1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
3 * new port: the Solaris operating system on x86 processors is now
4 mostly supported, though some rough edges in the environment
5 remain. (thanks to Daisuke Homma)
6 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
7 space for its heap. (thanks to Thiemo Seufer)
8 * optimization: SBCL on MIPS platforms now supports dynamic-extent
9 closures. (thanks to Thiemo Seufer)
10 * minor incompatible change: eof selects abort in the debugger.
11 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
12 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
14 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
15 deprecated, and will go away in a future revision of SBCL.
16 * minor incompatible change: GC-ON and GC-OFF are no longer
17 implemented with a counter, it does not matter how many times gc
19 * bug fix: discriminating functions for generic function classes
20 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
21 make invalid assumptions about method precedence order. (reported
23 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
24 platforms. (thanks to Thiemo Seufer)
25 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
26 * bug fix: correct pathname printing with printer escaping is on.
27 (thanks to Kevin Reid)
28 * bug fix: complex VOP definitions in "user-space" no longer trigger
29 package locks. (reported by Zach Beane)
30 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
31 for funcallable-instance functions, and is no different from
33 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
34 characters in the current readtable. (reported by Nicholas Neuss)
35 * bug fix: SUBTYPEP on various CONS types returns more a more
36 accurate acknowledgment of its certainty.
37 * optimizations: REMOVE-DUPLICATES now runs in linear time on
38 lists in some cases. This partially fixes bug 384.
39 * flush all standard streams before prompting in the REPL and the
41 * bug fix: signal handling and triggering gc do not conflict
42 directly anymore, in particular a high frequency sb-sprof does
43 not prevent gc from running
44 * bug fix: DECODE-UNIVERAL-TIME now uses a more reasonable
45 approximation for timezone and DST information between the
46 universal time epoch and the smallest negative 32-bit time_t.
47 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
48 year 1899 so long as the encoded time is non-negative.
49 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
50 the correct answer for non-constant rotate amounts when performing
51 an inline 32-bit rotation.
53 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
54 there is only one thread in the session
55 ** bug fix: memory leak for streams created in one thread and
57 ** bug fix: lockup when compiled with gcc4
58 ** bug fix: race that allows the gc to be triggered when gc is
60 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
61 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
63 ** bug fix: finished threads are gc'ed properly
64 * fixed some bugs revealed by Paul Dietz' test suite:
65 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
66 has arguments with a different set of applicable methods from
67 the orignal arguments.
68 ** The type error thrown by MAP now has a correct expected-type
70 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
71 name a compiled function.
72 ** Three MISC tests where a large bignum was improperly coerced to
73 a float that couldn't represent that bignum during type
74 derivation were fixed.
75 ** SUBTYPEP can now handle types involving the negation of a
76 list-form FUNCTION type.
77 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
78 as (COMPLEX (AND RATIO (NOT FIXNUM))).
79 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
81 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
82 * New feature: Experimental support for bivalent streams: streams
83 opened with :element-type :default now allow character and binary
85 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
86 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
88 * bug fix: no more highly sporadic "couldn't check whether ~S is
89 readable" when reading a stream and an interrupt hits in the middle
90 of a select system call
91 * compiler better recognizes complex arrays (reported by Eduardo
93 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
94 for variables on SBCL built with linkage-tables. (reported by Luis
96 * various error reporting improvements.
97 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
98 (Thanks to Thiemo Seufer)
99 * bug fix: nested reader invokations work correctly
100 * bug fix: it is possible to have simultaneous references to foreign
101 code and foreign data with the same name.
103 ** added x86-64 support
104 ** incompatible change: the threading api now works with thread
105 objects instead of thread ids
106 ** bug fix: threads are protected from signals and interruption when
107 starting up or going down
108 ** bug fix: a race where an exiting thread could lose its stack to gc
109 ** bug fix: don't halt on infinite error in threads if possible
110 ** fixed numerous gc deadlocks introduced in the pthread merge
111 ** bug fix: fixed thread safety issues in read and print
112 ** bug fix: debugger doesn't hang on session lock if interrupted at
113 an inappropriate moment
114 ** bug fix: run-program is now thread safe(r)
115 ** bug fix: inner with-recursive-lock no longer releases the mutex
116 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
117 * fixed some bugs revealed by Paul Dietz' test suite:
118 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
119 EXPECTED-TYPE slots filled.
120 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
122 ** MAKE-CONDITION accepts classes as type-designators.
123 ** COMPILE may never return NIL.
124 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
125 range before calling Unix time functions
127 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
128 * numerous signal handling fixes to increase stability
129 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
130 * minor incompatible change: we now correctly canonize default
131 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
132 as per the MOP, rather than the historical (INITARG INITFUNCTION
133 INITFORM). (reported by Bruno Haible)
134 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
135 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
136 for more information.
137 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
138 pathname is a directory pathname.
139 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
140 * dynamic space size on PPC has been increased to 768Mb. (thanks to
142 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
143 :METACLASS argument in addition to a class name. (reported by
144 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
145 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
146 (reported by Bruno Haible; patch from Gerd Moellmann)
147 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
149 * bug fix: more cleanups to the floating point exception handling on
150 x86-64 (thanks to James Knight)
151 * bug fix: the compiler does not try to constant fold calls of
152 COERCE and BIT-* functions when they return freshly constructed
153 objects. (reported by Nikodemus Siivola and Paul Dietz)
154 * optimization: improved the MIPS versions of generic (in the
155 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
156 * optimization: direct conversion of (unsigned-byte 32) to floats on
157 the PowerPC platform.
158 * optimization: structure instances with raw slots now use less
159 memory, and probably show better memory locality. (thanks to
161 * optimization: DYNAMIC-EXTENT declarations for lists and closures
162 are treated as requests for stack allocation on the x86-64,
163 Alpha-32, PPC and SPARC platforms.
164 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
165 the wrong file descriptor; implementation of SOCKET-OPEN-P.
166 (thanks to Tony Martinez)
168 ** gcing a dead thread can no longer lead to lockups
169 ** threads block signals until they are set up properly
170 ** errno is no longer shared by threads
171 ** interrupt-thread restores the eflags register on x86
172 ** fixed some lockups due to gc/thread interaction
173 * fixed some bugs revealed by Paul Dietz' test suite:
174 ** invalid dotted lists no longer raise a read error when
176 ** signal an error if a symbol that names a declaration is used
177 as the name of a type, or vice versa
178 ** allow using the (declare (typespec var*)) abbreviation for
179 (declare (type typespec var*)) with all type specifiers
180 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
181 ** always return NIL from PPRINT-POP when OBJECT is NIL
182 ** don't signal errors when pretty-printing malformed LABELS,
183 FLET or MACROLET forms
184 ** declarations in a DOLIST body are also in scope for the
186 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
188 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
191 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
192 * fixed cross-compiler leakages that prevented building a 32-bit
193 target with a 64-bit host compiler.
194 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
195 opened with :IF-EXISTS :OVERWRITE.
196 * fixed bug 281: error for an invalid qualifier in a short-form method
197 combination method is not signalled until the faulty method is called.
198 * bug fix: iteration variable type inferrer failed to deal with open
199 intervals. (reported by Alan Shields)
200 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
201 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
202 or more runtime options were provided to the sbcl binary.
203 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
204 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
205 * optimization: there's now a fast-path for fixnum arguments in the
206 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
207 * optimization: the code generated on x86-64 is more compact thanks
208 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
209 * bug fix: floating-point exception handling now partly works on
210 x86-64. (Thanks to James Knight)
211 * improvement to the MIPS backend from Thiemo Seufer: C-style
212 64-bit long long arguments and return values to alien functions
214 * fixed some bugs revealed by Paul Dietz' test suite:
215 ** the type-error signalled from WARN has a filled-in DATUM slot.
216 ** the type-error required when a stream is not associated with
217 a file has the stream as its datum.
218 ** type-errors on single-floats on x86-64 no longer have
219 :INVALID-OBJECT as the datum
220 ** the type-errors signalled for invalid function names now have
221 a correct expected type
222 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
223 for structures defined with a :TYPE. Documentation strings for
224 typed structures are no longer immediately discarded
225 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
226 broadcast streams and synonym streams. FILE-LENGTH now also works
227 on broadcast streams.
229 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
230 * incompatible change: the --noprogrammer option, deprecated since
231 version 0.7.5, has been removed. Please use the equivalent
232 --disable-debugger option instead.
233 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
235 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
236 inoperational for a while now) has been completely removed.
237 * null lexical environments are now printed as #<NULL-LEXENV>,
238 significantly reducing the amount of clutter in typical backtraces.
239 * documentation on weak pointers, finalization, and after GC hooks
240 has been added to the manual.
241 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
242 as other specialized array types, is much faster. SUBSEQ and
243 COPY-SEQ on such arrays have also been sped up.
244 * optimization: EQL is now more efficient when at least other argument
245 is known to be of type (OR FIXNUM (NOT NUMBER)).
246 * fixed bug: the runtime is now less vulnerable to changes in the
247 size of the SBCL object on OS X, and virtual memory is reserved for
248 all spaces that need to be at a fixed address.
249 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
250 * fixed bug: finalizers and after GC hooks that cause consing are now
252 * fixed bug: compiler error messages and summaries are now printed to
253 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
254 * fixed inference of the upper bound of an iteration variable.
255 (reported by Rajat Datta).
256 * fixed bug 211e: calling local functions with duplicated constant
257 keyword argument no longer causes a bogus style warning about an
259 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
260 ftype declaration to be disregarded. (reported by Dave Roberts)
261 * fixed bug 373: caused by erronous compilation of references to alien
262 variables in the runtime on ppc/darwin.
263 * fixed bug 376: CONJUGATE type deriver.
264 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
265 (reported by Baughn on #lisp)
266 * a cross-compiler bug on non-x86 platforms has been identified and
267 fixed. (thanks to Bruno Haible)
268 * improvements to the MIPS runtime code for increased stability.
269 (thanks to Thiemo Seufer)
270 * increased the maximimum compact environment size to allow
271 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
272 * partial workaround for bug 135: don't name the function we're
273 calling for hairy cases of EVAL, so as not to accumulate one environment
274 entry per call to EVAL. (reported by Kevin Reid)
275 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
276 * optimization: type testing for non-vector arrays should be faster.
277 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
279 * bug fix: setting 31st element of a bit vector to zero did not work
281 * bug fix: redefining a class definition which failed due to a
282 previous accessor / function clash now works (but see BUGS entry
283 #380 for more problems in this area). (thanks to Zach Beane)
284 * the long form of DEFINE-METHOD-COMBINATION disables method group
285 checking when given a single method group with pattern *. (thanks
287 * on x86 compiler supports stack allocation of results of simple
288 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
289 * support for building and running on Mac OS X 10.4 "Tiger" has been added
290 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
291 * fixed some bugs related to Unicode integration:
292 ** the restarts for recovering from input and output encoding
293 errors only appear when there is in fact such an error to
295 * fixed some bugs revealed by Paul Dietz' test suite:
296 ** MISC.549 and similar: late transformation of unsafe type
297 assertions into derived types caused unexpected code
299 ** SCALE-FLOAT type deriver is less wrong.
300 ** type derivers for EXP, LOG and similar functions compute result
301 types for complex arguments better.
302 ** (MISC.563) CONJUGATE type deriver works for very restricted
304 ** out-of-line type testers for character strings are available.
305 ** EQUAL compiler transform understands specialness of objects
307 ** accessing double-floats stored on the stack now works on x86-64.
308 ** debugger internals could sometimes create invalid lispobjs,
309 resulting in GC crashes.
310 ** MISC.548: type check weakening can convert required type into
312 ** initialization forms for bindings are not in scope of free special
315 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
316 * incompatible change: thread support for non-NPTL systems has
317 been removed - locking is buggy and unreliable. A threaded
318 SBCL build will now warn at startup and refuse to create
319 new threads, unless futex support is detected in the kernel
320 * incompatible change: the top level REPL now has only an ABORT
321 restart associated with it, not TOPLEVEL and ABORT as it used to.
322 TOP and TOPLEVEL are now available as debugger commands for
323 returning to the top level.
324 * incompatible change: forms evaluated in the REPL now use the
325 global optimization policy.
326 * incompatible change: user- and system-initialization files are
327 no longer processed with LOAD, but by READ and EVAL; hence the
328 global optimization policy, startup package, readtable, etc,
330 * The .fasl file format number has been incremented because of
331 various incompatible changes.
332 * internal entry point details and argument counts no longer appear
333 in backtraces unless explicitly requested by setting
334 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
335 * built-in and standard functions no longer have names like "top
336 level local call to FOO".
337 * fixed bug 32: functions defined in non-null lexical environments
338 now have more legible printed representation
339 * fixed bug 33: functions defined in non-null lexical environemnts
340 are now more amenable to inspection by INSPECT.
341 * workaround for bug 354: XEPs no longer appear in backtraces unless
342 explicitly requested.
343 * fixed bug: receiving the signal which results from attempting to
344 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
345 SIGBUS on other *BSDs) on architectures where the C stack is also
346 the Lisp stack (x86 and x86-64) from foreign code no longer leads
347 to debugger confusion or wild instability. (reported by Cheuksan
349 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
350 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
351 (reported by Lutz Euler)
352 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
353 compiled code. (reported by Paul Dietz)
354 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
355 specialized parameter to a method to something that is not TYPEP
356 the specializer is now possible.
357 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
358 face of package deletion.
359 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
360 pathname. (reported by David Tolpin on c.l.l)
361 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
362 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
363 STRING-STREAM and FILE-STREAM, respectively.
364 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
365 than previously; in addition, its test suite is now run on build.
366 (thanks to Luke Gorrie)
367 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
368 * optimization: added a immediate representation for single-floats
370 * fixed some bugs revealed by Paul Dietz' test suite:
371 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
372 ** attempting to create a package with a colliding nickname causes
373 correctable errors to be signalled.
374 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
375 ** several x86-64 backend bugs related to sign-extension of immediate
378 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
379 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
380 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
382 * fixed loading of multiply forward-referenced layouts.
383 (thanks to Cheuksan Wang)
384 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
385 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
386 (see COMPILE-FILE documentation for details of :PRINT option)
387 * fixed bugs 19 and 317: fixed-format floating point printing is
388 more accurate. This also fixes a bug reported by Adam Warner
389 related to the ~@F format directive.
390 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
392 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
393 dispatch table if the from-char is a dispatch macro character.
394 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
395 by the wordsize no longer ignore the last word. (reported by Lutz
397 * fixed bug in type checking of dynamic-extent variables. (reported
399 * optimization: sequence traversal functions use their freedom to
400 coerce function designators to functions.
401 * optimization: code with many calls to CLOS methods specialized on
402 CLOS classes has had redundant type checks removed.
403 * optimization: added declarations to speed up operations that access
404 the internal character database (for example STRING-UPCASE)
405 * optimization: comparison operations between floats and sufficiently small
406 fixnums no longer create extra rationals
407 * fixed some bugs related to Unicode integration:
408 ** portions of multibyte characters at the end of buffers for
409 character-based file input are correctly transferred to the
410 start of the buffer at the next read.
411 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
412 passing it through to OPEN.
413 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
414 argument given to internal calls to OPEN.
415 ** the built-in comment readers (introduced by character sequences
416 ";" and "#|") are more forgiving to encoding errors; they will
417 STYLE-WARN and then attempt to resync the stream at a character
418 boundary. (thanks to Teemu Kalvas)
419 * fixed some bugs in the x86-64 port:
420 ** Negative short int return values from c-calls are sign-extended
422 ** The stack is aligned to 16-bytes for c-calls, as required by
423 the ABI. (thanks to Cheuksan Wang)
424 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
425 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
426 for enums. (thanks to Vincent Arkesteijn)
427 ** Multiple small optimizations and bugfixes for floating point
429 * fixed some bugs revealed by Paul Dietz' test suite:
430 ** Space, Tab, Linefeed, Return and Page have the invalid
431 secondary constituent character trait.
432 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
434 ** WITH-INPUT-FROM-STRING should only update the index place on
436 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
438 ** Bit-array operations (BIT-AND and similar) worked incorrectly
439 with one-dimensional arrays with fill pointers.
440 ** TYPE-OF failed on a complex with an integer realpart and a
442 ** compiler failure during type inference for the code of form
443 (IF (EQL X (THE ...)) ...) (MISC.535).
445 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
446 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
447 platform. The port supports 61-bit fixnums, large memory spaces
448 and reloading shared object files.
449 * enhancement: saving cores with foreign code loaded is now
450 supported on ppc/Darwin in addition to the previously supported
452 * enhancement: the statistical profiler now walks deeper into the
453 call stack for more meaningful call-graphs and accrued time
454 reports (x86/x86-64 only). It also now reports time spent in
456 * enhancement: it is now possible to trace most individual methods
457 of a generic function in addition to tracing the generic function
459 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
461 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
462 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
463 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
464 produces an error. (thanks to Vincent Arkesteijn)
465 * bug fix: NAMESTRING on pathnames with :WILD components in their
466 directories works correctly. (thanks to Artem V. Andreev)
467 * fixed bug 125: compiler preserves identity of closures. (reported
469 * bug fixed: functions with &REST arguments sometimes failed with
470 "Undefined function" when compiled with (DEBUG 3). (reported by
472 * bug fix: overflow during compiling of setting element of a bit
473 vector with constant index and value. (reported by Timmy Douglas)
474 * build fix: fixed the dependence on *LOAD-PATHNAME* and
475 *COMPILE-FILE-PATHNAME* being absolute pathnames.
476 * on x86 compiler partially supports stack allocation of dynamic-extent
478 * GO and RETURN-FROM do not check the extent of their exit points
479 when compiled with SAFETY 0.
480 * fixed some bugs related to Unicode integration:
481 ** encoding and decoding errors are now much more robustly
482 handled; it should now be possible to recover even from invalid
483 input or output to the terminal. (thanks to Teemu Kalvas)
484 ** provided a first cut at implementing STRING-TO-OCTETS and
485 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
486 ** altered the SB-MD5 contributed module slightly, changing the
487 interface just enough for it to be supportable for builds where
488 lisp characters are not eight bits.
489 * fixed some bugs revealed by Paul Dietz' test suite:
490 ** the FORMATTER-generated functions for ~V[ conditionals require
491 the correct number of arguments.
492 ** READ-FROM-STRING returns the mandated second value when applied
493 to displaced strings.
494 ** the #\Rubout and #\Backspace characters are treated as invalid
495 constituent characters by the tokenizer.
497 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
498 * new feature: reloading changed shared object files with
499 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
500 * new feature: references to foreign variables and functions
501 can now be compiled and loaded before the corresponding shared
502 object file is loaded, as long as the foreign definitions are
503 available at runtime.
504 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
505 supported. (thanks to Dan Debertin)
506 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
507 just keywords. (thanks to Vincent Arkesteijn)
508 * fixed bug #331: structure-class instances corresponding to
509 DEFSTRUCT forms are now created eagerly.
510 * fixed bug #345: backtraces from calls to undefined functions work
511 on x86 as well. Related bug #61 is now also partially fixed on x86
512 (backtraces from throws to unknown catch tags.)
513 * bug fix: lambda-list parsing is now stricter vrt. order and number
514 of lambda-list keywords.
515 * bug fix: as specified by AMOP, an error is signalled if a
516 class-option appears multiple times in a DEFCLASS form. (reported
518 * bug fix: RANDOM can be compiled when the compiler derives the type
519 of its numeric argument as a disjoint set of small integers.
520 (reported by Paul Dietz)
521 * bug fix: starting a core saved with shared objects loaded when
522 those objects are not available no longer causes threaded SBCL to
523 hang. (reported by Sean Ross)
524 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
526 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
527 argument, which must always be NIL. (reported by Kalle Niemitalo)
528 * bug fix: printing 1.0d+23 no longer results in an error.
529 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
530 * bug fix: structure slot setters preserve evaluation order. (thanks
532 * fixed some bugs related to Unicode integration:
533 ** RUN-PROGRAM can allow its child to take input from a Lisp
534 stream. (reported by Stefan Scholl)
535 ** ASDF-INSTALL successfully downloads .tar.gz files in all
536 locales. (reported by Ken Causey)
537 * fixed some bugs revealed by Paul Dietz' test suite:
538 ** INCF, DECF and REMF evaluate their place form as specified in
540 ** FORMATTER expands ~{ ~} iteration directives with V or #
541 parameters correctly.
542 ** FORMATTER deals with the ~@[ ~] conditional directive where the
543 consequent uses no arguments correctly.
544 ** the system has a partial understanding of the (COMPLEX RATIO)
546 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
549 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
550 * new feature: a build-time option (controlled by the :SB-UNICODE
551 keyword feature, enabled by default) for building the system with
552 support for the entire 21-bit character space defined by the
554 * new feature: the system now has rudimentary external-format
555 support; the primary user-visible change at this time is that
556 characters with the high bit set (such as lower-case-e-acute) will
557 print correctly to a terminal in a UTF-8 environment.
558 * minor incompatible change: BASE-CHAR no longer names a class;
559 however, CHARACTER continues to do so, as required by ANSI.
560 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
561 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
562 has been moved to the SB-EXT package (temporarily re-exported from
564 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
565 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
567 * fixed bug #351: better error-handlind and reporting for malformed
569 * fixed bug #350: bignum-printing is now more memory-efficient,
570 allowing printing of very large bignums, eg. (expt 2 10000000).
571 (reported by Bruno Haible)
572 * fixed bug #302: better primitive-type selection for intersection
574 * fixed bug #308: non-graphic characters now all have names, as
575 required. (reported by Bruno Haible)
576 * bug fix: redefining a class with different superclasses now correctly
577 removes it from the direct-subclasses of its previous superclasses.
578 (reported by David Morse)
579 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
580 new value now works. (reported by Bruno Haible)
581 * bug fix: correct canonicalization of multiple non-standard slot
582 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
583 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
584 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
585 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
586 now exists, an signals an error.
587 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
588 during macroexpansion for non-iteration variables as well. (reported
589 by Bruno Haible for CMUCL)
590 * bug fix: Cyclic structures and unprintable objects in compiler
591 messages no longer cause errors. (reported by Bruno Haible)
592 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
593 and READ-FROM-STRING all now return a primary value of NIL if
594 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
595 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
596 (reported by Bruno Haible for CMUCL)
597 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
598 do not designate a condition. (reported by Bruno Haible for
600 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
601 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
602 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
603 permitting resolution in favour of any of the conflicting symbols.
604 (reported by Bruno Haible for CMUCL)
605 * bug fix: EQUAL compiler optimizations is less aggressive on
606 strings which can potentially compare true despite having distinct
607 specialized array element types.
608 * bug fix: unit enumerations can be defined without dividing by
609 zero. (thanks to Vincent Arkesteijn)
610 * FORMAT compile-time argument count checking has been enhanced.
611 (report from Bruno Haible for CMUCL)
612 * a partial workaround for the bug 262: the compiler does not try to
613 inline-expand a local function doing RETURN-FROM from a deleted
614 BLOCK. (thanks to Peter Denno for the bug report and to David
615 Wragg for the simple test case)
616 * fixed some bugs revealed by Paul Dietz' test suite:
617 ** RENAME-PACKAGE allows all package designators as new package
619 ** constraint propagation and lambda variable substitution are
620 more cautious in dealing with partially deleted code.
621 ** compiler sometimes lost reoptimization passes.
622 ** CERROR, given a condition as condition designator, passes its
623 remaining arguments to the continue format control without
625 ** Case-altering FORMAT directives work correctly on non-ASCII
627 ** The REAL type specifier handles bounds outside the floating
628 point ranges without signalling FLOATING-POINT-OVERFLOW.
629 ** Functions with IR1-transformations can create intercomponent
630 references to global functions.
631 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
633 ** FORMAT ~R treats a nil value for its first parameter correctly.
635 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
636 * enhancement: saving cores with foreign code loaded is now
637 supported on x86/NetBSD and sparc/Linux in addition to the previously
639 * bug fix: on some platforms repeated installations caused multiple
640 copies of HTML documentation to be installed -- should not happen
641 any more. (reported by Stefan Scholl)
642 * bug fix: parsing self-recursive alien record types multiple times
643 no longer causes infinite recursion. (reported by Thomas F. Burdick,
644 original patch by Helmut Eller for CMUCL)
645 * bug fix: stack-exhaustion detection works now on NetBSD as well.
646 (thanks to Richard Kreuter)
647 * bug fix: defining classes whose accessors are methods on existing
648 generic functions in other (locked) packages no longer signals
649 bogus package lock violations. (reported by François-René Rideau)
650 * bug fix: special variables as DEFMETHOD parameters no longer have
651 associated bogus type declarations. (reported by David Wragg and
653 * bug fix: read-write consistency on streams of element-type
654 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
655 * bug fix: redefiniton of the only method of a generic function with
656 no DEFGENERIC no longer emits a full WARNING. In addition,
657 redefinition of generic functions with no DEFGENERIC to an
658 incompatible lambda list now signals an error. (thanks to Zach
660 * bug fix: DEFGENERIC now works even when there's a function of the
661 same name in an enclosing lexical environment. (thanks to Zach
663 * fixed compiler failure, caused by instrumenting code during
664 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
665 * optimization: added loop analysis and improved register allocation
666 * fixed some bugs revealed by Paul Dietz' test suite:
667 ** POSITION on displaced vectors with non-zero displacement
668 returns the right answer.
669 ** (SIMPLE-STRING) is a valid type specifier for sequence
671 ** *PRINT-LEVEL* handling for slotless structures is pedantically
673 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
674 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
676 ** The justification version of the FORMAT ~< directive treats
677 non-zero minpad parameter correctly.
679 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
680 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
681 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
682 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
683 the supported interface.
684 * new feature: Single-stepping of code compiled with DEBUG 2 or
685 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
686 * new feature: saving cores with foreign code loaded is now
687 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
688 Timothy Moore's work for CMUCL)
689 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
690 parameters to * instead of NIL if no initform is supplied.
691 (reported by Johan Bockgaard)
692 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
693 correctly when FUNCALL appears as the car of the form. Note:
694 despite this FUNCALL forms are not currently subject to
695 compiler-macro expansion. (port of Raymond Toy's fix for the
696 same from CMUCL, reported by Johan Bockgaard)
697 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
698 (thanks for Teemu Kalvas)
699 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
700 case instead of terminating the iteration (thanks for Julian
701 Squires, Sean Champ and Raymond Toy)
702 * bug fix: incorrect expansion of defgeneric that caused a style
703 warning. (thanks for Zach Beane)
704 * bug fix: slot accessor effective method computation works properly
705 for classes with multiple non-standard applicable methods on
706 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
707 * on x86 compiler supports stack allocation of results of LIST and
708 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
709 CMUCL implementation by Gerd Moellmann)
710 * fixed some bugs revealed by Paul Dietz' test suite:
711 ** FORMAT strings with both the ~<~:;~> form of the justification
712 directive and pretty-printing directives cause an error.
714 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
715 * incompatible change: the internal functions
716 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
717 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
718 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
719 instead of the old functions.
720 * new feature: on platforms where "dladdr" is available foreign
721 function names now appear in backtraces. (based on Helmut Eller's
723 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
724 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
726 * documentation: ASDF manual is now distributed with SBCL.
727 * bug fix: SBCL can now load its contributed modules with REQUIRE
728 even if the system-provided entries have been removed from
729 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
731 * bug fix: backtraces involving undefined functions or assembly
732 routines are more informative. (thanks to Brian Downing)
733 * bug fix: mutually referent alien structures now work correctly.
734 (reported by Rick Taube)
735 * bug fix: structures defined by WITH-ALIEN can be referred to
736 within other definitions in the same WITH-ALIEN.
737 * bug fix: division operators (MOD, TRUNCATE and the like) with
738 constant zero divisors and integer dividends no longer generate
740 * bug fix: provide default methods for INPUT-STREAM-P and
741 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
742 * bug fix: improve the detection and resolution of MOP metacycles
743 (where effective method computation depends on the generic function
744 having its effective method computed). (reported by Bruno Haible)
745 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
746 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
747 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
748 * bug fix: pathnames with relative directory components can be
749 represented relative to default pathnames.
750 * optimization: in taking the GCD of bignums, reduce the two bignums
751 to approximately the same size (using Euclid's algorithm) before
752 applying the more sophisticated binary GCD. (thanks to Juho
754 * optimization: COUNT on bitvectors now operates word-at-a-time.
755 * optimization: ASH with a positive, but not necessarily constant,
756 (leftwards) shift, when in a modular context, is compiled to a
758 * fixed some bugs revealed by Paul Dietz' test suite:
759 ** FORMAT variable parameters ("~V<char>") are defaulted properly
760 if the corresponding argument is NIL.
761 ** FORMAT directives accepting a minpad parameter treat negative
763 ** literal commas as character parameters to FORMAT directives are
765 ** literal spaces directly after ~<Newline> directives within a
766 format-logical-block (~:< ~@:>) do not induce :FILL-style
767 conditional newlines.
768 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
769 as specified: it no longer includes conditional newlines).
770 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
772 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
773 :PER-LINE-PREFIX argument does not evaluate to a string.
775 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
776 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
777 the manual for details; for now, package locks can be disabled by
778 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
779 if no major problems are found then it is likely that they will be
780 compiled in unconditionally.
781 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
782 unsupported operators on all platforms. To load a shared library
783 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
784 object file, link it into a shared library outside of SBCL and
785 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
787 * bug workaround: ROOM T and the GC were not working together
788 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
789 reporting facility aren't preserved. That reporting has been
790 disabled, so now until and unless someone figures out how to make it
791 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
792 * minor incompatible change: as threatened around sbcl-0.8.0, the
793 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
794 The symbols are also exported from SB-PCL for backwards
795 compatibility, but more so than before SB-PCL should be treated as
796 an implementation-internal package.
797 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
799 * the SB-SPROF contrib now works on (most) non-x86 architectures.
800 It is known as of this release not to work on the Alpha, however.
801 * fixed bug #167: errors signalled due to illegal syntax in method
802 bodies are now more legible.
803 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
804 designators and can hence be used with TYPEP.
805 * fixed bug #333: CHECK-TYPE now ensures that the type error
806 signalled, if any, has the right object to be accessed by
807 TYPE-ERROR-DATUM. (reported by Tony Martinez)
808 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
809 5.1.2.3 for multiple-value place subforms. (reported by Kalle
811 * fixed bug #334: programmatic addition of slots using specialized
812 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
813 and :ALLOCATION :CLASS slots.
814 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
815 float exponents. (rereported by Peter Seibel)
816 * fixed a bug: #\Space (and other whitespace characters) are no
817 longer considered to be macro characters in standard syntax by
819 * fixed bug: initialization of condition class metaobjects no longer
820 causes an instance of the condition to be created. (reported by
822 * fixed bug: it is now possible to have slots such that
823 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
824 description is neither :INSTANCE nor :CLASS.
825 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
826 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
827 SLOT-VALUE-USING-CLASS), and no longer causes errors with
828 non-standard SLOT-DEFINITION-ALLOCATION values.
829 * fixed bugs: various race conditions handling exiting threads.
830 CL-PPCRE's thread torture test now passes.
831 * fixed arguably-a-bug: GC time proportional to number of threads in
832 system even when most of them are idle
833 * optimization: improved performance of BIT and SBIT on bit-vectors.
834 * .fasl file incompatibility: The fasl file version number has
835 been incremented because of changes associated with package locks.
837 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
838 * minor incompatible change: the system no longer provides
839 optimization hints (conditions which are TYPEP
840 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
841 mode; these hints are still emitted when higher SPEED optimization
843 * new contrib module: a sampling profiler (profiling by statistical
844 sampling, rather than by instrumenting functions) is available as
845 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
846 * the behaviour of the standard function ED is now customizeable by
847 third parties through a hook variable: see ED's documentation
848 string for information on the protocol.
849 * the compiler no longer emits efficiency notes for (FUNCALL X)
850 when the type of X is uncertain under default optimization
852 * fixed bug 276: mutating a binding of a specialized parameter to a
853 method to something that is not TYPEP the specializer is now
855 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
856 exhibit the required behaviour on the x86 platform. (thanks to
857 Peter van Eynde, Eric Marsden and Bruno Haible)
858 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
859 even for difficult arguments. (reported by Peter Graves)
860 * fixed bug 141a: the backquote printer now descends quoted
862 * fixed another bug in backquote printing: no more destructive
863 modification of the form's list structure. (reported by Brian
865 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
866 move between its address being taken and the call to
867 interrupt_thread, fixing a crashing race condition.
868 * the SB-POSIX contrib implementation has been adjusted so that it
869 no longer exhibits ridiculously poor performance when constructing
870 instances corresponding to C structs.
872 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
873 * minor incompatible change: the sb-grovel contrib now treats C
874 structures as alien (in the sense of SB-ALIEN) objects rather than
875 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
876 has implications for memory management of client code
877 (sb-grovel-returned objects must now be manually managed) and for
878 type safety (alien objects now have full types).
879 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
880 used to control emission of compiler diagnostics, rather than the
881 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
882 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
883 quality should be considered deprecated.
884 * install.sh now installs the user manual as well
885 * (not quite a new documentable feature, but worth considering in
886 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
887 beginnings of a semantically meaningful condition hierarchy is
888 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
890 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
891 designator as the defaults argument.
892 * fixed bug: Displaced arrays whose displaced-to array has become
893 too small now cause ARRAY-DIMENSION to signal an error, providing
894 for safer bounds-checking. (reported by Bruno Haible)
895 * fixed bug: DEFCLASS slot definitions with identical :READER and
896 :WRITER names now signal a reasonable error. (reported by Thomas
898 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
900 * fixed bug: Invalid slot specification errors now print correctly.
901 (thanks to Zach Beane)
902 * fixed bug 320: Shared to local slot value transfers in class
903 redefinitions now happen corrently from superclasses as
904 well. (reported by Bruno Haible)
905 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
907 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
908 improper lists correctly. (reported by Bruno Haible)
909 * fixed bug 313: source-transform for <fun-name> was erroneously
910 applied to a call of a value of a variable with name <fun-name>.
911 (reported by Antonio Menezes Leitao)
912 * fixed bug 307: The obsolete instance protocol ensures that
913 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
914 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
915 remove dead unknown-values globs from the middle of the stack.
916 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
917 easier to resume long computations after using *BREAK-ON-SIGNALS*
918 to diagnose and fix failures. (thanks to Nikodemus Siivola)
919 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
920 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
921 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
922 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
923 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
925 * fixed a (fairly theoretical) bug in string printing: if
926 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
927 does not have array-element-type equal to the most general string
929 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
930 function name. (thanks to Nikodemus Siivola)
931 * fixed bug: compile-time format string checker failed on
932 non-closed ~{. (reported by Thomas F Burdick)
933 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
934 don't warn on legal loop code involving more than one aggregate
935 boolean. (thanks to Nikodemus Siivola)
936 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
937 in keyword position to MAKE-PACKAGE was not being reported as
939 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
940 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
942 * fixed bugs 280 and 312: the checking for multiple definitions in a
943 file is less likely to become confused by uses of inline
945 * fixed bug: the #S reader macro performs the keyword coercion
946 specified for slot names. (reported by Kalle Niemitalo)
947 * fixed bug: lambda lists may contain symbols whose names start with
948 & but are not lambda-list-keywords; their occurrence triggers a
950 * fixed bug 321: define-method-combination argument lambda lists do
951 not cause invalid code to be generated when &optional or &aux
952 variables have default values. (reported by Bruno Haible)
953 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
954 have CLOS classes; this appears also to have fixed reported
955 instability in stack exhaustion detection.
956 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
957 MUFFLE-WARNING when no associated restart is present is now
959 * optimization: rearranged the expansion of various defining macros
960 so that each expands into only one top-level form in a
961 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
963 * optimization: used a previously little-used slot in symbols to
964 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
966 * fixed some bugs revealed by Paul Dietz' test suite:
967 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
969 ** Fixed an optimization bug related to inheritance of initforms
970 from local to shared slots.
971 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
972 ** CAST optimizer forgot to flush argument derived type.
973 ** print/read consistency on floats is now orders of magnitude
974 more likely. (thanks also to Bruno Haible for a similar report
976 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
977 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
978 for objects of type REAL. Make it so.
979 ** SXHASH is specified (infelicitously) to respect similarity,
980 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
981 it so. (thanks to Markus Ziegler)
982 ** on the Alpha, the compiler succeeds in compiling functions
983 returning a known number of arguments greater than 63.
984 ** fixed handling of invalid NIL arguments in keyword position
986 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
988 ** provided a readably-printable representation for RANDOM-STATE
990 ** ensured that pathnames, where they have a namestring, always
991 print using #P"..." syntax.
993 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
994 * Support for the forthcoming 2.0 version of the NetBSD kernel
995 running on x86 hardware has been added. (thanks to Perry
996 E. Metzger most immediately, and others for their past work)
997 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
998 current release version; SBCL's OpenBSD support had been broken
999 since about the time of OpenBSD's switch to ELF binary format.)
1000 * [placeholder for DX summary]
1001 ** user code with &REST lists declared dynamic-extent, under high
1002 speed or space and low safety and debug optimization policy.
1003 * The manual has been converted to Texinfo format and the debugger
1004 chapter from the cmucl manual has been added.
1005 * A facility has been added to extract documentation strings from
1006 sbcl and store them as Texinfo-formatted snippets for inclusion in
1007 the manual (via Texinfo's @include directive)
1008 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1009 the test case to Dave Roberts)
1010 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1011 pointers. (reported by Sean Ross)
1012 * bug fix: PROFILE output is printed nicely even for large numerical
1013 values. (thanks to Zach Beane)
1014 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1015 greater than 32 handle EOF correctly.
1016 * bug fix: on X86 an immediate argument of the IMUL instruction is
1017 correctly printed in disassembly. (thanks to Lutz Euler)
1018 * bug fix: class slots in redefined classes preserve their old
1019 values. (thanks to Bruno Haible and Nikodemus Siivola)
1020 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1021 succeeds. (reported by Marco Baringer)
1022 * fixed some bugs revealed by Paul Dietz' test suite:
1023 ** READ-SEQUENCE now works on ECHO-STREAMs.
1024 ** RATIONALIZE works more according to its specification. (thanks
1027 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1028 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1029 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1030 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1031 should matter to you only if you rebind the printer control
1032 variables and then find you want different bindings in the
1033 debugger than in the ordinary execution of your program.)
1034 * The runtime build system has been tweaked to support building
1035 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1036 assembler and linker. (thanks to Nikodemus Siivola)
1037 * Unbound, undefined, undeclared variables now trigger full
1038 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1039 more often programmer error than deliberate exploitation of undefined
1041 * optimization: the hash algorithm for strings has changed to one
1042 that is less vulnerable to spurious collisions. (thanks to Juho
1044 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1045 needless bounds checking. (thanks to Juho Snellman)
1046 * optimization: implemented multiplication as a modular
1047 (UNSIGNED-BYTE 32) operation on the PPC backend.
1048 * fixed some bugs revealed by Paul Dietz' test suite:
1049 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1051 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1052 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1053 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1055 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1057 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1059 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1060 their output stream on EOF from read.
1061 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1062 have been read to end-of-file.
1063 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1065 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1066 description of determination of which consecutive characters
1068 ** Printing the "Space" character with escaping on now yields "#\\ ",
1069 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1070 ** Reading floating-point numbers with *READ-BASE* set to a number
1071 less than 10 works correctly.
1072 ** Reading floating-point numbers with *READ-BASE* set to a number
1073 more than 10 works correctly.
1074 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1075 the readtable currently in effect.
1077 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1078 * minor incompatible change: parsing of namestrings on a physical
1079 (Unix) host has changed; numbers after the final #\. in a
1080 namestring are no longer interpreted as a version field. This is
1081 intented to be largely invisible to the user, except that the
1082 meaning of the namestring "*.*.*" has changed: it now refers to a
1083 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1084 should usually be replaced by
1085 (make-pathname :name :wild :type :wild :version :wild)
1086 with the added benefit that this is more likely to be portable.
1087 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1088 option to OPEN now signals an error if the file being opened
1089 exists; this may have an impact on existing code.
1090 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1091 signals received, on the PowerPC platforms (both Linux and
1092 Darwin). (thanks to Pierre Mai for pointing out the location of
1094 * several fixes on OS X: The system now builds and runs cleanly on
1095 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1096 (thanks to Brian Mastenbrook)
1097 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1098 second-resolution: integer multiples of 1/3600 between -24 and 24.
1099 (thanks to Vincent Arkesteijn)
1100 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1101 type when called with 1 argument; PEEK-CHAR checked type of
1102 PEEK-TYPE only after having read first character from a
1103 stream. (reported by Peter Graves)
1104 * bug fix: the garbage collector now has much better locality
1105 behaviour, and in particular no longer treats all memory as being
1106 exhausted when many small objects point to each other in a deeply
1108 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1109 recognized as being TYPEP their class.
1110 * bug fix: the PUSHNEW documentation string has been corrected.
1111 (thanks to Vincent Arkesteijn)
1112 * bug fix: defaulting of the value for the last of an atypically
1113 large number of multiple values being bound was not being
1114 performed correctly on the Alpha or PPC platforms
1115 * optimization: implemented multiplication as a modular
1116 (UNSIGNED-BYTE 32) operation on the x86 backend.
1117 * optimization: SEARCH on simple-base-strings can now be open-coded.
1118 (see also contrib/compiler-extras.lisp for inspiration for
1119 teaching the compiler about the Boyer-Moore algorithm).
1120 * value, returned by MAX (and MIN) called with several EQUALP, but
1121 not EQL, arguments now does not depend on compiler settings.
1122 * fixed some bugs revealed by Paul Dietz' test suite:
1123 ** in stack analysis liveness information is propagated from
1124 non-local entry points.
1125 ** pathwise CAST removing failed when the CAST node did not start
1127 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1128 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1130 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1131 host is already defined.
1132 ** RENAME-FILE works on streams instead of signalling an internal
1134 ** PEEK-CHAR uses the current readtable when determining whether
1135 or not a character is whitespace.
1136 ** MERGE-PATHNAMES handles the case when the pathname does not
1137 specify a name while the default-pathname specifies a version.
1138 ** Pathnames now stand a better chance of respecting print/read
1140 ** Attempting to use standardized file system operators with a
1141 pathname with invalid :DIRECTORY components signals a
1143 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1145 ** DIRECTORY on logical pathnames is more correct.
1146 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1147 signal a TYPE-ERROR if their argument is not a stream.
1148 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1149 designator argument does not designate a stream.
1150 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1151 examining the synonym.
1152 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1154 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1155 element-type larger than ([UN]SIGNED-BYTE 32).
1157 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1158 * When built with the :SB-FUTEX feature, threaded builds now take
1159 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1160 for faster/more reliable mutex and condition variable support.
1161 * Incompatible change (but one you probably shouldn't have been using
1162 anyway): the interface and code for arbitrating between multiple
1163 threads in the same user session has been redesigned.
1164 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1165 error when called without an explicit environment argument.
1166 (thanks to Nikodemus Siivola)
1167 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1168 confused about their position. (thanks to Adam Warner and Gerd
1170 * bug fix: Pretty printing backquoted forms with unquotations in the
1171 argument list position of various code constructs such as LAMBDA
1172 now works correctly. (reported by Paul Dietz)
1173 * bug fix: Pretty printing unquotations no longer loses all
1174 stream position information.
1175 * optimization: performance of string output streams is now less
1176 poor for multiple small sequence writes.
1177 * optimization: performance of CSUBTYPEP in the presence of complex
1178 expressions involving CONS and NOT many times has been improved.
1179 (reported by Paul Dietz)
1180 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1182 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1183 streams for the REPL, for the debugger, and so on. (thanks to
1185 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1186 * fixed some bugs revealed by Paul Dietz' test suite:
1187 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1188 is ignored if it is too small, rather than propagating through
1189 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1190 ** extremely complex negations of CONS types were not being
1191 sufficiently canonicalized, leading to inconsistencies in
1193 ** VALUES tranformer lost derived type.
1195 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1196 * fixed a bootstrapping bug: the build process no longer assumes
1197 that the various BOOLE-related constants have the same value in
1198 host and target lisps. (noted by Paul Dietz' test suite on an
1199 SBCL binary built from CLISP)
1200 * The system can now be dynamically linked on the MIPS platform,
1201 which enables dynamic loading of foreign code from Lisp. (thanks
1202 to Ralf Baechle for discussions on the MIPS ABI)
1203 * The system now records debugging information for its own source
1204 files in a filesystem-position-independent manner, relative to
1205 the "SYS" logical host.
1206 * fixed a compiler bug: MV-LET convertion did not check references
1207 to the "max args" entry point. (reported by Brian Downing)
1208 * tweaked disassembly notes to be less confident about proclaiming
1209 some instruction as an LRA. (thanks to Brian Downing)
1210 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1211 now each have their own history, command character, and other
1212 characteristics. (thanks to David Lichteblau)
1213 * fixed some bugs revealed by Paul Dietz' test suite:
1214 ** compiler failure in compiling LOGAND expressions including a
1216 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1217 shift greater than 32.
1218 ** FLUSH-DEST did not mark blocks for type check regeneration.
1219 ** HANDLER-CASE failed to accept declarations in handler clauses
1220 in some circumstances.
1222 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1223 * New code in contrib/sb-introspect (still probably not entirely
1224 stable yet) provides some support for smart Lisp development
1225 environments like SLIME.
1226 * The conditions signalled for errors occurring when loading .fasl
1227 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1228 in a way which should help ASDF recover gracefully.
1229 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1230 stuff has been cleaned up. If you code contrib/ stuff, this might
1231 affect you, and you can look at contrib/README, contrib/STANDARDS,
1232 and/or the 0.8.4.27 diff to check.
1233 * In full calls the compiler now does not generate checks for declared
1234 argument types for all arguments.
1235 * various threading fixes
1236 ** and some experimental patches which didn't make it into
1237 the main tree for this release, but which are shipped in
1238 contrib/experimental-thread.patch as a possible fix for some
1239 failures (deadlock, spinning...) in GC-intensive multithreaded
1241 * fixed PPC build problem (source code incompatibility of different
1242 library versions): added offsetof() hackery which attempts to divine
1243 where glibc maintainers put uc_mcontext today
1244 * fixed bug 282: compiler does not trust type assertions while passing
1245 arguments to a full call.
1246 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1247 &OPTIONAL VALUES type parameter.
1248 * fix bug 214: algorithm for noting rejected templates is now more
1249 similar to that of template seletion. (also reported by rydis on
1251 * fixed bug 141b: printing backquoted information readably and prettily
1252 inserts a space where necessary.
1253 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1254 (CHARACTER 10) are now reported as errors, rather than propagated
1255 as unknown types. (reported by piso on #lisp)
1256 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1257 with values NIL and :ERROR. (thanks to Milan Zamazal)
1258 * fixed bug 191c: CLOS now does proper keyword argument checking as
1259 described in CLHS 7.6.5 and 7.6.5.1.
1260 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1261 longer raise an error; further, using a list as a for-as-arithmetic
1262 counter now raises a meaningful error.
1263 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1264 now understood by sequence creation functions such as MAKE-SEQUENCE
1266 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1267 a STRING-INPUT-STREAM.
1268 * compiler enhancement: SIGNUM is now better able to derive the type
1270 * type declarations inside WITH-SLOTS are checked. (reported by
1272 * fixed some bugs revealed by Paul Dietz' test suite:
1273 ** incorrect optimization of TRUNCATE for a positive first
1274 argument and negative second.
1275 ** compiler failure in let-convertion during flushing dead code.
1276 ** compiler failure while deriving type of TRUNCATE on an
1277 interval, containing 0.
1278 ** ASH of a negative bignum by a negative bignum count now returns
1280 ** intersection of CONS types now canonicalizes properly, fixing
1281 inconsistencies in SUBTYPEP.
1283 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1284 * incompatible change: The --disable-debugger command line
1285 option now clobbers the debugger at a more fundamental
1286 level, by redefining #'INVOKE-DEBUGGER instead of by
1287 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1288 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1289 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1290 enter the debugger REPL and then suffer recursive errors
1291 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1292 while under the new system, BREAK in --disable-debugger mode
1293 terminates the system just as an unhandled error would.
1294 * fixed compiler performance when processing loops with a step >1;
1295 * bug fix: DOCUMENTATION now retrieves generic function
1296 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1297 support has been systematized, and now supports the methods
1298 specified by ANSI, along with a default method and a method for
1299 slot documentation. (reported by Nathan Froyd)
1300 * bug fix: effective methods associated with a generic function are
1301 no longer cached over a change of that generic function's method
1302 combination. (reported by Andreas Fuchs)
1303 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1304 names a non-existent directory. (thanks to Andreas Fuchs)
1305 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1306 the PPC platform, lead to stack corruption; this has been fixed.
1307 (reported by Rainer Joswig)
1308 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1309 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1310 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1311 arithmetic in general, is now much more reliable on the PPC
1313 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1314 platform now returns the right answer.
1315 * optimization: restored some effective method precomputation in
1316 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1317 precomputation is now tunable.
1318 * optimization: compiler-internal data structure use has been
1319 reviewed, and changes have been made that should improve the
1320 performance of the compiler by about 20%.
1321 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1322 simple-base-strings and simple-bit-vectors is improved.
1323 * optimization: the optimization of 32-bit logical and arithmetic
1324 functions introduced in version 0.8.3 on the x86 has been
1325 implemented on the mips, ppc and sparc platforms; an
1326 implementation of the same facility, but for 64-bit arithmetic,
1327 has been added for the alpha.
1328 * microoptimization: the compiler is better able to make use of the
1329 x86 LEA instruction for multiplication by constants.
1330 * bug fix: in some situations compiler did not report usage of
1331 generic arithmetic in (SPEED 3) policy.
1332 * bug 145b fix: compiler used wrong type specifier while converting
1333 MEMBER-types to numeric.
1334 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1336 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1337 index argument now works properly on the Alpha platform.
1338 * bug fix: floating point exception treatment on the Alpha platform
1340 * bug fix: FILE-POSITION works much better on string input and
1341 output streams. (thanks to Nikodemus Siivola)
1342 * bug fix: many threading/garbage collection symptoms sorted.
1343 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1344 might be pseudo-atomic.
1345 * internal change: Stopping for GC is now done with signals not
1346 ptrace. GC is now done in whichever thread wanted it, instead of
1348 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1350 * bug fix: result form in DO is not contained in the implicit
1352 * incompatible change: ICR structure is changed; the value part of
1353 CONTINUATION is now called LVAR; corresponding functions are
1354 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1355 * added type deriver for ISQRT (thanks to Robert E. Brown).
1356 * bug fix: better support for loading from the command line when an
1357 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1359 * fixed some bugs revealed by Paul Dietz' test suite:
1360 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1361 ** ROUND and FROUND now give the right answer when given very
1362 small float arguments.
1363 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1365 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1366 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1367 ** LCM should return a non-negative integer.
1368 ** PARSE-INTEGER returned the index of a terminator instead of the
1369 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1370 ** PARSE-INTEGER returned an incorrect index being applied to a
1372 ** LCM with two arguments of 0 returns 0 rather than signalling
1374 ** unsigned addition of a 32-bit constant with the high bit set no
1375 longer causes an internal compiler error on the x86.
1376 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1377 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1378 with negative last argument.
1379 ** byte specifiers with zero size and position no longer cause
1380 an error during type derivation.
1381 ** bignum multiplication on the Alpha platform now returns the
1383 * porting: The system now builds on SuSE AMD64, although it still
1384 generates a 32-bit binary.
1385 * .fasl file incompatibility: The fasl file version number has
1386 been incremented (because of the changes to internal compiler
1387 data structures referred to above).
1389 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1390 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1391 more accurately, on the Darwin kernel running on PowerPC hardware.
1392 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1393 * Compiler code deletion notes now signal a condition of type
1394 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1395 an associated MUFFLE-WARNING restart.
1396 * The compiler now performs limited argument count validation of
1397 constant format strings in FORMAT, and where appropriate in ERROR,
1398 CERROR and WARN. (thanks to Gerd Moellmann)
1399 * New ASDF-INSTALL contrib can be used for automatic download and
1400 installation of third-party Lisp code from CCLAN or other sites
1402 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1403 which forces another thread to execute a function supplied by the
1405 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1406 now accept and act upon their :ELEMENT-TYPE keyword argument.
1407 (reported by Martin Atzmueller, Edi Weitz)
1408 * bug fix: FILE-POSITION now accepts position designators up to
1409 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1410 is the greater. (thanks to Patrik Nordebo)
1411 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1412 Kalle Olavi Niemitalo)
1413 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1414 printable. (reported by Eric Marsden)
1415 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1416 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1417 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1418 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1419 (thanks to Stig Sandoe)
1420 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1421 UNREAD-CHAR now correctly includes the unread character in the
1422 target sequence. (thanks to Gerd Moellmann)
1423 * bug fix in threaded builds: the system can now be suspended and
1424 resumed by shell job control with minimal disruption.
1425 * bug fixes in times and timezones >2038AD
1426 * better handling of "where is GNU make?" problem in build scripts
1427 (thanks to Nikodemus Siivola)
1428 * new optimization: inside a named function any reference to a
1429 function with the same name is considered to be a self-reference;
1430 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1431 optimization quality.
1432 * new optimization on x86: logical functions and + now have
1433 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1434 used when the result is truncated to 32 bits.
1435 * VALUES declaration is partially enabled.
1436 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1437 * fixed some bugs revealed by Paul Dietz' test suite:
1438 ** The system now obeys the constraint imposed by
1439 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1440 types form a lattice under type intersection.
1441 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1442 ** ASSOC now ignores NIL elements in an alist.
1443 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1444 and (1+ MOST-POSITIVE-FIXNUM) answers.
1445 ** The addition of a method with invalid qualifiers to a generic
1446 function does not cause an error to be signalled immediately;
1447 a warning is signalled, and the error is generated only on
1448 calling the generic function.
1449 * changed .fasl file version number, in part to add type codes for
1450 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1451 obscure ANSI requirements
1453 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1454 * fixed bug 148: failure to inline-expand a local function left
1455 garbage, confusing the compiler.
1456 * fixed bugs 3cd: structure slot readers perform type check if the
1457 slot can have an invalid value (i.e. it is either not initialized
1458 or can be written with a less specific slot writer).
1459 * bug fix: the compiler now traps array references to elements off
1460 the end of an array; previously, the bounds checking in some
1461 circumstances could go off-by-one.
1462 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1464 * type declarations for array element types now obey the description
1465 on the CLHS page "Declaration TYPE", as per discussions on
1466 sbcl-help around 2003-05-08. This means that a declaration
1467 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1468 declaration, all references to BAR will be asserted or assumed
1469 (with THE, so dependent on compiler policy) to involve objects of
1470 type FOO. Note that no such declaration is implied in
1471 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1472 * declared types of functions from the "Conditions"
1473 chapter. (reported by Paul Dietz)
1474 * bug fix: CERROR accepts a function as its first argument.
1475 * bug fix: NTH an NTHCDR accept a bignum as index
1476 arguments. (reported by Adam Warner)
1477 * optimization: character compare routines now optimize comparing
1478 against a constant character. (reported by Gilbert Baumann)
1479 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1480 argument now works properly on the MIPS platform.
1481 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1482 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1483 declared to have a functional type, does not cause an error
1485 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1486 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1487 * fixed bug: if last continuation of a deleted block has a
1488 destination, this destination should be deleted too. (reported by
1490 * fixed a bug in the bootstrap process: the host compiler's values
1491 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1492 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1493 test case from Patrik Nordebo)
1494 * improved the ability of the disassembler on the PPC platform to
1495 provide helpful disassembly notes.
1496 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1497 the class in more cases than previously.
1498 * bug fix: FILE-POSITION now understands :START and :END for
1499 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1500 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1501 8) data. (thanks to David Lichteblau)
1502 * bug fix: it is possible to add a method to a generic function
1503 without lambda list.
1504 * bug fix: reader failed to signal END-OF-FILE inside an
1505 object representation. (reported by Nikodemus Siivola)
1506 * fixed some bugs revealed by Paul Dietz' test suite:
1507 ** LAST and [N]BUTLAST should accept a bignum.
1508 ** condition slot accessors are methods.
1509 ** (VECTOR NIL) is a subtype of STRING.
1511 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1512 * minor incompatible change: some nonsensical specialized lambda
1513 lists (used in DEFMETHOD) which were previously ignored now signal
1515 * minor incompatible change: the system is now aware of the types of
1516 variables in the COMMON-LISP package, and will signal errors for
1517 most violations of these type constraints (where previously they
1518 were silently accepted).
1519 * minor incompatible change: COMPILE-FILE now uses the freedom
1520 afforded (ANSI 3.2.2.3) to use derived function types for
1521 functions defined in the same file. This also permits the system
1522 to warn on static type mismatches and function
1523 redefinition. (Currently it does not work with high DEBUG level.)
1524 * minor incompatible change: VALUES declaration is disabled.
1525 * When issuing notes, the compiler now signals a condition of type
1526 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1527 restart for use in user handlers. It is expected that the
1528 COMPILER-NOTE condition will eventually become a condition
1529 supertype to a hierarchy of note types, which will then be
1530 handleable in a similar fashion. However, at the moment, no such
1531 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1532 but it's an implementation detail, not a classification for the
1534 * Changes in type checking closed the following bugs:
1535 ** type checking of unused values (192b, 194d, 203);
1536 ** template selection based on unsafe type assertions (192c, 236);
1537 ** type checking in branches (194bc).
1538 * A short form of VALUES type specifier has ANSI meaning (it has
1539 increased the number of situations when SBCL cannot perform type
1541 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1542 keywords or constants is permissible.
1543 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1544 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1545 classes. (thanks to Antonio Martinez)
1546 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1547 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1548 argument to a pretty-print stream. Instead, it leaves any such
1549 operations to DESCRIBE-OBJECT methods.
1550 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1552 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1553 lambda lists. (thanks to David Lichteblau)
1554 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1555 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1556 (thanks to Gerd Moellmann)
1557 * fixed bug 246: increased compilation speed of long
1558 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1560 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1562 * DEFCONSTANT now throws a condition of type
1563 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1564 constant to a non-EQL value; CONTINUE and ABORT restarts
1565 respectively change and preserve the value.
1566 * fixed bug 63: The code walker, part of the implementation of CLOS,
1567 is now better at handling symbol macros.
1568 * bug fix: There is no longer an internal implementation type named
1569 CL:LENGTH. (reported by Raymond Toy)
1570 * bug fix: In macro-like defining macros/special operators the
1571 implicit block does not enclose the lambda list.
1572 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1573 enclosed in parentheses) are not suitable as type specifiers, and
1574 their use properly signals an error now.
1575 * bug fix: An argument count mismatch for a type specifier in code
1576 being compiled no longer causes an unhandled error at compile
1577 time, but signals a compile-time warning.
1578 * fixed simple vector readable printing
1579 * bug fix: DESCRIBE takes more care over whether the class
1580 precedence list slot of a class is bound before accessing it.
1581 (reported by Markus Krummenacker)
1582 * bug fix: FORMATTER can successfully compile pretty-printer format
1583 strings which use variants of the ~* directive inside.
1584 * bug fix: SEARCH now applies its TEST predicate to the elements of
1585 the arguments in the correct order. (thanks to Wolfhard Buss)
1586 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1587 in type inference. (thanks to Robert E. Brown)
1588 * bug fix: Reading in symbols with an explicit package name of ""
1589 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1590 * fixed some bugs revealed by Paul Dietz' test suite:
1591 ** NIL is now allowed as a structure slot name.
1592 ** Arbitrary numbers, not just REALs, are allowed in certain
1593 circumstances in LOOP for-as-arithmetic clauses.
1594 ** Multiple class redefinitions before slot access no longer
1595 causes a type error.
1596 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1597 association between the name and a class.
1598 ** Generic functions with non-standard method-combination and over
1599 five methods all of which return constants no longer return NIL
1600 after the first few invocations. (thanks to Gerd Moellmann)
1601 ** CALL-NEXT-METHOD with no arguments now passes the original
1602 values of the arguments, even in the presence of assignment.
1603 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1605 ** DISASSEMBLE works with closures and funcallable instances.
1606 ** ADD-METHOD now returns the generic function, not the new method.
1607 ** FIND-METHOD signals an error if the lengths of the specializers
1608 is incompatible with the generic function, even if the ERRORP
1610 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1611 which its argument is a member.
1612 ** DEFCLASS only redefines the class named by its class-name
1613 argument if that name is the proper name of the class;
1614 otherwise, it creates a new class.
1615 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1616 of the UNBOUND-SLOT condition to the name of the slot.
1617 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1618 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1619 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1620 SLOT-MAKUNBOUND in the specified fashion.
1622 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1623 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1624 cross-compilation host. As a consequence, we can now bootstrap our
1625 way up to SBCL starting with a bare gcc toolchain and human-readable
1626 source code (first the source to CLISP, then the source to SBCL).
1627 * A contributed module containing a partial implementation of the
1628 simple-streams interface has been included. (thanks to Rudi
1630 * A contributed module implementing the RFC1321 Message Digest
1631 Algorithm, known as MD5, has been included.
1632 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1633 no longer has any effect, as the code controlled by this feature
1634 has been deleted. (As far as we know, no-one has ever built using
1635 this feature, and its semantics were confused in any case).
1636 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1637 work on conditions (as required by the ANSI specification),
1638 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1639 expected behaviour on conditions. Users should note, however,
1640 that such behaviour is not required by the ANSI specification,
1641 and so use of this behaviour may render their code unportable.
1642 * fixed some bugs revealed by Paul Dietz' test suite:
1643 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1645 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1646 ** SXHASH on condition objects no longer returns NIL.
1647 ** :ALLOCATION :CLASS slots are better treated; their values are
1648 updated on class redefinition, and initforms inherited from
1649 superclasses are applied.
1650 ** REMOVE-METHOD returns its generic function argument even when
1651 no method was removed.
1652 ** SHARED-INITIALIZE now initializes the values of the requested
1653 slots, including those with :ALLOCATION :CLASS.
1654 ** ALLOCATE-INSTANCE now works on structure classes defined via
1655 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1657 ** SLOT-EXISTS-P now works on conditions, as well as structures
1659 ** MAKE-LOAD-FORM now has the required methods on
1660 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1661 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1662 keyword, and now implements the SLOT-NAMES argument.
1663 ** methods with &OPTIONAL arguments no longer allow too many
1664 arguments to be passed in the call without error.
1665 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1666 option is consistent with the required arguments of the generic
1667 function lambda list.
1668 * bug fix: REQUIRE accepts a string designator. (Thanks to
1670 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1671 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1672 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1673 * bug fix: The compiler now checks for duplicated variables in macro
1675 * bug fix: SETQ on globals returns the correct value.
1676 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1677 not a COMPILER-ERROR (followed by some other strange error on
1678 choosing the CONTINUE restart).
1679 * bug fix: make.sh and friends are now more consistent in the way that
1680 they look for GNU "make".
1682 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1683 * experimental native threads support (on x86 Linux >=2.4 only).
1684 This is not compiled in by default: you need to add :SB-THREAD to
1685 the target features. See the "Beyond ANSI" chapter of the manual
1687 * fix for longstanding nonANSIism: The old distinction between
1688 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1689 The return value from CL:FIND-CLASS is now a CLOS class, and
1690 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1691 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1692 * An interface to the MetaObject Protocol, as described in Kiczales,
1693 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1694 MIT Press, 1991, is available from the SB-MOP package.
1695 * incompatible change: the SB-PCL package should now be considered
1696 a private implementation detail, and no longer a semi-private MOP
1698 * minor incompatible change: due to rearrangement for threads, the
1699 control stack and binding stack are now allocated at arbitrary
1700 addresses instead of being hardcoded per-port. Users affected by
1701 this probably have to be doing advanced things with shared
1702 libraries, and will know who they are.
1703 * minor incompatible change: Previously, all --eval forms used were
1704 processed with READ before any of them were processed with EVAL.
1705 Now each --eval form is processed with both READ and EVAL before
1706 the next --eval form is processed. (Thus package operations like
1707 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1708 work as the user might reasonably expect.)
1709 * minor incompatible change: *STANDARD-INPUT* is now only an
1710 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1712 * minor incompatible change: Y-OR-N-P is now character-oriented,
1713 not line oriented. Also, YES-OR-NO-P now works without errors.
1714 (thanks to Antonio Martinez)
1715 * sb-aclrepl module improvements: an integrated inspector, added
1716 repl features, and a bug fix to :trace command.
1717 * Known functions, which cannot be open coded by the backend, are
1718 considered to be able to check types of their arguments. (fixing
1719 a bug report by Nathan J. Froyd)
1720 * fixed a bug in computing method discriminating functions: It is
1721 now possible to define methods specialized on classes which have
1722 forward-referenced superclasses. (thanks to Gerd Moellmann)
1723 * fixed evaluation order in optional entries (reported by Gilbert
1725 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1726 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1727 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1728 required-by-AMOP NAME argument, as well as CLASS and
1729 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1730 * fixed bug 20: DEFMETHOD can define methods using names that are
1731 not the proper names of classes to designate class specializers.
1732 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1733 Unix file descriptors, instead of blowing up. (thanks to Antonio
1735 * Garbage collection refactoring: user-visible change is that a
1736 call to the GC function during WITHOUT-GCING will not do garbage
1737 collection until the end of the WITHOUT-GCING. If you were doing
1738 this you were probably losing anyway.
1739 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1740 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1741 TYPEP the latter but not the former.
1742 * The compiler issues a full WARNING for calls to undefined functions
1743 with names from the CL package.
1744 * MAP-INTO for a vector destination is open coded. (reported by
1745 Brian Downing on c.l.l)
1746 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1747 documentation string.
1748 * fixed some bugs revealed by Paul Dietz' test suite:
1749 ** COPY-ALIST now signals an error if its argument is a dotted
1751 ** Condition slots are now accessed more correctly in the presence
1752 of multiple initargs for a given slot.
1753 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1754 exclude restarts of the same name associated with a different
1756 ** DEFCLASS of forward-referenced classes with another
1757 forward-referenced class in the superclasses list no longer
1759 ** Condition slots are now initialized once each, not multiple
1760 times. (thanks to Gerd Moellmann)
1761 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1763 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1764 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1765 arguments contain duplicated elements.
1766 ** RESTART-CASE understands local macros.
1767 ** RESTART-CASE associates exactly its own restarts with a condition.
1768 ** ENDP in safe mode checks its argument to be of type LIST.
1769 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1770 in question is unbound.
1771 ** Optimized MAKE-INSTANCE functions no longer cause internal
1772 assertion failures in the presence of duplicate initargs.
1773 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1774 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1775 propagated even in the ctor optimized implementation of
1777 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1779 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1780 * a better implementation of SXHASH on (simple) bit vectors,
1781 measured both in execution speed and in distribution of results
1782 over the positive fixnums, has been installed. Likewise, a better
1783 implementation of EQUAL for simple bit vectors is now available.
1784 * fixed CEILING optimization for a divisor of form 2^k.
1785 * fixed bug 240 (emitting extra style warnings "using the lexical
1786 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1787 by Antonio Martinez)
1788 * fixed SXHASH, giving different results for NIL depending on type
1789 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1790 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1791 variable. (found by Rolf Wester)
1792 * SBCL does not ignore type declarations for special
1793 variables. (reported by rif on c.l.l 2003-03-05)
1794 * some bug fixes in contrib/sb-aclrepl/
1795 * fixed some bugs revealed by Paul Dietz' test suite:
1796 ** a bug in the CONS type specifier, whereby the CAR and CDR
1797 types got intertwined, has been fixed;
1798 ** the type system is now able to reason about the interaction
1799 between INTEGER and RATIO types more completely;
1800 ** APPEND, [N]REVERSE and NRECONC check that those their
1801 arguments, which must be proper lists, are really so;
1802 ** An array specialized to be unable to hold elements has been
1803 implemented, as required -- yes, really -- by ANSI;
1804 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1805 on malformed property lists;
1807 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1808 * incompatible packaging change: in line with Unix convention,
1809 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1810 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1811 * REQUIRE and PROVIDE are now optionally capable of doing something
1812 useful. See the documentation string for REQUIRE.
1813 * infrastructure for a managed SBCL contrib system: contributed
1814 modules in this release include:
1815 ** the ASDF system definition facility;
1816 ** an interface to the BSD Sockets API;
1817 ** an ACL-like convenience interface to the repl;
1818 (thanks to Kevin Rosenberg)
1819 ** an implementation of ROTATE-BYTE, with efficient implementation
1821 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1822 gives a closer approximation to the right answer than previously.
1823 (thanks to Raymond Toy)
1824 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1825 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1826 optional environment arguments, as required by ANSI.
1827 * fixed bugs in other functions taking environment objects, allowing
1828 calls with an explicit NIL environment argument to be compiled
1830 * fixed bug 228: primary return values from
1831 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1832 COMPILE or FUNCTION.
1833 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1834 :TYPE will no longer signal errors on innocuous objects.
1835 * fixed bug 231b: SETQ is better at respecting type declarations in
1836 the lexical environment.
1837 * fixed a bug in DEFCLASS: classes named by symbols with no or
1838 unprintable packages can now be defined.
1839 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1840 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1841 * fixed compiler failure related to checking types of functions.
1842 (reported by Robert E. Brown)
1843 * the compiler is now much more consistent in its error-checking
1844 treatment of bounding index arguments to sequence functions: in
1845 (SAFETY 3) code, errors will be signalled in almost all cases if
1846 invalid sequence bounding indices are passed to functions defined
1847 by ANSI to operate on sequences.
1848 * fixed a bug in the build procedure: documentation of SBCL-specific
1849 packages is now preserved and available in the final Lisp image.
1850 * lifted FDEFINITION lookup out of loops in the implementation of
1851 many list operations. (thanks to Robert E. Brown)
1852 * fixed a bug in the reader: the #n# reader macro now works for
1853 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1854 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1855 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1856 E. Brown and Tony Martinez respectively)
1857 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1858 SLOT-BOUNDP: in particular, it is now able to optimize them much
1859 better, and is now not vulnerable to having packages renamed.
1860 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1861 always called when a slot is not present in an instance. (thanks
1863 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1864 longer signal an error on unreading a character following EOF on
1865 the previous constituent stream. (thanks to Tony Martinez)
1866 * fixed some bugs revealed by Paul Dietz' test suite:
1867 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1868 not just nonnegative fixnums;
1869 ** the logical bit-array operators such as BIT-AND now accept an
1870 explicit NIL for their "opt-arg" argument (to indicate a
1871 freshly-consed result bit-array);
1872 ** ELT now signals an error on an invalid sequence index in safe
1874 ** the type system is now cleverer about negations of numeric
1875 types, and consequently understands the BIGNUM and RATIO types
1877 ** the type system is now cleverer about the interaction between
1878 INTEGER and RATIO types: while bugs still remain, many more
1879 cases are accurately computed;
1880 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1881 if it is in the last clause;
1882 ** CONSTANTLY now correctly returns a side-effect-free function in
1884 ** DECLARE is no longer treated as a special-operator; in
1885 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1886 * incremented fasl file version number due to the change in the
1887 DEFSTRUCT-SLOT-DESCRIPTION structure.
1889 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1890 * minor incompatible change: code processed by the "interpreter" or
1891 EVAL now has a compilation optimization policy of (DEBUG 2)
1892 (changed from (DEBUG 1)) to improve debuggability of interactive
1893 development, and to allow the use of the debug RETURN command in
1895 * an experimental implementation of the RETURN command for the
1896 debugger has been included. (thanks to Frederik Kuivinen)
1897 * fixed bug 62: constraints were not propagated into a loop.
1898 * fixed bug in embedded calls of SORT (reported and investigated by
1900 * fixