there's a function of the
666 same name in an enclosing lexical environment. (thanks to Zach
668 * fixed compiler failure, caused by instrumenting code during
669 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
670 * optimization: added loop analysis and improved register allocation
671 * fixed some bugs revealed by Paul Dietz' test suite:
672 ** POSITION on displaced vectors with non-zero displacement
673 returns the right answer.
674 ** (SIMPLE-STRING) is a valid type specifier for sequence
676 ** *PRINT-LEVEL* handling for slotless structures is pedantically
678 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
679 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
681 ** The justification version of the FORMAT ~< directive treats
682 non-zero minpad parameter correctly.
684 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
685 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
686 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
687 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
688 the supported interface.
689 * new feature: Single-stepping of code compiled with DEBUG 2 or
690 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
691 * new feature: saving cores with foreign code loaded is now
692 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
693 Timothy Moore's work for CMUCL)
694 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
695 parameters to * instead of NIL if no initform is supplied.
696 (reported by Johan Bockgaard)
697 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
698 correctly when FUNCALL appears as the car of the form. Note:
699 despite this FUNCALL forms are not currently subject to
700 compiler-macro expansion. (port of Raymond Toy's fix for the
701 same from CMUCL, reported by Johan Bockgaard)
702 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
703 (thanks for Teemu Kalvas)
704 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
705 case instead of terminating the iteration (thanks for Julian
706 Squires, Sean Champ and Raymond Toy)
707 * bug fix: incorrect expansion of defgeneric that caused a style
708 warning. (thanks for Zach Beane)
709 * bug fix: slot accessor effective method computation works properly
710 for classes with multiple non-standard applicable methods on
711 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
712 * on x86 compiler supports stack allocation of results of LIST and
713 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
714 CMUCL implementation by Gerd Moellmann)
715 * fixed some bugs revealed by Paul Dietz' test suite:
716 ** FORMAT strings with both the ~<~:;~> form of the justification
717 directive and pretty-printing directives cause an error.
719 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
720 * incompatible change: the internal functions
721 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
722 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
723 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
724 instead of the old functions.
725 * new feature: on platforms where "dladdr" is available foreign
726 function names now appear in backtraces. (based on Helmut Eller's
728 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
729 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
731 * documentation: ASDF manual is now distributed with SBCL.
732 * bug fix: SBCL can now load its contributed modules with REQUIRE
733 even if the system-provided entries have been removed from
734 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
736 * bug fix: backtraces involving undefined functions or assembly
737 routines are more informative. (thanks to Brian Downing)
738 * bug fix: mutually referent alien structures now work correctly.
739 (reported by Rick Taube)
740 * bug fix: structures defined by WITH-ALIEN can be referred to
741 within other definitions in the same WITH-ALIEN.
742 * bug fix: division operators (MOD, TRUNCATE and the like) with
743 constant zero divisors and integer dividends no longer generate
745 * bug fix: provide default methods for INPUT-STREAM-P and
746 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
747 * bug fix: improve the detection and resolution of MOP metacycles
748 (where effective method computation depends on the generic function
749 having its effective method computed). (reported by Bruno Haible)
750 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
751 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
752 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
753 * bug fix: pathnames with relative directory components can be
754 represented relative to default pathnames.
755 * optimization: in taking the GCD of bignums, reduce the two bignums
756 to approximately the same size (using Euclid's algorithm) before
757 applying the more sophisticated binary GCD. (thanks to Juho
759 * optimization: COUNT on bitvectors now operates word-at-a-time.
760 * optimization: ASH with a positive, but not necessarily constant,
761 (leftwards) shift, when in a modular context, is compiled to a
763 * fixed some bugs revealed by Paul Dietz' test suite:
764 ** FORMAT variable parameters ("~V<char>") are defaulted properly
765 if the corresponding argument is NIL.
766 ** FORMAT directives accepting a minpad parameter treat negative
768 ** literal commas as character parameters to FORMAT directives are
770 ** literal spaces directly after ~<Newline> directives within a
771 format-logical-block (~:< ~@:>) do not induce :FILL-style
772 conditional newlines.
773 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
774 as specified: it no longer includes conditional newlines).
775 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
777 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
778 :PER-LINE-PREFIX argument does not evaluate to a string.
780 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
781 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
782 the manual for details; for now, package locks can be disabled by
783 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
784 if no major problems are found then it is likely that they will be
785 compiled in unconditionally.
786 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
787 unsupported operators on all platforms. To load a shared library
788 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
789 object file, link it into a shared library outside of SBCL and
790 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
792 * bug workaround: ROOM T and the GC were not working together
793 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
794 reporting facility aren't preserved. That reporting has been
795 disabled, so now until and unless someone figures out how to make it
796 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
797 * minor incompatible change: as threatened around sbcl-0.8.0, the
798 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
799 The symbols are also exported from SB-PCL for backwards
800 compatibility, but more so than before SB-PCL should be treated as
801 an implementation-internal package.
802 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
804 * the SB-SPROF contrib now works on (most) non-x86 architectures.
805 It is known as of this release not to work on the Alpha, however.
806 * fixed bug #167: errors signalled due to illegal syntax in method
807 bodies are now more legible.
808 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
809 designators and can hence be used with TYPEP.
810 * fixed bug #333: CHECK-TYPE now ensures that the type error
811 signalled, if any, has the right object to be accessed by
812 TYPE-ERROR-DATUM. (reported by Tony Martinez)
813 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
814 5.1.2.3 for multiple-value place subforms. (reported by Kalle
816 * fixed bug #334: programmatic addition of slots using specialized
817 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
818 and :ALLOCATION :CLASS slots.
819 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
820 float exponents. (rereported by Peter Seibel)
821 * fixed a bug: #\Space (and other whitespace characters) are no
822 longer considered to be macro characters in standard syntax by
824 * fixed bug: initialization of condition class metaobjects no longer
825 causes an instance of the condition to be created. (reported by
827 * fixed bug: it is now possible to have slots such that
828 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
829 description is neither :INSTANCE nor :CLASS.
830 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
831 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
832 SLOT-VALUE-USING-CLASS), and no longer causes errors with
833 non-standard SLOT-DEFINITION-ALLOCATION values.
834 * fixed bugs: various race conditions handling exiting threads.
835 CL-PPCRE's thread torture test now passes.
836 * fixed arguably-a-bug: GC time proportional to number of threads in
837 system even when most of them are idle
838 * optimization: improved performance of BIT and SBIT on bit-vectors.
839 * .fasl file incompatibility: The fasl file version number has
840 been incremented because of changes associated with package locks.
842 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
843 * minor incompatible change: the system no longer provides
844 optimization hints (conditions which are TYPEP
845 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
846 mode; these hints are still emitted when higher SPEED optimization
848 * new contrib module: a sampling profiler (profiling by statistical
849 sampling, rather than by instrumenting functions) is available as
850 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
851 * the behaviour of the standard function ED is now customizeable by
852 third parties through a hook variable: see ED's documentation
853 string for information on the protocol.
854 * the compiler no longer emits efficiency notes for (FUNCALL X)
855 when the type of X is uncertain under default optimization
857 * fixed bug 276: mutating a binding of a specialized parameter to a
858 method to something that is not TYPEP the specializer is now
860 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
861 exhibit the required behaviour on the x86 platform. (thanks to
862 Peter van Eynde, Eric Marsden and Bruno Haible)
863 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
864 even for difficult arguments. (reported by Peter Graves)
865 * fixed bug 141a: the backquote printer now descends quoted
867 * fixed another bug in backquote printing: no more destructive
868 modification of the form's list structure. (reported by Brian
870 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
871 move between its address being taken and the call to
872 interrupt_thread, fixing a crashing race condition.
873 * the SB-POSIX contrib implementation has been adjusted so that it
874 no longer exhibits ridiculously poor performance when constructing
875 instances corresponding to C structs.
877 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
878 * minor incompatible change: the sb-grovel contrib now treats C
879 structures as alien (in the sense of SB-ALIEN) objects rather than
880 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
881 has implications for memory management of client code
882 (sb-grovel-returned objects must now be manually managed) and for
883 type safety (alien objects now have full types).
884 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
885 used to control emission of compiler diagnostics, rather than the
886 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
887 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
888 quality should be considered deprecated.
889 * install.sh now installs the user manual as well
890 * (not quite a new documentable feature, but worth considering in
891 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
892 beginnings of a semantically meaningful condition hierarchy is
893 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
895 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
896 designator as the defaults argument.
897 * fixed bug: Displaced arrays whose displaced-to array has become
898 too small now cause ARRAY-DIMENSION to signal an error, providing
899 for safer bounds-checking. (reported by Bruno Haible)
900 * fixed bug: DEFCLASS slot definitions with identical :READER and
901 :WRITER names now signal a reasonable error. (reported by Thomas
903 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
905 * fixed bug: Invalid slot specification errors now print correctly.
906 (thanks to Zach Beane)
907 * fixed bug 320: Shared to local slot value transfers in class
908 redefinitions now happen corrently from superclasses as
909 well. (reported by Bruno Haible)
910 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
912 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
913 improper lists correctly. (reported by Bruno Haible)
914 * fixed bug 313: source-transform for <fun-name> was erroneously
915 applied to a call of a value of a variable with name <fun-name>.
916 (reported by Antonio Menezes Leitao)
917 * fixed bug 307: The obsolete instance protocol ensures that
918 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
919 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
920 remove dead unknown-values globs from the middle of the stack.
921 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
922 easier to resume long computations after using *BREAK-ON-SIGNALS*
923 to diagnose and fix failures. (thanks to Nikodemus Siivola)
924 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
925 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
926 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
927 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
928 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
930 * fixed a (fairly theoretical) bug in string printing: if
931 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
932 does not have array-element-type equal to the most general string
934 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
935 function name. (thanks to Nikodemus Siivola)
936 * fixed bug: compile-time format string checker failed on
937 non-closed ~{. (reported by Thomas F Burdick)
938 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
939 don't warn on legal loop code involving more than one aggregate
940 boolean. (thanks to Nikodemus Siivola)
941 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
942 in keyword position to MAKE-PACKAGE was not being reported as
944 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
945 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
947 * fixed bugs 280 and 312: the checking for multiple definitions in a
948 file is less likely to become confused by uses of inline
950 * fixed bug: the #S reader macro performs the keyword coercion
951 specified for slot names. (reported by Kalle Niemitalo)
952 * fixed bug: lambda lists may contain symbols whose names start with
953 & but are not lambda-list-keywords; their occurrence triggers a
955 * fixed bug 321: define-method-combination argument lambda lists do
956 not cause invalid code to be generated when &optional or &aux
957 variables have default values. (reported by Bruno Haible)
958 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
959 have CLOS classes; this appears also to have fixed reported
960 instability in stack exhaustion detection.
961 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
962 MUFFLE-WARNING when no associated restart is present is now
964 * optimization: rearranged the expansion of various defining macros
965 so that each expands into only one top-level form in a
966 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
968 * optimization: used a previously little-used slot in symbols to
969 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
971 * fixed some bugs revealed by Paul Dietz' test suite:
972 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
974 ** Fixed an optimization bug related to inheritance of initforms
975 from local to shared slots.
976 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
977 ** CAST optimizer forgot to flush argument derived type.
978 ** print/read consistency on floats is now orders of magnitude
979 more likely. (thanks also to Bruno Haible for a similar report
981 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
982 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
983 for objects of type REAL. Make it so.
984 ** SXHASH is specified (infelicitously) to respect similarity,
985 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
986 it so. (thanks to Markus Ziegler)
987 ** on the Alpha, the compiler succeeds in compiling functions
988 returning a known number of arguments greater than 63.
989 ** fixed handling of invalid NIL arguments in keyword position
991 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
993 ** provided a readably-printable representation for RANDOM-STATE
995 ** ensured that pathnames, where they have a namestring, always
996 print using #P"..." syntax.
998 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
999 * Support for the forthcoming 2.0 version of the NetBSD kernel
1000 running on x86 hardware has been added. (thanks to Perry
1001 E. Metzger most immediately, and others for their past work)
1002 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1003 current release version; SBCL's OpenBSD support had been broken
1004 since about the time of OpenBSD's switch to ELF binary format.)
1005 * [placeholder for DX summary]
1006 ** user code with &REST lists declared dynamic-extent, under high
1007 speed or space and low safety and debug optimization policy.
1008 * The manual has been converted to Texinfo format and the debugger
1009 chapter from the cmucl manual has been added.
1010 * A facility has been added to extract documentation strings from
1011 sbcl and store them as Texinfo-formatted snippets for inclusion in
1012 the manual (via Texinfo's @include directive)
1013 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1014 the test case to Dave Roberts)
1015 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1016 pointers. (reported by Sean Ross)
1017 * bug fix: PROFILE output is printed nicely even for large numerical
1018 values. (thanks to Zach Beane)
1019 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1020 greater than 32 handle EOF correctly.
1021 * bug fix: on X86 an immediate argument of the IMUL instruction is
1022 correctly printed in disassembly. (thanks to Lutz Euler)
1023 * bug fix: class slots in redefined classes preserve their old
1024 values. (thanks to Bruno Haible and Nikodemus Siivola)
1025 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1026 succeeds. (reported by Marco Baringer)
1027 * fixed some bugs revealed by Paul Dietz' test suite:
1028 ** READ-SEQUENCE now works on ECHO-STREAMs.
1029 ** RATIONALIZE works more according to its specification. (thanks
1032 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1033 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1034 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1035 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1036 should matter to you only if you rebind the printer control
1037 variables and then find you want different bindings in the
1038 debugger than in the ordinary execution of your program.)
1039 * The runtime build system has been tweaked to support building
1040 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1041 assembler and linker. (thanks to Nikodemus Siivola)
1042 * Unbound, undefined, undeclared variables now trigger full
1043 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1044 more often programmer error than deliberate exploitation of undefined
1046 * optimization: the hash algorithm for strings has changed to one
1047 that is less vulnerable to spurious collisions. (thanks to Juho
1049 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1050 needless bounds checking. (thanks to Juho Snellman)
1051 * optimization: implemented multiplication as a modular
1052 (UNSIGNED-BYTE 32) operation on the PPC backend.
1053 * fixed some bugs revealed by Paul Dietz' test suite:
1054 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1056 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1057 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1058 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1060 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1062 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1064 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1065 their output stream on EOF from read.
1066 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1067 have been read to end-of-file.
1068 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1070 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1071 description of determination of which consecutive characters
1073 ** Printing the "Space" character with escaping on now yields "#\\ ",
1074 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1075 ** Reading floating-point numbers with *READ-BASE* set to a number
1076 less than 10 works correctly.
1077 ** Reading floating-point numbers with *READ-BASE* set to a number
1078 more than 10 works correctly.
1079 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1080 the readtable currently in effect.
1082 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1083 * minor incompatible change: parsing of namestrings on a physical
1084 (Unix) host has changed; numbers after the final #\. in a
1085 namestring are no longer interpreted as a version field. This is
1086 intented to be largely invisible to the user, except that the
1087 meaning of the namestring "*.*.*" has changed: it now refers to a
1088 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1089 should usually be replaced by
1090 (make-pathname :name :wild :type :wild :version :wild)
1091 with the added benefit that this is more likely to be portable.
1092 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1093 option to OPEN now signals an error if the file being opened
1094 exists; this may have an impact on existing code.
1095 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1096 signals received, on the PowerPC platforms (both Linux and
1097 Darwin). (thanks to Pierre Mai for pointing out the location of
1099 * several fixes on OS X: The system now builds and runs cleanly on
1100 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1101 (thanks to Brian Mastenbrook)
1102 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1103 second-resolution: integer multiples of 1/3600 between -24 and 24.
1104 (thanks to Vincent Arkesteijn)
1105 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1106 type when called with 1 argument; PEEK-CHAR checked type of
1107 PEEK-TYPE only after having read first character from a
1108 stream. (reported by Peter Graves)
1109 * bug fix: the garbage collector now has much better locality
1110 behaviour, and in particular no longer treats all memory as being
1111 exhausted when many small objects point to each other in a deeply
1113 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1114 recognized as being TYPEP their class.
1115 * bug fix: the PUSHNEW documentation string has been corrected.
1116 (thanks to Vincent Arkesteijn)
1117 * bug fix: defaulting of the value for the last of an atypically
1118 large number of multiple values being bound was not being
1119 performed correctly on the Alpha or PPC platforms
1120 * optimization: implemented multiplication as a modular
1121 (UNSIGNED-BYTE 32) operation on the x86 backend.
1122 * optimization: SEARCH on simple-base-strings can now be open-coded.
1123 (see also contrib/compiler-extras.lisp for inspiration for
1124 teaching the compiler about the Boyer-Moore algorithm).
1125 * value, returned by MAX (and MIN) called with several EQUALP, but
1126 not EQL, arguments now does not depend on compiler settings.
1127 * fixed some bugs revealed by Paul Dietz' test suite:
1128 ** in stack analysis liveness information is propagated from
1129 non-local entry points.
1130 ** pathwise CAST removing failed when the CAST node did not start
1132 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1133 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1135 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1136 host is already defined.
1137 ** RENAME-FILE works on streams instead of signalling an internal
1139 ** PEEK-CHAR uses the current readtable when determining whether
1140 or not a character is whitespace.
1141 ** MERGE-PATHNAMES handles the case when the pathname does not
1142 specify a name while the default-pathname specifies a version.
1143 ** Pathnames now stand a better chance of respecting print/read
1145 ** Attempting to use standardized file system operators with a
1146 pathname with invalid :DIRECTORY components signals a
1148 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1150 ** DIRECTORY on logical pathnames is more correct.
1151 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1152 signal a TYPE-ERROR if their argument is not a stream.
1153 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1154 designator argument does not designate a stream.
1155 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1156 examining the synonym.
1157 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1159 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1160 element-type larger than ([UN]SIGNED-BYTE 32).
1162 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1163 * When built with the :SB-FUTEX feature, threaded builds now take
1164 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1165 for faster/more reliable mutex and condition variable support.
1166 * Incompatible change (but one you probably shouldn't have been using
1167 anyway): the interface and code for arbitrating between multiple
1168 threads in the same user session has been redesigned.
1169 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1170 error when called without an explicit environment argument.
1171 (thanks to Nikodemus Siivola)
1172 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1173 confused about their position. (thanks to Adam Warner and Gerd
1175 * bug fix: Pretty printing backquoted forms with unquotations in the
1176 argument list position of various code constructs such as LAMBDA
1177 now works correctly. (reported by Paul Dietz)
1178 * bug fix: Pretty printing unquotations no longer loses all
1179 stream position information.
1180 * optimization: performance of string output streams is now less
1181 poor for multiple small sequence writes.
1182 * optimization: performance of CSUBTYPEP in the presence of complex
1183 expressions involving CONS and NOT many times has been improved.
1184 (reported by Paul Dietz)
1185 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1187 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1188 streams for the REPL, for the debugger, and so on. (thanks to
1190 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1191 * fixed some bugs revealed by Paul Dietz' test suite:
1192 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1193 is ignored if it is too small, rather than propagating through
1194 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1195 ** extremely complex negations of CONS types were not being
1196 sufficiently canonicalized, leading to inconsistencies in
1198 ** VALUES tranformer lost derived type.
1200 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1201 * fixed a bootstrapping bug: the build process no longer assumes
1202 that the various BOOLE-related constants have the same value in
1203 host and target lisps. (noted by Paul Dietz' test suite on an
1204 SBCL binary built from CLISP)
1205 * The system can now be dynamically linked on the MIPS platform,
1206 which enables dynamic loading of foreign code from Lisp. (thanks
1207 to Ralf Baechle for discussions on the MIPS ABI)
1208 * The system now records debugging information for its own source
1209 files in a filesystem-position-independent manner, relative to
1210 the "SYS" logical host.
1211 * fixed a compiler bug: MV-LET convertion did not check references
1212 to the "max args" entry point. (reported by Brian Downing)
1213 * tweaked disassembly notes to be less confident about proclaiming
1214 some instruction as an LRA. (thanks to Brian Downing)
1215 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1216 now each have their own history, command character, and other
1217 characteristics. (thanks to David Lichteblau)
1218 * fixed some bugs revealed by Paul Dietz' test suite:
1219 ** compiler failure in compiling LOGAND expressions including a
1221 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1222 shift greater than 32.
1223 ** FLUSH-DEST did not mark blocks for type check regeneration.
1224 ** HANDLER-CASE failed to accept declarations in handler clauses
1225 in some circumstances.
1227 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1228 * New code in contrib/sb-introspect (still probably not entirely
1229 stable yet) provides some support for smart Lisp development
1230 environments like SLIME.
1231 * The conditions signalled for errors occurring when loading .fasl
1232 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1233 in a way which should help ASDF recover gracefully.
1234 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1235 stuff has been cleaned up. If you code contrib/ stuff, this might
1236 affect you, and you can look at contrib/README, contrib/STANDARDS,
1237 and/or the 0.8.4.27 diff to check.
1238 * In full calls the compiler now does not generate checks for declared
1239 argument types for all arguments.
1240 * various threading fixes
1241 ** and some experimental patches which didn't make it into
1242 the main tree for this release, but which are shipped in
1243 contrib/experimental-thread.patch as a possible fix for some
1244 failures (deadlock, spinning...) in GC-intensive multithreaded
1246 * fixed PPC build problem (source code incompatibility of different
1247 library versions): added offsetof() hackery which attempts to divine
1248 where glibc maintainers put uc_mcontext today
1249 * fixed bug 282: compiler does not trust type assertions while passing
1250 arguments to a full call.
1251 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1252 &OPTIONAL VALUES type parameter.
1253 * fix bug 214: algorithm for noting rejected templates is now more
1254 similar to that of template seletion. (also reported by rydis on
1256 * fixed bug 141b: printing backquoted information readably and prettily
1257 inserts a space where necessary.
1258 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1259 (CHARACTER 10) are now reported as errors, rather than propagated
1260 as unknown types. (reported by piso on #lisp)
1261 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1262 with values NIL and :ERROR. (thanks to Milan Zamazal)
1263 * fixed bug 191c: CLOS now does proper keyword argument checking as
1264 described in CLHS 7.6.5 and 7.6.5.1.
1265 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1266 longer raise an error; further, using a list as a for-as-arithmetic
1267 counter now raises a meaningful error.
1268 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1269 now understood by sequence creation functions such as MAKE-SEQUENCE
1271 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1272 a STRING-INPUT-STREAM.
1273 * compiler enhancement: SIGNUM is now better able to derive the type
1275 * type declarations inside WITH-SLOTS are checked. (reported by
1277 * fixed some bugs revealed by Paul Dietz' test suite:
1278 ** incorrect optimization of TRUNCATE for a positive first
1279 argument and negative second.
1280 ** compiler failure in let-convertion during flushing dead code.
1281 ** compiler failure while deriving type of TRUNCATE on an
1282 interval, containing 0.
1283 ** ASH of a negative bignum by a negative bignum count now returns
1285 ** intersection of CONS types now canonicalizes properly, fixing
1286 inconsistencies in SUBTYPEP.
1288 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1289 * incompatible change: The --disable-debugger command line
1290 option now clobbers the debugger at a more fundamental
1291 level, by redefining #'INVOKE-DEBUGGER instead of by
1292 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1293 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1294 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1295 enter the debugger REPL and then suffer recursive errors
1296 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1297 while under the new system, BREAK in --disable-debugger mode
1298 terminates the system just as an unhandled error would.
1299 * fixed compiler performance when processing loops with a step >1;
1300 * bug fix: DOCUMENTATION now retrieves generic function
1301 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1302 support has been systematized, and now supports the methods
1303 specified by ANSI, along with a default method and a method for
1304 slot documentation. (reported by Nathan Froyd)
1305 * bug fix: effective methods associated with a generic function are
1306 no longer cached over a change of that generic function's method
1307 combination. (reported by Andreas Fuchs)
1308 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1309 names a non-existent directory. (thanks to Andreas Fuchs)
1310 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1311 the PPC platform, lead to stack corruption; this has been fixed.
1312 (reported by Rainer Joswig)
1313 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1314 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1315 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1316 arithmetic in general, is now much more reliable on the PPC
1318 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1319 platform now returns the right answer.
1320 * optimization: restored some effective method precomputation in
1321 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1322 precomputation is now tunable.
1323 * optimization: compiler-internal data structure use has been
1324 reviewed, and changes have been made that should improve the
1325 performance of the compiler by about 20%.
1326 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1327 simple-base-strings and simple-bit-vectors is improved.
1328 * optimization: the optimization of 32-bit logical and arithmetic
1329 functions introduced in version 0.8.3 on the x86 has been
1330 implemented on the mips, ppc and sparc platforms; an
1331 implementation of the same facility, but for 64-bit arithmetic,
1332 has been added for the alpha.
1333 * microoptimization: the compiler is better able to make use of the
1334 x86 LEA instruction for multiplication by constants.
1335 * bug fix: in some situations compiler did not report usage of
1336 generic arithmetic in (SPEED 3) policy.
1337 * bug 145b fix: compiler used wrong type specifier while converting
1338 MEMBER-types to numeric.
1339 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1341 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1342 index argument now works properly on the Alpha platform.
1343 * bug fix: floating point exception treatment on the Alpha platform
1345 * bug fix: FILE-POSITION works much better on string input and
1346 output streams. (thanks to Nikodemus Siivola)
1347 * bug fix: many threading/garbage collection symptoms sorted.
1348 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1349 might be pseudo-atomic.
1350 * internal change: Stopping for GC is now done with signals not
1351 ptrace. GC is now done in whichever thread wanted it, instead of
1353 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1355 * bug fix: result form in DO is not contained in the implicit
1357 * incompatible change: ICR structure is changed; the value part of
1358 CONTINUATION is now called LVAR; corresponding functions are
1359 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1360 * added type deriver for ISQRT (thanks to Robert E. Brown).
1361 * bug fix: better support for loading from the command line when an
1362 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1364 * fixed some bugs revealed by Paul Dietz' test suite:
1365 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1366 ** ROUND and FROUND now give the right answer when given very
1367 small float arguments.
1368 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1370 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1371 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1372 ** LCM should return a non-negative integer.
1373 ** PARSE-INTEGER returned the index of a terminator instead of the
1374 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1375 ** PARSE-INTEGER returned an incorrect index being applied to a
1377 ** LCM with two arguments of 0 returns 0 rather than signalling
1379 ** unsigned addition of a 32-bit constant with the high bit set no
1380 longer causes an internal compiler error on the x86.
1381 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1382 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1383 with negative last argument.
1384 ** byte specifiers with zero size and position no longer cause
1385 an error during type derivation.
1386 ** bignum multiplication on the Alpha platform now returns the
1388 * porting: The system now builds on SuSE AMD64, although it still
1389 generates a 32-bit binary.
1390 * .fasl file incompatibility: The fasl file version number has
1391 been incremented (because of the changes to internal compiler
1392 data structures referred to above).
1394 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1395 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1396 more accurately, on the Darwin kernel running on PowerPC hardware.
1397 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1398 * Compiler code deletion notes now signal a condition of type
1399 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1400 an associated MUFFLE-WARNING restart.
1401 * The compiler now performs limited argument count validation of
1402 constant format strings in FORMAT, and where appropriate in ERROR,
1403 CERROR and WARN. (thanks to Gerd Moellmann)
1404 * New ASDF-INSTALL contrib can be used for automatic download and
1405 installation of third-party Lisp code from CCLAN or other sites
1407 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1408 which forces another thread to execute a function supplied by the
1410 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1411 now accept and act upon their :ELEMENT-TYPE keyword argument.
1412 (reported by Martin Atzmueller, Edi Weitz)
1413 * bug fix: FILE-POSITION now accepts position designators up to
1414 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1415 is the greater. (thanks to Patrik Nordebo)
1416 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1417 Kalle Olavi Niemitalo)
1418 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1419 printable. (reported by Eric Marsden)
1420 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1421 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1422 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1423 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1424 (thanks to Stig Sandoe)
1425 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1426 UNREAD-CHAR now correctly includes the unread character in the
1427 target sequence. (thanks to Gerd Moellmann)
1428 * bug fix in threaded builds: the system can now be suspended and
1429 resumed by shell job control with minimal disruption.
1430 * bug fixes in times and timezones >2038AD
1431 * better handling of "where is GNU make?" problem in build scripts
1432 (thanks to Nikodemus Siivola)
1433 * new optimization: inside a named function any reference to a
1434 function with the same name is considered to be a self-reference;
1435 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1436 optimization quality.
1437 * new optimization on x86: logical functions and + now have
1438 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1439 used when the result is truncated to 32 bits.
1440 * VALUES declaration is partially enabled.
1441 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1442 * fixed some bugs revealed by Paul Dietz' test suite:
1443 ** The system now obeys the constraint imposed by
1444 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1445 types form a lattice under type intersection.
1446 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1447 ** ASSOC now ignores NIL elements in an alist.
1448 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1449 and (1+ MOST-POSITIVE-FIXNUM) answers.
1450 ** The addition of a method with invalid qualifiers to a generic
1451 function does not cause an error to be signalled immediately;
1452 a warning is signalled, and the error is generated only on
1453 calling the generic function.
1454 * changed .fasl file version number, in part to add type codes for
1455 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1456 obscure ANSI requirements
1458 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1459 * fixed bug 148: failure to inline-expand a local function left
1460 garbage, confusing the compiler.
1461 * fixed bugs 3cd: structure slot readers perform type check if the
1462 slot can have an invalid value (i.e. it is either not initialized
1463 or can be written with a less specific slot writer).
1464 * bug fix: the compiler now traps array references to elements off
1465 the end of an array; previously, the bounds checking in some
1466 circumstances could go off-by-one.
1467 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1469 * type declarations for array element types now obey the description
1470 on the CLHS page "Declaration TYPE", as per discussions on
1471 sbcl-help around 2003-05-08. This means that a declaration
1472 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1473 declaration, all references to BAR will be asserted or assumed
1474 (with THE, so dependent on compiler policy) to involve objects of
1475 type FOO. Note that no such declaration is implied in
1476 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1477 * declared types of functions from the "Conditions"
1478 chapter. (reported by Paul Dietz)
1479 * bug fix: CERROR accepts a function as its first argument.
1480 * bug fix: NTH an NTHCDR accept a bignum as index
1481 arguments. (reported by Adam Warner)
1482 * optimization: character compare routines now optimize comparing
1483 against a constant character. (reported by Gilbert Baumann)
1484 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1485 argument now works properly on the MIPS platform.
1486 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1487 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1488 declared to have a functional type, does not cause an error
1490 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1491 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1492 * fixed bug: if last continuation of a deleted block has a
1493 destination, this destination should be deleted too. (reported by
1495 * fixed a bug in the bootstrap process: the host compiler's values
1496 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1497 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1498 test case from Patrik Nordebo)
1499 * improved the ability of the disassembler on the PPC platform to
1500 provide helpful disassembly notes.
1501 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1502 the class in more cases than previously.
1503 * bug fix: FILE-POSITION now understands :START and :END for
1504 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1505 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1506 8) data. (thanks to David Lichteblau)
1507 * bug fix: it is possible to add a method to a generic function
1508 without lambda list.
1509 * bug fix: reader failed to signal END-OF-FILE inside an
1510 object representation. (reported by Nikodemus Siivola)
1511 * fixed some bugs revealed by Paul Dietz' test suite:
1512 ** LAST and [N]BUTLAST should accept a bignum.
1513 ** condition slot accessors are methods.
1514 ** (VECTOR NIL) is a subtype of STRING.
1516 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1517 * minor incompatible change: some nonsensical specialized lambda
1518 lists (used in DEFMETHOD) which were previously ignored now signal
1520 * minor incompatible change: the system is now aware of the types of
1521 variables in the COMMON-LISP package, and will signal errors for
1522 most violations of these type constraints (where previously they
1523 were silently accepted).
1524 * minor incompatible change: COMPILE-FILE now uses the freedom
1525 afforded (ANSI 3.2.2.3) to use derived function types for
1526 functions defined in the same file. This also permits the system
1527 to warn on static type mismatches and function
1528 redefinition. (Currently it does not work with high DEBUG level.)
1529 * minor incompatible change: VALUES declaration is disabled.
1530 * When issuing notes, the compiler now signals a condition of type
1531 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1532 restart for use in user handlers. It is expected that the
1533 COMPILER-NOTE condition will eventually become a condition
1534 supertype to a hierarchy of note types, which will then be
1535 handleable in a similar fashion. However, at the moment, no such
1536 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1537 but it's an implementation detail, not a classification for the
1539 * Changes in type checking closed the following bugs:
1540 ** type checking of unused values (192b, 194d, 203);
1541 ** template selection based on unsafe type assertions (192c, 236);
1542 ** type checking in branches (194bc).
1543 * A short form of VALUES type specifier has ANSI meaning (it has
1544 increased the number of situations when SBCL cannot perform type
1546 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1547 keywords or constants is permissible.
1548 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1549 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1550 classes. (thanks to Antonio Martinez)
1551 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1552 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1553 argument to a pretty-print stream. Instead, it leaves any such
1554 operations to DESCRIBE-OBJECT methods.
1555 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1557 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1558 lambda lists. (thanks to David Lichteblau)
1559 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1560 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1561 (thanks to Gerd Moellmann)
1562 * fixed bug 246: increased compilation speed of long
1563 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1565 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1567 * DEFCONSTANT now throws a condition of type
1568 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1569 constant to a non-EQL value; CONTINUE and ABORT restarts
1570 respectively change and preserve the value.
1571 * fixed bug 63: The code walker, part of the implementation of CLOS,
1572 is now better at handling symbol macros.
1573 * bug fix: There is no longer an internal implementation type named
1574 CL:LENGTH. (reported by Raymond Toy)
1575 * bug fix: In macro-like defining macros/special operators the
1576 implicit block does not enclose the lambda list.
1577 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1578 enclosed in parentheses) are not suitable as type specifiers, and
1579 their use properly signals an error now.
1580 * bug fix: An argument count mismatch for a type specifier in code
1581 being compiled no longer causes an unhandled error at compile
1582 time, but signals a compile-time warning.
1583 * fixed simple vector readable printing
1584 * bug fix: DESCRIBE takes more care over whether the class
1585 precedence list slot of a class is bound before accessing it.
1586 (reported by Markus Krummenacker)
1587 * bug fix: FORMATTER can successfully compile pretty-printer format
1588 strings which use variants of the ~* directive inside.
1589 * bug fix: SEARCH now applies its TEST predicate to the elements of
1590 the arguments in the correct order. (thanks to Wolfhard Buss)
1591 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1592 in type inference. (thanks to Robert E. Brown)
1593 * bug fix: Reading in symbols with an explicit package name of ""
1594 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1595 * fixed some bugs revealed by Paul Dietz' test suite:
1596 ** NIL is now allowed as a structure slot name.
1597 ** Arbitrary numbers, not just REALs, are allowed in certain
1598 circumstances in LOOP for-as-arithmetic clauses.
1599 ** Multiple class redefinitions before slot access no longer
1600 causes a type error.
1601 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1602 association between the name and a class.
1603 ** Generic functions with non-standard method-combination and over
1604 five methods all of which return constants no longer return NIL
1605 after the first few invocations. (thanks to Gerd Moellmann)
1606 ** CALL-NEXT-METHOD with no arguments now passes the original
1607 values of the arguments, even in the presence of assignment.
1608 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1610 ** DISASSEMBLE works with closures and funcallable instances.
1611 ** ADD-METHOD now returns the generic function, not the new method.
1612 ** FIND-METHOD signals an error if the lengths of the specializers
1613 is incompatible with the generic function, even if the ERRORP
1615 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1616 which its argument is a member.
1617 ** DEFCLASS only redefines the class named by its class-name
1618 argument if that name is the proper name of the class;
1619 otherwise, it creates a new class.
1620 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1621 of the UNBOUND-SLOT condition to the name of the slot.
1622 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1623 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1624 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1625 SLOT-MAKUNBOUND in the specified fashion.
1627 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1628 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1629 cross-compilation host. As a consequence, we can now bootstrap our
1630 way up to SBCL starting with a bare gcc toolchain and human-readable
1631 source code (first the source to CLISP, then the source to SBCL).
1632 * A contributed module containing a partial implementation of the
1633 simple-streams interface has been included. (thanks to Rudi
1635 * A contributed module implementing the RFC1321 Message Digest
1636 Algorithm, known as MD5, has been included.
1637 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1638 no longer has any effect, as the code controlled by this feature
1639 has been deleted. (As far as we know, no-one has ever built using
1640 this feature, and its semantics were confused in any case).
1641 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1642 work on conditions (as required by the ANSI specification),
1643 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1644 expected behaviour on conditions. Users should note, however,
1645 that such behaviour is not required by the ANSI specification,
1646 and so use of this behaviour may render their code unportable.
1647 * fixed some bugs revealed by Paul Dietz' test suite:
1648 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1650 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1651 ** SXHASH on condition objects no longer returns NIL.
1652 ** :ALLOCATION :CLASS slots are better treated; their values are
1653 updated on class redefinition, and initforms inherited from
1654 superclasses are applied.
1655 ** REMOVE-METHOD returns its generic function argument even when
1656 no method was removed.
1657 ** SHARED-INITIALIZE now initializes the values of the requested
1658 slots, including those with :ALLOCATION :CLASS.
1659 ** ALLOCATE-INSTANCE now works on structure classes defined via
1660 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1662 ** SLOT-EXISTS-P now works on conditions, as well as structures
1664 ** MAKE-LOAD-FORM now has the required methods on
1665 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1666 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1667 keyword, and now implements the SLOT-NAMES argument.
1668 ** methods with &OPTIONAL arguments no longer allow too many
1669 arguments to be passed in the call without error.
1670 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1671 option is consistent with the required arguments of the generic
1672 function lambda list.
1673 * bug fix: REQUIRE accepts a string designator. (Thanks to
1675 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1676 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1677 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1678 * bug fix: The compiler now checks for duplicated variables in macro
1680 * bug fix: SETQ on globals returns the correct value.
1681 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1682 not a COMPILER-ERROR (followed by some other strange error on
1683 choosing the CONTINUE restart).
1684 * bug fix: make.sh and friends are now more consistent in the way that
1685 they look for GNU "make".
1687 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1688 * experimental native threads support (on x86 Linux >=2.4 only).
1689 This is not compiled in by default: you need to add :SB-THREAD to
1690 the target features. See the "Beyond ANSI" chapter of the manual
1692 * fix for longstanding nonANSIism: The old distinction between
1693 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1694 The return value from CL:FIND-CLASS is now a CLOS class, and
1695 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1696 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1697 * An interface to the MetaObject Protocol, as described in Kiczales,
1698 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1699 MIT Press, 1991, is available from the SB-MOP package.
1700 * incompatible change: the SB-PCL package should now be considered
1701 a private implementation detail, and no longer a semi-private MOP
1703 * minor incompatible change: due to rearrangement for threads, the
1704 control stack and binding stack are now allocated at arbitrary
1705 addresses instead of being hardcoded per-port. Users affected by
1706 this probably have to be doing advanced things with shared
1707 libraries, and will know who they are.
1708 * minor incompatible change: Previously, all --eval forms used were
1709 processed with READ before any of them were processed with EVAL.
1710 Now each --eval form is processed with both READ and EVAL before
1711 the next --eval form is processed. (Thus package operations like
1712 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1713 work as the user might reasonably expect.)
1714 * minor incompatible change: *STANDARD-INPUT* is now only an
1715 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1717 * minor incompatible change: Y-OR-N-P is now character-oriented,
1718 not line oriented. Also, YES-OR-NO-P now works without errors.
1719 (thanks to Antonio Martinez)
1720 * sb-aclrepl module improvements: an integrated inspector, added
1721 repl features, and a bug fix to :trace command.
1722 * Known functions, which cannot be open coded by the backend, are
1723 considered to be able to check types of their arguments. (fixing
1724 a bug report by Nathan J. Froyd)
1725 * fixed a bug in computing method discriminating functions: It is
1726 now possible to define methods specialized on classes which have
1727 forward-referenced superclasses. (thanks to Gerd Moellmann)
1728 * fixed evaluation order in optional entries (reported by Gilbert
1730 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1731 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1732 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1733 required-by-AMOP NAME argument, as well as CLASS and
1734 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1735 * fixed bug 20: DEFMETHOD can define methods using names that are
1736 not the proper names of classes to designate class specializers.
1737 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1738 Unix file descriptors, instead of blowing up. (thanks to Antonio
1740 * Garbage collection refactoring: user-visible change is that a
1741 call to the GC function during WITHOUT-GCING will not do garbage
1742 collection until the end of the WITHOUT-GCING. If you were doing
1743 this you were probably losing anyway.
1744 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1745 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1746 TYPEP the latter but not the former.
1747 * The compiler issues a full WARNING for calls to undefined functions
1748 with names from the CL package.
1749 * MAP-INTO for a vector destination is open coded. (reported by
1750 Brian Downing on c.l.l)
1751 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1752 documentation string.
1753 * fixed some bugs revealed by Paul Dietz' test suite:
1754 ** COPY-ALIST now signals an error if its argument is a dotted
1756 ** Condition slots are now accessed more correctly in the presence
1757 of multiple initargs for a given slot.
1758 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1759 exclude restarts of the same name associated with a different
1761 ** DEFCLASS of forward-referenced classes with another
1762 forward-referenced class in the superclasses list no longer
1764 ** Condition slots are now initialized once each, not multiple
1765 times. (thanks to Gerd Moellmann)
1766 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1768 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1769 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1770 arguments contain duplicated elements.
1771 ** RESTART-CASE understands local macros.
1772 ** RESTART-CASE associates exactly its own restarts with a condition.
1773 ** ENDP in safe mode checks its argument to be of type LIST.
1774 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1775 in question is unbound.
1776 ** Optimized MAKE-INSTANCE functions no longer cause internal
1777 assertion failures in the presence of duplicate initargs.
1778 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1779 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1780 propagated even in the ctor optimized implementation of
1782 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1784 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1785 * a better implementation of SXHASH on (simple) bit vectors,
1786 measured both in execution speed and in distribution of results
1787 over the positive fixnums, has been installed. Likewise, a better
1788 implementation of EQUAL for simple bit vectors is now available.
1789 * fixed CEILING optimization for a divisor of form 2^k.
1790 * fixed bug 240 (emitting extra style warnings "using the lexical
1791 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1792 by Antonio Martinez)
1793 * fixed SXHASH, giving different results for NIL depending on type
1794 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1795 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1796 variable. (found by Rolf Wester)
1797 * SBCL does not ignore type declarations for special
1798 variables. (reported by rif on c.l.l 2003-03-05)
1799 * some bug fixes in contrib/sb-aclrepl/
1800 * fixed some bugs revealed by Paul Dietz' test suite:
1801 ** a bug in the CONS type specifier, whereby the CAR and CDR
1802 types got intertwined, has been fixed;
1803 ** the type system is now able to reason about the interaction
1804 between INTEGER and RATIO types more completely;
1805 ** APPEND, [N]REVERSE and NRECONC check that those their
1806 arguments, which must be proper lists, are really so;
1807 ** An array specialized to be unable to hold elements has been
1808 implemented, as required -- yes, really -- by ANSI;
1809 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1810 on malformed property lists;
1812 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1813 * incompatible packaging change: in line with Unix convention,
1814 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1815 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1816 * REQUIRE and PROVIDE are now optionally capable of doing something
1817 useful. See the documentation string for REQUIRE.
1818 * infrastructure for a managed SBCL contrib system: contributed
1819 modules in this release include:
1820 ** the ASDF system definition facility;
1821 ** an interface to the BSD Sockets API;
1822 ** an ACL-like convenience interface to the repl;
1823 (thanks to Kevin Rosenberg)
1824 ** an implementation of ROTATE-BYTE, with efficient implementation
1826 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1827 gives a closer approximation to the right answer than previously.
1828 (thanks to Raymond Toy)
1829 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1830 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1831 optional environment arguments, as required by ANSI.
1832 * fixed bugs in other functions taking environment objects, allowing
1833 calls with an explicit NIL environment argument to be compiled
1835 * fixed bug 228: primary return values from
1836 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1837 COMPILE or FUNCTION.
1838 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1839 :TYPE will no longer signal errors on innocuous objects.
1840 * fixed bug 231b: SETQ is better at respecting type declarations in
1841 the lexical environment.
1842 * fixed a bug in DEFCLASS: classes named by symbols with no or
1843 unprintable packages can now be defined.
1844 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1845 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1846 * fixed compiler failure related to checking types of functions.
1847 (reported by Robert E. Brown)
1848 * the compiler is now much more consistent in its error-checking
1849 treatment of bounding index arguments to sequence functions: in
1850 (SAFETY 3) code, errors will be signalled in almost all cases if
1851 invalid sequence bounding indices are passed to functions defined
1852 by ANSI to operate on sequences.
1853 * fixed a bug in the build procedure: documentation of SBCL-specific
1854 packages is now preserved and available in the final Lisp image.
1855 * lifted FDEFINITION lookup out of loops in the implementation of
1856 many list operations. (thanks to Robert E. Brown)
1857 * fixed a bug in the reader: the #n# reader macro now works for
1858 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1859 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1860 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1861 E. Brown and Tony Martinez respectively)
1862 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1863 SLOT-BOUNDP: in particular, it is now able to optimize them much
1864 better, and is now not vulnerable to having packages renamed.
1865 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1866 always called when a slot is not present in an instance. (thanks
1868 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1869 longer signal an error on unreading a character following EOF on
1870 the previous constituent stream. (thanks to Tony Martinez)
1871 * fixed some bugs revealed by Paul Dietz' test suite:
1872 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1873 not just nonnegative fixnums;
1874 ** the logical bit-array operators such as BIT-AND now accept an
1875 explicit NIL for their "opt-arg" argument (to indicate a
1876 freshly-consed result bit-array);
1877 ** ELT now signals an error on an invalid sequence index in safe
1879 ** the type system is now cleverer about negations of numeric
1880 types, and consequently understands the BIGNUM and RATIO types
1882 ** the type system is now cleverer about the interaction between
1883 INTEGER and RATIO types: while bugs still remain, many more
1884 cases are accurately computed;
1885 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1886 if it is in the last clause;
1887 ** CONSTANTLY now correctly returns a side-effect-free function in
1889 ** DECLARE is no longer treated as a special-operator; in
1890 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1891 * incremented fasl file version number due to the change in the
1892 DEFSTRUCT-SLOT-DESCRIPTION structure.
1894 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1895 * minor incompatible change: code processed by the "interpreter" or
1896 EVAL now has a compilation optimization policy of (DEBUG 2)
1897 (changed from (DEBUG 1)) to improve debuggability of interactive
1898 development, and to allow the use of the debug RETURN command in
1900 * an experimental implementation of the RETURN command for the
1901 debugger has been included. (thanks to Frederik Kuivinen)
1902 * fixed bug 62: constraints were not propagated into a loop.
1903 * fixed bug in embedded calls of SORT (reported and investigated by
1905 * fixed some bugs revealed by Paul Dietz' test suite:
1906 ** printing and reading of arrays with some dimensions having
1907 length 0 (thanks to Gerd Moellmann);
1908 ** BOA constructor with &AUX argument without a default value does
1909 not cause a type error;
1910 ** CONSTANTP now returns true for all self-evaluating objects.
1912 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1913 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1914 accessors that are related by inheritance, as specified in the
1915 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1916 Valtteri Vuorikoski)
1917 * The compiler is now able to inline functions that were defined in
1918 a complex lexical environment (e.g. inside a MACROLET).
1919 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1921 * Support for the upcoming FreeBSD-5.0 release has been included.
1922 (thanks to Dag-Erling Smorgrav)
1923 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1924 effect when it is not in a toplevel context.
1925 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1926 stand a better chance of being correct. (thanks to Gerd
1928 * fixed bug in COERCE, which now signals an error on coercing a
1929 rational to a bounded real type which excludes the expected
1931 * The compiler is now able to derive types more accurately from the
1932 COERCE and COMPILE functions.
1933 * fixed bug 223: functional binding is considered to be constant
1934 only for symbols in the CL package.
1935 * fixed bug 231: SETQ did not check the type of a variable being set
1936 (reported by Robert E. Brown)
1937 * A new optimization for MAKE-INSTANCE has been included, fixing
1938 various bugs (including relating to :ALLOCATION :CLASS slots and
1939 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1941 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1942 ** As required by ANSI, LOOP now disallows anonymous collection
1943 clauses such as COLLECT I in conjunction with aggregate boolean
1944 clauses such as THEREIS (= I 1);
1945 ** LOOP now signals an error when any variable is reused in the
1946 same loop (including the potentially useful construct analogous
1947 to WITH A = 1 WITH A = (1+ A);
1948 ** IT is only a special loop symbol within the first clause of a
1949 conditional loop clause;
1950 ** LOOP with a typed iteration variable over a hashtable now
1951 signals a type error iff it should.
1952 * fixed some other bugs revealed by Paul Dietz' test suite:
1953 ** FILE-STREAM now names the class previously known as FD-STREAM;
1954 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1955 argument) no longer signals an error;
1956 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1957 of the slot symbol, rather than using the current package
1958 ((:CONC-NAME "") continues to intern the slot's name in the
1960 * incremented fasl file version number, because of the incompatible
1961 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1962 of the new implementation of DEFINE-COMPILER-MACRO.
1964 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1965 * Support for building SBCL for MIPS platforms running in
1966 little-endian mode has now been checked in, and basic
1967 functionality on said platforms verified.
1968 * minor incompatible change: PCL now records the pathname of a file
1969 in which methods and the like are defined, rather than its
1971 * minor incompatible change: TRUENAME now considers the truename of
1972 a file naming a directory to be the pathname with :DIRECTORY
1973 component indicating that directory.
1974 * minor incompatible change: a NAMED clause in the extended form of
1975 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1976 reason for the previous behaviour is unclear.
1977 * more systematization and improvement of CLOS and MOP conformance
1978 in PCL (thanks to Gerd Moellman and Pierre Mai):
1979 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1981 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1982 primary methods with no specializers;
1983 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1985 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1986 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1987 FORWARD-REFERENCED-CLASSes; error reporting on
1988 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1990 ** SXHASH on CLOS instances now uses a slot internal to the
1991 instance to return different numbers on distinct instances,
1992 while preserving the same return value through invocations of
1994 ** DEFMETHOD signals errors when methods with longer incongruent
1995 lambda lists are added to generic functions;
1996 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1997 CLASS, as specified in AMOP;
1998 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1999 on the order returned by the primary method for classes of
2000 class STANDARD-CLASS;
2001 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2002 * fixed some bugs shown by Paul Dietz' test suite:
2003 ** DOLIST puts its body in TAGBODY;
2004 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2006 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2007 value producing form;
2008 ** if more variables are given to PROGV than values, extra
2009 variables are bound and made to have no value;
2010 ** NSUBSTITUTE on list arguments gets the right answer with
2012 ** ELT signals an error of type TYPE-ERROR when the index argument
2013 is not a valid sequence index;
2014 ** LOOP signals (at macroexpansion time) an error of type
2015 PROGRAM-ERROR when duplicate variable names are found;
2016 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2017 ** FUNCALL of special-operators now cause an error of type
2019 ** PSETQ now works as required in the presence of side-effecting
2020 symbol-macro places;
2021 ** NCONC accepts any object as its last argument;
2022 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2024 ** loop-for-as-package does not require a package to be explicitely
2026 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2028 * fixed bug 166: compiler preserves "there is a way to go"
2029 invariant when deleting code.
2030 * fixed bug 172: macro lambda lists with required arguments after
2031 &REST arguments now cause an error to be signalled. (thanks to
2033 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2034 bug. (thanks to Matthew Danish)
2035 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2037 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2038 arguments are lexically rebound. (thanks to Gerd Moellmann and
2040 * fixed bug 194: error messages are now more informative when there
2041 is no primary method applicable in a call to a generic
2042 function. (thanks to Gerd Moellmann)
2043 * fixed bug in command line argument checking (thanks to Julian
2045 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2046 (thanks to Matthew Danish)
2047 * incremented fasl file version number, because of the
2048 SXHASH-related changes in the layout of CLOS data structures
2050 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2051 * minor incompatible change: The runtime (the Unix executable named
2052 "sbcl") is now much pickier about the .core files it will load.
2053 Essentially it now requires .core files to descend from the same
2054 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2055 as the runtime does. (The intent is to prevent the crashes which
2056 can occur, and which can even be reported as mysterious failures,
2057 when people patch the sources or change the build parameters
2058 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2059 sbcl and .core files.)
2060 * fixed bug: VALUES-LIST is no longer optimized away.
2061 * fixed bug 142: The FFI conversion of C string values to Lisp
2062 string values no longer conses excessively. (thanks to Nathan
2063 Froyd porting Raymond Toy's fix to CMU CL)
2064 * began to systematize and improve MOP conformance in PCL (thanks to
2065 Nathan Froyd, Gerd Moellman and Pierre Mai):
2066 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2068 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2069 ** FINALIZE-INHERITANCE is now called on class finalization;
2070 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2071 argument precedence order.
2072 * fixed bug 202: The compiler no longer fails on functions whose
2073 derived types contradict their declared type.
2074 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2075 so it can be non-toplevel.
2076 * The fasl file version number has changed (because of the new
2077 implementation of DEFMACRO).
2078 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2079 safe code, that any length requirement by their type-specifier
2080 argument is valid. The exceptions to this are described in bug
2082 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2083 functions now check the types of their inputs as required by ANSI.
2084 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2085 introduced symbol is DECLAREd to be SPECIAL.
2086 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2087 * fixed bug: :COUNT argument to sequence functions may be negative
2088 * fixed bug: body of DO-SYMBOLS may contain declarations
2089 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2090 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2091 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2092 symbol macro only once
2093 * fixed printing of call frame when argument list is unavailable
2094 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2095 * compiler no longer signals WARNING on unknown keyword
2098 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2099 * A beta-quality port to the mips architecture running Linux,
2100 based on the old CMUCL backend, has been made. It has been tested
2101 on a big-endian kernel, and works sufficiently well to be able to
2102 rebuild itself; it has not been tested in little-endian mode.
2103 * fixed an inconsistency between gencgc.c and purify.c which made
2104 dumping/loading .core files unreliable
2105 * fixed bug 120a: The compiler now deals correctly with IFs where
2106 the consequent is the same as the alternative, instead of
2107 misderiving the return type. (thanks to Alexey Dejneka)
2108 * fixed bug 113: Logical pathnames are now dumpable (the logical
2109 host is resolved at load-time, throwing an error if it is not
2111 * fixed bug 174: FORMAT's error message is slightly clearer when a
2112 non-printing character is used in a format directive.
2113 * fixed several bugs in compiler checking of type declarations, i.e.
2114 violations of the Python "declarations are assertions" principle
2115 (thanks to Alexey Dejneka)
2116 * fixed several bugs in PCL's error checking (thanks to Gerd
2118 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2120 * fixed bug in compilation of functions as first class values
2121 (thanks to Antonio Martinez-Shotton)
2122 * The compiler's handling TYPE-ERRORs which it can prove will
2123 inevitably happen at runtime has been cleaned up and corrected
2124 in several ways. (thanks to Alexey Dejneka)
2125 * improved argument type checking for various basic arithmetic
2126 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2127 lavished on them in the past that they can be compiled in many
2128 ways in different special cases
2129 * fixed bug 181: compiler checks validity of user supplied type
2131 * cleaned up code flushing in optimization: Function calls which
2132 should signal errors for safety purposes (e.g. which ANSI says
2133 should signal errors when their arguments are of incorrect type)
2134 are no longer optimized away.
2135 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2136 * incremented fasl file version number, because changes in the
2137 implementation of sequence functions like COERCE caused
2138 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2139 in old inline expansions) to become undefined. (Actually these
2140 changes were later undone, so we might very well be binary
2141 compatible with 0.7.7 after all, but leaving the version number
2142 incremented seemed like the simplest and most conservative
2145 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2146 * An alpha-quality port to the parisc architecture running Linux,
2147 based on the old CMUCL backend, has been made. This, even more so
2148 than the other backends, should be considered still a work in
2149 progress; known problems include that the Linux kernel in 64-bit
2150 mode does not propagate the correct sigcontext structure to
2151 userspace, and consequently SBCL on a parisc64 kernel will not
2153 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2154 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2155 suppressed.) Also now that the compiler is looking at declarations
2156 in the environment, it checks optimization declarations as well,
2157 and suppresses inlining when (> DEBUG SPEED).
2158 * More fixes have been made to treatment of floating point exception
2159 treatment and other Unix signals. In particular, floating point
2160 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2161 * The detection and handling of control stack exhaustion (infinite
2162 or very deeply nested recursion) has changed. Stack exhaustion
2163 detection is now done by write-protecting pages at the OS level
2164 and applies at all optimization settings; when found, a
2165 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2166 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2167 be caught using IGNORE-ERRORS.
2168 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2169 correctly when transforming recursive LABELS functions to LETs.
2170 (thanks to Alexey Dejneka)
2171 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2172 that are names of constants or global variables.
2173 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2174 alien routines with docstrings.
2175 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2176 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2178 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2179 (thanks to Eric Marsden)
2180 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2181 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2182 * Bug fix: The SPARC backend can now compile functions involving
2183 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2184 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2185 to a functions expecting a PATHNAME-DESIGNATOR.
2186 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2187 lambda lists. (thanks to Alexey Dejneka)
2188 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2189 functions so that EXPORTing the name of the class doesn't cause
2190 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2191 to Antonio Martinez for reporting this)
2192 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2193 constant-folded without failing an assertion. (thanks to Einar
2194 Floystad Dorum for reporting this)
2195 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2196 types are now dealt with more correctly by the compiler.
2197 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2198 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2199 behaviour analogous to RENAME-FILE. This puts its behaviour more
2200 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2202 * The fasl file version number has changed again. (because of the
2203 bug fix involving the names of PCL MAKE-INSTANCE functions)
2205 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2206 * bug fix: Floating point exceptions are treated much more
2207 consistently on the x86/Linux and PPC/Linux platforms.
2208 * Array initialization with :INITIAL-ELEMENT is now much faster for
2209 cases when the compiler cannot open code the array creation, but
2210 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2211 array accesses have also seen a speed increase.
2212 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2213 specified. (This was at the root of some bad interactions between
2214 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2216 * bug fix: Internal error arguments for undefined functions are now
2217 computed correctly on the PPC/Linux platform.
2218 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2219 Raymond Toy's patch for CMU CL)
2220 * Support for the Solaris 9 operating environment has been included
2221 (thanks to Daniel Merritt)
2222 * A very ugly but hopefully complete draft of the missing FFI chapter
2223 of the manual has been created by reformatting the corresponding
2224 CMU CL manual chapter into (currently very ugly and incoherent)
2225 DocBook and bringing it up to date for SBCL behavior. Thus, the
2226 manual is now essentially complete, at least by my extreme
2227 once-and-only-once standards, whereby it's acceptable to refer to
2228 the doc strings of SB-EXT functions as the primary documentation.
2229 * The fasl file version number has changed again, due to cleanup of
2230 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2231 is no longer a static symbol.)
2233 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2234 * SBCL now builds with OpenMCL (version 0.12) as the
2235 cross-compilation host; also, more progress has been made toward
2236 bootstrapping under CLISP.
2237 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2239 * bug 158 fixed: The compiler can now deal with integer loop
2240 increments different from 1; fixing this turned out also to fix
2242 * bug 169 fixed: no more bogus warnings about using lexical bindings
2243 despite the presence of perfectly good SPECIAL declarations (thanks
2244 to David Lichteblau)
2245 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2246 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2247 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2249 * bug 180 fixed: Method combination specifications no longer ignore
2250 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2251 * bug fix: Structure type predicate functions now check their argument
2252 count as they should.
2253 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2254 correctly. (thanks to Pierre Mai)
2255 * minor incompatible change: The --noprogrammer option is deprecated
2256 in favor of the new --disable-debugger option, which is very similar.
2257 (The major difference is that it takes effect at a slightly different
2258 time at startup, causing handling of errors in --sysinit and
2259 --userinit files will be affected differently.) The
2260 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2261 been added to allow this functionality to be controlled from ordinary
2262 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2263 maintainers, who might want to run non-interactive scripts to
2264 build SBCL cores which will later be used interactively.)
2265 * minor incompatible change: The LOAD function no longer, when given
2266 a wild pathname to load, loads all files matching that pathname.
2267 Instead, an error of type FILE-ERROR is signalled.
2269 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2270 * bug 147 fixed: The compiler preserves its block link/count
2271 invariants more correctly now so that it doesn't crash. (thanks
2273 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2275 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2276 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2277 no-value-for-this-character as NIL (as specified by ANSI).
2278 * HOST-NAMESTRING on physical pathnames now returns a string that is
2279 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2280 (thanks to Christophe Rhodes)
2281 * The Alpha port handles icache flushing more correctly. (thanks to
2283 * More progress has been made toward bootstrapping under CLISP. (thanks
2284 to Christophe Rhodes)
2285 * The fasl file format has changed again, because dynamic loading
2286 on OpenBSD (which has non-ELF object files) motivated some cleanups
2287 in the way that foreign symbols are transformed and passed around.
2288 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2289 is no longer treated as whitespace by the reader, but instead as
2290 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2291 |A
\7fB|, instead of A as it used to.
2293 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2294 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2295 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2296 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2297 this as of 0.7.1.45, but was left out of the previous news section
2298 (thanks to Dan Barlow)
2299 * SBCL now runs on the Solaris operating system on SPARC architectures
2300 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2301 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2302 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2303 * SPARC backend cleanups, allowing builds of cores optimized for V8
2304 and V9 SPARCS, and also emission of code targeted to a particular
2305 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2307 * SBCL is closer to bootstrapping under CLISP, thanks to various
2308 fixes by Christophe Rhodes.
2309 * The fasl file format has changed again, to allow the compiler's
2310 INFO database to support symbol macros.
2311 * The user manual (in doc/) is formatted into HTML more nicely.
2312 (thanks to coreythomas)
2313 * The system is smarter about SUBTYPEP relationships, especially
2314 those involving NOT types (including types such as ATOM which are
2315 represented internally using NOT types). Thus SUBTYPEP is less
2316 likely to return (VALUES NIL NIL) in general, and in particular
2317 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2319 * The fasl file format has changed again, because the internal
2320 representation of types now includes a new slot to support the new
2321 SUBTYPEP-of-NOT-types logic.
2322 * (not a change in the main branch of SBCL, but a related prototype
2323 which can hopefully be merged into the main branch of SBCL in the
2324 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2325 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2326 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2327 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2328 dispatch macro character combinations. (thanks to Alexey Dejneka)
2329 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2330 that superficially look like logical namestrings correctly.
2331 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2333 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2334 default supplied-p) for &optional and &key arguments. (thanks to
2337 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2338 * incompatible change: The compiler is now less aggressive about
2339 tail call optimization, doing it only when (> SPACE DEBUG) or
2340 (> SPEED DEBUG). (This is an incompatible change because there are
2341 programs which relied on the old CMU-CL-style behavior to optimize
2342 away their unbounded recursion which will now die of stack overflow.)
2343 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2344 for non-GENCGC systems has been increased to 20M (since that
2345 seems much closer to the likely performance optimum for modern
2346 systems than the old 4M value was)
2347 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2348 and *DEBUG-PRINT-LEVEL*
2349 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2350 of CMU CL's support for SPARC, and various endianness and other
2351 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2352 * new syntactic sugar for the Unix command line: --load foo.bar is now
2353 an alternate notation for --eval '(load "foo.bar")'.
2355 ** The system now detects stack overflow and handles it gracefully,
2356 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2357 optimization settings. (This is a good thing in general, and
2358 its introduction in this version should be particularly timely
2359 for anyone whose code fails because of suppression of tail
2361 ** The system now hunts for the C variable "environ" in a more
2362 devious way, to avoid segfaults when the C library version differs
2363 between compile time and run time. (thanks to Christophe Rhodes)
2364 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2365 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2366 ** The compiler no longer issues bogus style warnings for undefined
2367 classes in the same source file as the DEFCLASSes which defined
2368 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2370 ** fixes in CONDITION class precedence list for undefined function
2371 errors (thanks to Alexei Dejneka)
2372 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2373 correctly. (thanks to Dan Barlow)
2374 ** portability fixes aiming at bootstrapping under CLISP (thanks
2375 to Dave McDonald and Christophe Rhodes)
2376 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2377 ** fixes in type translation and and type inference (thanks to
2379 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2380 ** various fixes in the new ports (thanks to Dan Barlow)
2381 * several changes related to debugging:
2382 ** suppression of tail recursion, as noted above
2383 ** stack overflow detection, as noted above
2384 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2385 is now the default. (For some time encapsulation has been more
2386 reliable than the breakpoint-based :ENCAPSULATE NIL
2387 implementation, at least on X86 systems; and I just noticed that
2388 encapsulation also seems closer to the spirit of the ANSI
2391 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2393 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2394 up properly again. (There was a packaging bug in 0.7.0 which
2395 left their definitions in SB-SYS::LOAD-FOREIGN and
2396 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2397 vital for most things which interface to C-level interfaces,
2398 like extensions working with sockets or databases or
2399 Perl-compatible regexes or whatever, and the need to fix
2400 this bug is the main reason that 0.7.1 was released so
2402 ** DEFGENERIC is now choosier about the methods it redefines, so that
2403 reLOADing a previously-LOADed file containing DEFGENERICs does
2404 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2405 cycle now works as it should. (thanks to Alexey Dejneka)
2406 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2407 fixed by Christophe Rhodes. (He actually submitted this patch
2408 months ago, and I delayed until after 0.7.0.)
2409 ** Bug 111 (internal compiler confusion about runtime checks on
2410 FUNCTION types) was fixed by Alexey Dejneka.
2411 * Some internal cleanups (getting rid of variables which aren't
2412 needed now that the byte interpreter is gone) caused the fasl
2413 file format number to change again.
2415 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2416 * major incompatible change: The default fasl file extension, i.e. the
2417 default extension for files produced by COMPILE-FILE, has changed
2418 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2420 ** There are many changes in the implementation of the compiler.
2421 SBCL is now essentially a compiler-only implementation of ANSI
2422 Common Lisp. EVAL still "interprets" a few special cases, but
2423 almost all the interesting cases are handled by creating
2424 a LAMBDA expression, calling COMPILE on it, then calling
2425 FUNCALL on the result.
2426 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2427 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2428 Since the code is newer, there might still be some new bugs
2429 (though not as many as before Martin Atzmueller's fixes:-). But
2430 the new code is substantially simpler and clearer, and hopefully
2431 any remaining bugs will be simpler, less fundamental, and more
2432 fixable then the bugs in the old code.
2433 ** The revised compiler is still a little unsteady on its feet.
2435 *** The debugging information it produces (particularly the names
2436 of FUNCTION objects) is sometimes much less useful than what
2437 the old compiler produced.
2438 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2439 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2440 within a MACROLET) has been temporarily weakened.
2441 ** There are new compiler optimizations for various functions:
2442 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2443 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2444 *** the math functions TRUNCATE, FLOOR, and CEILING
2445 *** the function-of-all-trades COERCE
2446 Mostly these should be transparent, but there's one
2447 potentially-annoying problem (bug 117): when the compiler
2448 inline-expands a function and does type analysis on the result,
2449 it can create control paths which have type mismatches, and
2450 when it can't prove that those control paths aren't taken,
2451 it will issue WARNINGs about the type mismatches. This is
2452 a particular problem in practice for the new sequence functions.
2453 It's not clear how this should be fixed, and for now, a
2454 workaround is given in the entry for 117 in the BUGS file.
2455 ** (Because of the interaction between the two previous items --
2456 occasional inlining problems and new inline expansions -- some
2457 of the new sequence function optimizations won't really kick in
2458 completely until debugging information, and then inlining, are
2459 straightened out in some future version.)
2460 * minor incompatible changes:
2461 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2462 DIRECTORY no longer implicitly promotes NIL slots of its
2463 pathname argument to :WILD. In particular, when you ask for the
2464 contents of a directory (which you used to be able to do without
2465 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2466 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2467 ** changes in behavior that ANSI explicitly defines to be
2468 implementation dependent:
2469 *** The new compiler-only implementation still conforms with ANSI,
2470 but acts a little different than before. Besides the obvious
2471 changes in performance tradeoffs (that the cost per form passed
2472 to EVAL has gone up, and the cost per form executed by EVAL
2473 has gone down), the behavior of the system changes a little
2474 because there are no longer any interpreted function objects.
2475 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2476 e.g. doing COMPILE on the output of interactive DEFUN is
2478 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2480 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2481 has changed from (:CL) to NIL.
2482 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2483 e.g. #\Newline and #\Tab, don't have names specified in the
2484 ANSI Common Lisp standard, is now based on their ASCII symbolic
2485 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2486 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2487 are no longer used for output.
2488 ** changes in internal implementation constants:
2489 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2490 4 million. (If your application spends a lot of time GCing and
2491 you have a lot of RAM, you might want to experiment with
2492 increasing it even more.)
2493 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2494 However, almost all old code should still continue to work without
2495 immediate update, as SB-C-CALL is now a (deprecated) nickname
2497 ** Old operator names in the style DEF-FOO are now deprecated in
2498 favor of new corresponding names DEFINE-FOO, for consistency with
2499 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2500 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2501 internal symbols, but a few supported extensions like
2502 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2503 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2504 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2505 etc. as you get deeper into recursive calls to the debugger
2506 command loop, instead of the old "5]", "5]]", "5]]]"
2507 sequence. (I was motivated to do this when squabbles between
2508 ILISP and SBCL left me very deeply nested in the debugger. In the
2509 short term, this change will probably provoke more ILISP/SBCL
2510 squabbles, but hopefully it will be an improvement in the long run.)
2511 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2512 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2513 function, since declaiming return types involving aliens is
2514 (1) annoyingly messy to do by hand and (2) vital to efficient
2515 compilation of code which calls such functions.
2516 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2517 longer reexported by the SB-EXT package. They're solely useful
2518 for alien code, so it seems more logical that you should get
2519 them from the SB-ALIEN package, not in SB-EXT.
2520 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2521 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2522 features. Instead, the code that they used to control is always
2523 built into the system.
2524 * many other bug fixes
2525 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2526 advantage of the new EVAL-WHEN stuff and to clean them up in
2527 general, and they are now more ANSI-compliant in a number of
2528 ways. Martin Atzmueller is responsible for a lot of this.
2529 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2531 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2532 *** correct ERROR type for various file operations
2533 *** some fixes for Lisp streams
2534 *** DEFMETHOD syntax checking
2535 *** changing old weird representation of debug information as
2536 strings (which, among their other deficiencies, don't transform
2537 correctly when you rename packages, and don't change their
2538 print representation when you change things like *PACKAGE*
2539 and *PRINT-LENGTH*) to symbols and lists of symbols
2540 He also made several improvements and fixed several bugs in DESCRIBE.
2541 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2543 *** misbehavior of WRITE-STRING/WRITE-LINE
2544 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2545 and several other LOOP problems as well
2546 *** DIRECTORY when similar filenames are present
2547 *** DEFGENERIC with :METHOD options
2548 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2549 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2550 *** argument ordering in FIND with :TEST option
2551 *** mishandled package designator argument in APROPOS-LIST
2552 *** various problems in the backquote readmacro
2553 *** a bug in APROPOS
2554 *** probably some others that I'm not describing very well here,
2555 since the CVS log documents them by reference to sbcl-devel
2556 messages, and the SourceForge archives aren't working well.:-(
2557 ** Dan Barlow improved the Alpha port (and is making progress on the
2558 PPC port, for those of you who think different).
2559 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2560 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2561 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2562 continues to work on the SPARC port (for those of you in a position
2563 to look down upon our little PC-compatible boxes from a great height).
2564 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2565 and uses that copy, rather than the current dynamic value, when
2566 it's trying to decide whether to truncate output. Thus e.g.
2567 (let ((*print-lines* 50))
2568 (pprint-logical-block (stream nil)
2570 (let ((*print-lines* 8))
2571 (print (aref possiblybigthings i) stream)))))
2572 should now truncate the logical block only at 50 lines, instead of
2573 often truncating it at 8 lines, as it did before.
2574 * The doc/cmucl/ directory, containing old CMU CL documentation
2575 from the time of the fork, is no longer part of the base system.
2576 SourceForge has shut down its anonymous FTP service, and with it
2577 my original plan for distributing the old CMU CL documentation
2578 there. For now, if you need these files you can download an old
2579 SBCL source release and extract them from it.
2580 * The fasl file version number changed again, for dozens of reasons,
2581 some of which are apparent above.
2583 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2584 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2585 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2586 patch, so that DISASSEMBLE output is much nicer.
2587 * The code in the SB-PROFILE package now seems reasonably stable.
2588 I still haven't decided what the final interface should look like
2589 (I'd like PROFILE to interact cleanly with TRACE, since both
2590 facilities use function encapsulation) but if you have a need
2591 for profiling now, you can probably use it successfully with
2592 the current CMU-CL-style interface.
2593 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2594 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2595 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2596 link now returns the dangling link itself, and for similar
2597 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2598 link itself. (In these cases the old code signalled an error and
2599 looped endlessly, respectively.) Thus, DIRECTORY now works even
2600 in the presence of dangling and cyclic symbolic links.
2601 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2602 is now a supported extension again, since the consensus on
2603 sbcl-devel was that it can be useful for ordinary development
2604 work, not just for debugging SBCL itself.
2605 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2606 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2607 that currently-defined functions might be redefined later with
2608 different return types.
2609 * Hash tables can be printed readably, as inspired by CMU CL code
2610 of Eric Marsden and SBCL code of Martin Atzmueller.
2611 * better error handling in CLOS method combination, thanks to
2612 Martin Atzmueller porting Pierre Mai's CMU CL patches
2613 * more overflow fixes for >16Mbyte I/O buffers
2614 * A bug in READ has been fixed, so that now a single Ctrl-D
2615 character suffices to cause end-of-file on character streams.
2616 In particular, now you only need one Ctrl-D at the command
2617 line (not two) to exit SBCL.
2618 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2620 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2621 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2622 does the right thing.
2623 * The implementation of some type tests, especially for CONDITION
2624 types, is now tidier and maybe faster, due to CMU CL code
2625 originally by Douglas Crosher, ported by Martin Atzmueller.
2626 * Some math functions have been fixed, and there are new
2627 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2628 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2629 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2630 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2631 likely to be built into sbcl-0.7.0.)
2632 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2633 CL, ported to SBCL by Martin Atzmueller)
2634 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2635 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2636 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2637 * There's a new slam.sh hack to shorten the edit/compile/debug
2638 cycle for low-level changes to SBCL itself, and a new
2639 :SB-AFTER-XC-CORE target feature to control the generation of
2640 the after-xc.core file needed by slam.sh.
2641 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2642 COMPILE-FILE is no longer supported, so that now every function
2643 gets an entry point, so that block compilation looks a little
2644 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2645 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2646 returns the number of bytes consed since the system started,
2647 rather than the number consed since the first time the function
2648 was called. (The new definition parallels ANSI functions like
2649 CL:GET-INTERNAL-RUN-TIME.)
2650 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2651 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2652 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2653 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2654 * Fasl file version numbers are now independent of the target CPU,
2655 since historically most system changes which required version
2656 number changes have affected all CPUs equally. Similarly,
2657 the byte fasl file version is now equal to the ordinary
2660 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2661 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2662 is no longer recognized. I apologize for this, because it was
2663 listed in SB-EXT as a supported extension, but I found that
2664 its existing behavior was poorly specified, as well as incorrectly
2665 specified, and it looked like too much of a mess to straighten it
2666 out. I have enough on my hands trying to get ANSI stuff to work..
2667 * many patches ported from CMU CL by Martin Atzmueller, with
2668 half a dozen bug fixes in pretty-printing and the debugger, and
2669 half a dozen others elsewhere
2670 * fixed bug 13: Floating point infinities are now supported again.
2671 They might still be a little bit flaky, but thanks to bug reports
2672 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2673 as flaky as they were.
2674 * The --noprogrammer command line option is now supported. (Its
2675 behavior is slightly different in detail from what the old man
2676 page claimed it would do, but it's still appropriate under the
2677 same circumstances that the man page talks about.)
2678 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2679 are now supported, and enabled by default. Thus, the compiler can
2680 handle many floating point and complex operations much less
2681 inefficiently. (Thus e.g. you can implement a complex FFT
2683 * The compiler now detects type mismatches between DECLAIM FTYPE
2684 and DEFUN better, and implements CHECK-TYPE more correctly, and
2685 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2686 thanks to patches from Martin Atzmueller.
2687 * various fixes to make the cross-compiler more portable to
2688 ANSI-conforming-but-different cross-compilation hosts (notably
2689 Lispworks for Windows, following bug reports from Arthur Lemmens)
2690 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2691 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2692 fixed thanks to Pierre Mai's CMU CL patches.
2693 * fixes to keep the system from overflowing internal counters when
2694 it tries to use i/o buffers larger than 16M bytes
2695 * fixed bug 45a: Various internal functions required to support
2696 complex special functions have been merged from CMU CL sources.
2697 (When I was first setting up SBCL, I misunderstood a compile-time
2698 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2699 * improved support for type intersection and union, fixing bug 12
2700 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2701 more obscure bugs as well
2702 * some steps toward byte-compiling non-performance-critical
2703 parts of the system, courtesy of patches from Martin Atzmueller
2704 * Christophe Rhodes has made some debian packages of sbcl at
2705 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2706 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2707 stable, but are nonetheless usable. When he's ready, I'd be happy
2708 to add them to the SourceForge "File Releases" section. (And if
2709 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2710 * new fasl file format version number (because of changes in
2711 internal representation of (OR ..) types to accommodate the new
2712 support for (AND ..) types, among other things)
2714 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2715 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2717 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2719 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2720 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2721 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2722 * DESCRIBE now works on structure objects again.
2723 * Most function call argument type mismatches are now handled as
2724 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2725 know whether the function will be redefined before the call is
2726 executed. (The compiler could flag local calls with full WARNINGs,
2727 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2728 it doesn't keep track of enough information to know whether calls
2729 are local in this sense.)
2730 * Compiler output is now more verbose, with messages truncated
2731 later than before. (There should be some supported way for users
2732 to override the default verbosity, but I haven't decided how to
2733 provide it yet, so this behavior is still controlled by the internal
2734 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2735 src/compiler/ir1util.lisp.)
2736 * Fasl file format version numbers have increased again, because
2737 support for the Gray streams extension changes the layout of the
2738 system's STREAM objects.
2739 * The Gray subclassable streams extension now works, thanks to a
2740 patch from Martin Atzmueller.
2741 * The full LOAD-FOREIGN extension (not just the primitive
2742 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2743 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2744 but like most other programs, it defaults to copying the Unix
2745 environment from the original process instead of starting the
2746 new process in an empty environment.
2747 * Extensions which manipulate the Unix environment now support
2748 an :ENVIRONMENT keyword option which doesn't smash case or
2749 do other bad things. The CMU-CL-style :ENV option is retained
2750 for porting convenience.
2751 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2752 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2754 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2756 * A patch from Martin Atzmueller seems to have solved the SIGINT
2757 problem, and as far as we know, signal-handling now works cleanly.
2758 (If you find any new bugs, please report them!)
2759 * The system no longer defaults Lisp source file names to types
2760 ".l", ".cl", or ".lsp", but only to ".lisp".
2761 * The compiler no longer uses special default file extensions for
2762 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2763 seems to expect a single default extension for all compiled code,
2764 and there's no compelling reason to try to stretch the standard
2765 to allow two different extensions.) Instead, byte-compiled files
2766 default to the same extension as native-compiled files.
2767 * Fasl file format version numbers have increased again, because
2768 a rearrangement of internal implementation packages made some
2769 dumped symbols in old fasl files unreadable in new cores.
2770 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2771 many fewer weird special cases.
2772 * Bug #17 (differing COMPILE-FILE behavior between logical and
2773 physical pathnames) has been fixed, and some related misbehavior too,
2774 thanks to a patch from Martin Atzmueller.
2775 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2776 by Tim Moore, ported to SBCL by Martin Atzmueller.
2777 * Martin Atzmueller fixed several filesystem-related problems,
2778 including bug #36, in part by porting CMU CL patches, which were
2779 written in part by Paul Werkowski.
2780 * More compiler warnings in src/runtime/ are gone, thanks to
2781 more patches from Martin Atzmueller.
2782 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2785 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2787 * DESCRIBE now works on CONDITION objects.
2788 * The debugger now handles errors which arise when trying to print
2789 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2791 * The build system now uses an additional file, customize-target-features.lisp,
2792 to allow local modifications to the target *FEATURES* list. (The point of
2793 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2794 debugging features enabled, without having to worry about propagating it
2795 into everyone's system when I do a "cvs update".) When no
2796 customize-target-features.lisp file exists, the target *FEATURES* list
2797 should be constructed the same way as before.
2798 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2799 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2800 warn about behavior which is undefined under the ANSI spec. Note: This
2801 is specified by ANSI, but it's not very popular with programmers.
2802 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2803 macro in the SBCL sources for an example of a workaround which you
2804 might use to make portable ANSI-standard code which does what you want.
2805 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2806 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2807 when it's used as a non-toplevel form. (This is required in order
2808 to implement the DEFCONSTANT-EQX macro.)
2809 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2810 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2811 funny things, due to bugs in the implementation of EVAL-WHEN
2812 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2813 non-toplevel DEFCONSTANTs are uncommon.)
2814 * The core file version number and fasl file version number have been
2815 incremented, because the old noncompliant DEFCONSTANT behavior involved
2816 calling functions which no longer exist, and because I also took the
2817 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2818 * fixed bug 1 (error handling before read-eval-print loop starts), and
2819 redid debugger restarts and related debugger commands somewhat while
2821 ** The QUIT debugger command is gone, since it did something
2822 rather different than the SB-EXT:QUIT command, and since it never
2823 worked properly outside the main toplevel read/eval/print loop.
2824 Invoking the new TOPLEVEL restart provides the same functionality.
2825 ** The GO debugger command is also gone, since you can just invoke
2826 the CONTINUE restart directly instead.
2827 ** The TOP debugger command is also gone, since it's redundant with the
2828 FRAME 0 command, and since it interfered with abbreviations for the
2830 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2832 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2833 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2834 some porting work by Martin Atzmueller)
2835 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2836 with a boatload of patches which he ported to SBCL. Now that those
2838 ** The system tries to make sure that its low-priority messages
2839 are prefixed by semicolons, to help people who like to use
2840 syntax highlighting in their ILISP buffer. (This patch
2841 was originally due to Raymond Toy.)
2842 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2843 patches originally written by Raymond Toy.
2844 ** The compiler understands coercion between single-value and
2845 multiple-VALUES type expressions better, getting rid of some very
2846 weird behavior, thanks to patches originally by Robert MacLachlan
2847 and Douglas Crosher.
2848 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2849 lambda lists, thanks to a patch originally by Pierre Mai.
2850 ** The system no longer bogusly warns about "abbreviated type
2852 ** The compiler gets less confused by inlining and RETURN-FROM,
2853 thanks to some patches originally by Tim Moore.
2854 ** The system no longer hangs when dumping circular lists to fasl
2855 files, thanks to a patch originally from Douglas Crosher.
2856 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2857 undefined function error.
2858 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2859 in structure slot initforms) for now, documented workaround instead:-|
2860 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2861 * fixed bug 5: added stubs for various Gray stream functions called
2862 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2863 installed, at least appropriate type errors are generated
2864 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2865 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2866 consistently in DEFMETHOD forms.
2867 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2868 it doesn't seem to affect SBCL after all
2869 * The C runtime system now builds with better optimization and many
2870 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2872 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2874 * The system is now under CVS at SourceForge (instead of the
2875 CVS repository on my home machine).
2876 * The new signal handling code has been tweaked to treat register
2877 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2878 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2879 downstream, e.g. in debug-int.lisp, has implicit dependencies
2880 on the unsignedness of integer representation of machine words,
2881 and that caused the system to bomb out with infinite regress
2882 when trying to recover from type errors involving signed values,
2883 e.g. (BUTLAST '(1 2 3) -1).)
2884 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2885 (This was one of the bugs Peter Van Eynde reported back in July.)
2886 * The system now uses code inspired by Colin Walters' O(N)
2887 implementation of MAP (from the cmucl-imp@cons.org mailing
2888 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2889 inline the MAP operation, and there is more than one
2890 sequence argument to the MAP call (so that it can't just
2891 do ETYPECASE once and for all based on the type of the
2892 single sequence argument). (The old non-inline implementation
2893 of the general M-argument sequence-of-length-N case required
2894 O(M*N*N) time when any of the sequence arguments were LISTs.)
2895 * The QUIT :UNIX-CODE keyword argument has been renamed to
2896 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2897 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2899 * Raymond Wiker's patches to port dynamic loading from Linux to
2900 FreeBSD have been added.
2901 * The BUGS file is now more nearly up to date, thanks in large part
2902 to Martin Atzmueller's review of it.
2903 * The debugger now flushes standard output streams before it begins
2904 its output ("debugger invoked" and so forth).
2905 * The core version number and fasl file version number have both
2906 been incremented, because of incompatible changes in the layout
2908 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2909 used to not be called for a saved Lisp image.)
2910 * Martin Atzmueller's version of a patch to fix a compiler crash,
2911 as posted on sbcl-devel 13 September 2000, has been installed.
2912 * Instead of installing Martin Atzmueller's patch for the
2913 compiler transform for SUBSEQ, I deleted the compiler transform,
2914 and transforms for some similar consing operations.
2915 * A bug in signal handling which kept TRACE from working on OpenBSD
2917 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2918 away by constant folding
2919 * The system now defines its address space constants in one place
2920 (in the Lisp sources), and propagates them automatically elsewhere
2921 (through GENESIS and the sbcl.h file). Therefore, patching the
2922 address map is less unnecessarily tedious and error-prone. The
2923 Lisp names of address space constants have also been systematized.
2924 * CVS tags like dollar-Header-dollar have been removed from
2925 the sources, because they have never saved me trouble and
2926 they've been source of trouble working with patches and other
2927 diff-related operations.
2928 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2929 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2931 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2933 * The system has been ported to OpenBSD.
2934 * The system now compiles with a simple "sh make.sh" on the systems
2935 that it's supported on. I.e., now you no longer need to tweak
2936 text in the target-features.lisp-expr and symlinks in src/runtime/
2937 by hand, the make.sh takes care of it for you.
2938 * The system is no longer so grossly inefficient when compiling code
2939 involving vectors implemented as general (not simple) vectors (VECTOR T),
2940 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2941 which dares to use the various sequence functions on non-simple
2942 vectors, takes less of a performance hit.
2943 * There is now a primitive type predicate VECTOR-T-P
2944 to test for the (VECTOR T) type, so that e.g.
2945 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2946 can now be compiled with some semblance of efficiency. (The old code
2947 turned the type declaration into a full call to %TYPEP at runtime!)
2948 * AREF on (VECTOR T) is still not fast, since it's still compiled
2949 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2950 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2951 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2952 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2953 without making full calls to SUBTYPEP at runtime.
2954 (Some analogous efficiency issues for non-simple vectors specialized to
2955 element types other than T, or for non-simple multidimensional arrays,
2956 have not been addressed. They could almost certainly be handled the
2957 same way if anyone is motivated to do so.)
2958 * The changes in array handling break binary compatibility, so
2959 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2960 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2961 instead of (VALUES T T).
2962 * By following the instructions that Dan Barlow posted to sbcl-devel
2963 on 2 July 2000, I was able to enable primitive dynamic object
2964 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2965 functionality is not implemented (since it calls ld to resolve
2966 library references automatically, requiring RUN-PROGRAM for its
2967 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2968 try to resolve library references) is now supported.
2969 * The system now flushes the standard output streams when it terminates,
2970 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2971 them at several other probably-convenient times, e.g. in each pass of
2972 the toplevel read-eval-print loop, and after evaluating a form given
2973 as an "--eval" command-line option. (These changes were motivated by a
2974 discussion of stream flushing issues on cmucl-imp in August 2000.)
2975 * The source transform for TYPEP of array types no longer assumes
2976 that an array whose element type is a not-yet-defined type
2977 is implemented as an array of T, but instead punts, so that the
2978 type will be interpreted at runtime.
2979 * There is now some support for cross-compiling in make.sh: each of
2980 the phases of make.sh has its own script. (This should be transparent
2981 to people doing ordinary, non-cross-compile builds.)
2982 * Since my laptop doesn't have hundreds of megabytes of memory like
2983 my desktop machine, I became more motivated to do some items on
2984 my to-do list in order to reduce the size of the system a little:
2985 ** Arrange for various needed-only-at-cold-init things to be
2986 uninterned after cold init. To support this, those things have
2987 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2988 symbols with such names are now uninterned after cold init).
2989 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2990 fasl files for cold load.
2991 ** Remove the old compiler structure pooling code (which used to
2992 be conditional on the target feature :SB-ALLOC) completely.
2993 ** Redo the representation of some data in cold init to be more compact.
2994 (I also looked into supporting byte compiled code at bootstrap time,
2995 which would probably reduce the size of the system a lot, but that
2996 looked too complicated, so I punted for now.)
2997 * The maximum signal nesting depth in the src/runtime/ support code has
2998 been reduced from 4096 to 256. (I don't know any reason for the very
2999 large old value. If the new smaller value turns out to break something,
3000 I'll probably just bump it back up.)
3001 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3003 * Many, many bugs reported by Peter Van Eynde have been added to
3004 the BUGS list; some have even been fixed.
3005 * While enabling dynamic object file loading, I tried to make the
3006 code easier to understand, renaming various functions and variables
3007 with less ambiguous names, and changing some function calling
3008 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3009 * While trying to figure out how to do the OpenBSD port, I tried to
3010 clean up some of the code in src/runtime/. In particular, I dropped
3011 support for non-POSIX signal handling, added various comments,
3012 tweaked the code to reduce the number of compilation warnings, and
3013 renamed some files to increase consistency.
3014 * To support the new automatic configuration functionality in make.sh,
3015 the source file target-features.lisp-expr has been replaced with the
3016 source file base-target-features.lisp-expr and the machine-generated
3017 file local-target-features.lisp-expr.
3018 * fixed a stupid quoting error in make.sh so that using CMU CL
3019 "lisp -batch" as cross-compilation host works again
3021 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3023 * DESCRIBE no longer tries to call itself recursively to describe
3024 bound/fbound values, so that it no longer fails on symbols which are
3025 bound to themselves (like keywords, T, and NIL).
3026 * DESCRIBE now works on generic functions.
3027 * The printer now prints less-screwed-up representations of closures
3028 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3029 * A private symbol is used instead of the :EMPTY keyword previously
3030 used to mark empty slots in hash tables. Thus
3031 (DEFVAR *HT* (MAKE-HASH-TABLE))
3032 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3033 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3034 now does what ANSI says that it should. (You can still get
3035 similar noncompliant behavior if bang on the hash table
3036 implementation with all the symbols you get back from
3037 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3038 This breaks binary compatibility, since tests for equality to
3039 :EMPTY are wired into things like the macroexpansion of
3040 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3042 * There's now a minimal placeholder implementation for CL:STEP,
3043 as required by ANSI.
3044 * An obscure bug in the interaction of the normal compiler, the byte
3045 compiler, inlining, and structure predicates has been patched
3046 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3047 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3049 * Missing ordinary arguments in a macro call are now detected even
3050 when the macro lambda list contains &KEY or &REST.
3051 * The debugger no longer complains about encountering the top of the
3052 stack when you type "FRAME 0" to explicitly instruct it to go to
3053 the top of the stack. And it now prints the frame you request even
3054 if it's the current frame (instead of saying "You are here.").
3055 * As specified by ANSI, the system now always prints keywords
3056 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3058 * The default initial SIZE of HASH-TABLEs is now smaller.
3059 * Type information from CLOS class dispatch is now propagated
3060 into DEFMETHOD bodies, so that e.g.
3061 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3063 is now basically equivalent to
3064 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3065 (DECLARE (TYPE SINGLE-FLOAT X))
3067 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3068 operation, without having to do run-time type dispatch.
3069 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3070 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3071 * Since I don't seem to be making big file reorganizations very often
3072 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3073 is overflowing my ability to conveniently back them up), I've finally
3074 checked the system into CVS. (The CVS repository is on my home system,
3075 not at SourceForge -- putting it on SourceForge might come later.)
3076 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3077 high-level GC-NOTIFY-FOO functions send their output. (There's
3078 still very little control of where low-level verbose GC functions
3079 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3080 less than it used to -- the GC-NOTIFY-FOO functions are now under
3081 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3082 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3083 in only one place in the source code, and propagates it automatically
3084 everywhere that it's needed. Thus e.g. when I bump the version from
3085 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3086 * The C source files now include boilerplate legalese and documentation
3087 at the head of each file (just as the Lisp source files already did).
3088 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3089 to his page will be replaced with a link to his new CLiki service.
3091 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3093 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3094 * The build process now looks for GNU make under the default name "gmake",
3095 instead of "make" as it used to. If GNU make is not available as "gmake"
3096 on your system, you can change this default behavior by setting the
3097 GNUMAKE environment variable.
3098 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3099 can build without error under CMU CL.
3101 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3103 * There is now a partial SBCL user manual (with some new text and some
3104 text cribbed from the CMU CL manual).
3105 * The beginnings of a profiler have been added (starting with the
3106 CMU CL profiler and simplifying and cleaning up). Eventually the
3107 main interface should be through the TRACE macro, but for now,
3108 it's still accessed through vaguely CMU-CL-style functions and macros
3109 exported from the package SB-PROFILE.
3110 * Some problems left over from porting CMU CL to the new
3111 cross-compilation bootstrap process have been cleaned up:
3112 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3113 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3114 part when getting rid of the compiler *BACKEND* stuff.)
3115 ** The value of *NULL-TYPE* was screwed up, because it was
3116 being initialized before the type system knew the final
3117 definition of the 'NULL type. This screwed up several key
3118 optimizations in the compiler, causing inefficiency in all sorts
3119 of places. (I found it because I wanted to understand why
3120 GET-INTERNAL-RUN-TIME was consing.)
3121 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3122 PROCLAIM FTYPE information. Unfortunately this broke binary
3123 compatibility again, since now the forms output by DEFGENERIC
3124 to refer to functions which didn't exist in 0.6.3.
3125 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3126 can use the new (as of 0.6.3) transform for SOME into MAP into
3128 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3129 values to (INTEGER 0 1000000), so that the time code will no longer
3130 occasionally get blown up by Linux returning 1000000 microseconds
3131 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3132 its output conform to the ANSI spec. (Alas, this makes its output
3133 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3134 * A full call to MAP NIL with a single sequence argument no longer conses.
3135 * fixes to problems pointed out by Martin Atzmueller:
3136 * The manual page no longer talks about multiprocessing as though
3137 it were currently supported.
3138 * The ILISP support patches have been removed from the distribution,
3139 because as of version 5.10.1, ILISP now supports SBCL without us
3140 having to maintain patches.
3141 * added a modified version of Raymond Toy's recent CMU CL patch for
3142 EQUALP comparison of HASH-TABLE
3144 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3146 * The system still can't cross-compile itself with
3147 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3148 entails), but at least it can compile more of itself that way
3149 than it used to be able to, and various buglets which were uncovered
3150 by trying to cross-compile itself that way have now been fixed.
3151 * This release breaks binary compatibility again. This time
3152 at least I've incremented the FASL file format version to 2, so that the
3153 problem can be detected reliably instead of just causing weird errors.
3154 * various new style warnings:
3155 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3156 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3157 ** using the lexical binding of a variable named in the *FOO* style
3158 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3159 as specified by ANSI.
3160 * *RANDOM-STATE* is no longer automatically initialized from
3161 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3162 default behavior of the system is to repeat its behavior every time
3163 it's run. If you'd like to change this behavior, you can always
3164 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3165 old convention there was no comparably easy way to get the system to
3166 repeat its behavior every time it was run.
3167 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3168 FTYPE declarations has been removed, in favor of their ANSI
3169 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3170 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3171 the types of their sequence arguments can be determined at compile time.
3172 This is done through a new open code expansion for MAP which eliminates
3173 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3174 argument types can be determined at compile time.
3175 * The optimizer now transforms COERCE into an identity operation when it
3176 can prove that the coerced object is already of the correct type. (This
3177 can be a win for machine generated code, including the output of other
3178 optimization transforms, such as the MAP transform above.)
3179 * Credit information has been moved from source file headers into CREDITS.
3180 * Source file headers have been made more standard.
3181 * The CASE macro now compiles without complaining even when it has
3184 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3186 * (Note that the way that the PCL macroexpansions were rewritten
3187 to accommodate the change in DEFGENERIC below breaks binary
3188 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3189 not run under sbcl-0.6.2. Once we get out of alpha releases,
3190 i.e. hit release 1.0.0, we'll probably try to maintain binary
3191 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3192 and sbcl-1.4.4. Until then, however, it might be fairly common
3193 for maintenance releases to break binary compatibility.)
3194 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3196 * The reporting of unhandled signals has been changed to print some
3197 explanatory text as well as the report form. (Previously only
3198 the report form was printed.)
3199 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3200 it defines, so that the compiler no longer issues undefined function
3201 warnings for compiled-but-not-yet-loaded generic functions.
3202 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3203 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3204 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3205 section of the ANSI spec are supported.
3206 * The "" nickname for the "KEYWORD" package has been removed.
3207 The reader still handles symbol tokens which begin with a package marker
3208 as keywords, but it doesn't expose its mechanism for doing so in the
3209 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3210 * The system now issues STYLE-WARNINGs for contradictory TYPE
3211 proclamations. (Warnings for contradictory FTYPE proclamations would
3212 be nice too, but those can't be done usefully unless the type system
3213 is made smarter about FUNCTION types.)
3214 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3215 systematized, so that "*target-*.lisp is supposed to exist only on the
3216 target and imply that there's a related file which exists on the
3217 host, and *host-*.lisp is supposed to exist only on the host and imply
3218 that there's a related file which exists on the target. This involves a
3219 lot of renaming. Hopefully the acute confusion caused by the renaming
3220 will be justified by the reduction in chronic confusion..
3221 ** runtime-type.lisp -> early-target-type.lisp
3222 ** target-type.lisp -> late-target-type.lisp
3223 ** early-host-format.lisp -> early-format.lisp
3224 ** late-host-format.lisp -> late-format.lisp
3225 ** host-error.lisp -> misc-error.lisp
3226 ** early-error.lisp -> early-target-error.lisp
3227 ** late-error.lisp -> late-target-error.lisp
3228 ** host-defboot.lisp -> early-defboot.lisp
3229 ** code/misc.lisp -> code/target-misc.lisp
3230 ** code/host-misc.lisp -> code/misc.lisp
3231 ** code/numbers.lisp -> code/target-numbers.lisp
3232 ** code/early-numbers.lisp -> numbers.lisp
3233 ** early-host-type.lisp -> early-type.lisp
3234 ** late-host-type.lisp -> late-type.lisp
3235 ** host-typep.lisp -> typep.lisp
3236 ** load.lisp -> target-load.lisp
3237 ** host-load.lisp -> load.lisp
3238 ** host-disassem.lisp -> disassem.lisp
3239 ** host-insts.lisp -> insts.lisp
3240 ** byte-comp.lisp -> target-byte-comp.lisp
3241 ** host-byte-comp.lisp -> byte-comp.lisp
3242 ** host-signal.lisp -> signal.lisp
3243 ** host-defstruct.lisp -> defstruct.lisp
3244 ** late-target-type.lisp -> deftypes-for-target.lisp
3245 Furthermore, several other previously target-only files foo.lisp (e.g.
3246 hash-table.lisp and random.lisp) have been split into a target-and-host
3247 foo.lisp file and a target-only target-foo.lisp file, with their key type
3248 definitions in the target-and-host part, so that the cross-compiler will
3249 know more about target types.
3250 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3251 gone away. In their place are various *BACKEND-FOO* variables
3252 corresponding to the slots of the old structure.
3253 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3254 into the target SBCL has been fixed.
3255 * The chill.lisp system for loading cold code into a running SBCL
3257 * Support for the CMU CL "scavenger hook" extension has been removed.
3258 (It was undocumented and unused in the CMU CL sources that SBCL was
3259 derived from, and stale in sbcl-0.6.1.)
3260 * Various errors in the cross-compiler type system were detected
3261 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3262 (enabling various consistency checks). Many of them were fixed,
3263 but some hard problems remain, so the compiler is back to
3264 running without *TYPE-SYSTEM-INITIALIZED* for now.
3265 * As part of the cross-compiler type system cleanup, I implemented
3266 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3267 * I have started adding UNCROSS calls throughout the type system
3268 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3269 on cross-compiler input files will be able to go away, and various
3271 * CONSTANTP now returns true for quoted forms (as explicitly required
3274 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3276 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3277 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3278 (MAKE-LIST -1). In the longer term, it ought to become true
3279 that declarations are assertions even at SAFETY 1. For now, it's not
3280 quite true even at SAFETY 3, but it's at least more nearly true..
3281 (Note that this change seems to increases the size of the system by
3282 O(5%) and to decrease the speed of the compiler by 20% or more.)
3283 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3284 for the problem of printing dozens of lines of distracting information
3285 about low-level system machinery as part of the top stack frame
3286 on entry to the debugger when an undefined function was called.
3287 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3288 is not reset to COMMON-LISP-USER.
3289 * Compilation of stuff related to dyncount.lisp has been made conditional
3290 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3291 smaller. The various dyncount-related symbols have been moved into
3292 a new "SB-DYNCOUNT" package.
3293 * tty-inspect.lisp has been renamed to inspect.lisp.
3294 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3295 feature has gone away. (When we eventually port to other flavors of
3296 libc and/or Unix, we'll try to make the differences between flavors
3297 invisible at the user level.)
3298 * Various other *FEATURES* tags, and/or their associated conditionals,
3299 have been removed if obsolescent, or given better documentation, or
3300 sometimes given more-mnemonic names.
3302 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3304 * tidied up "make.sh" script
3305 * tidied up system directory structure
3306 * better "clean.sh" behavior
3307 * added doc/FOR-CMUCL-DEVELOPERS
3308 * many many small tweaks to output format, e.g. removing possibly-confusing
3309 trailing #\. character in DESCRIBE-INSTANCE
3310 * (EQUALP #\A 'A) no longer signals an error.
3311 * new hashing code, including EQUALP hashing
3312 * tidied up Lisp initialization and toplevel
3313 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3314 * command line argument processing
3315 * added POSIX-GETENV function to deal with Unix-ish environment variables
3316 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3317 terminating SBCL on EOF
3318 * non-verbose GC by default
3319 * There is no more "sbcl" shell script; the sbcl file is now the C
3320 runtime executable (just like CMU CL).
3321 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3323 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3324 symbol and package information as Lisp native symbol and package objects
3325 instead of strings naming symbols and strings naming packages. This way,
3326 whenever packages are renamed (as in warm init), debug information is
3327 transformed along with everything else.
3328 * tweaked the optimization policy declarations which control the building
3329 of SBCL itself. Now, among other things, the system no longer saves
3330 source location debugging information. (This helps two problems at once
3331 by reducing SBCL size and by keeping SBCL from trying to look for its
3332 sources -- which may not exist -- when reporting errors.)
3333 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3334 debugging and testing purposes
3335 * cleaned up printing, making the printer call PRINT-OBJECT for
3336 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3337 methods, giving nearly-ANSI behavior
3338 * converted almost all special variables to use *FOO* naming convention
3339 * deleted PARSE-TIME functionality, since it can be done portably
3340 * moved some files out of cold init into warm init
3341 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3343 * regularized formatting of source files
3344 * added an install.sh script
3345 * fixed ridiculous memory usage of cross-compiler by making
3346 compiler/alloc.lisp not try to do pooling unless it can hook
3347 itself into the GC of the cross-compilation host. Now the system
3348 builds nicely on my old laptop.
3349 * added :SB-ALLOC in target-features.lisp-expr
3350 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3351 was not implemented)
3352 * re-did condition handling and note reporting in the compiler. Notes
3353 are no longer handled by signalling conditions. Style warnings
3354 and warnings are handled more correctly and reported in such a way
3355 that it's easy to find one or the other in your output (so that you
3356 can e.g. figure out which of many problems caused COMPILE-FILE to
3358 * changed the severity of several compiler warnings from full WARNING
3359 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3360 compiler note reporting so that it doesn't use the condition system
3361 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3362 COMPILE-FILE command)
3363 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3364 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3365 no longer supported, since their implementation was deeply intertwingled
3366 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3367 * removed bogus "support" for compiler macros named (SETF FOO), and
3368 removed the compiler macro for SETF INFO (but only after making a fool
3369 of myself on the cmucl-imp mailing list by posting a bogus patch for
3370 DEFINE-COMPILER-MACRO..)
3371 * Compiled files containing forms which have side effects on the Lisp
3372 reader (such as DEFPACKAGE forms) are now handled more correctly.
3373 (Compiler queuing of top level lambdas has been suppressed by setting
3374 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3375 * deleted various currently-unused source files, e.g. gengc.lisp. They
3376 may be added back at some point e.g. when porting to other architectures,
3377 but until they are it's distracting to distribute them and to try to
3379 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3380 there were so many of them they're just distractions, and UNCROSS is
3381 known to be able to handle the current sources
3382 * moved PROFILE functionality into TRACE, so that it will be clear
3383 how the wrapping and unwrapping of functions when you profile them
3384 interacts with the wrapping and unwrapping of functions when you
3385 trace them. (Actually, the functionality isn't there yet, but at least
3386 the interface specification is there. Hopefully, the functionality will
3387 arrive with some future maintenance release.)
3388 * removed host-oops.lisp
3389 * changed signature of QUIT function to allow UNIX-CODE argument
3390 * fixed READ-SEQUENCE bug
3391 * tweaked verbose GC output so that it looks more like the progress
3392 output that ANSI specifies for functions like LOAD
3393 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3394 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob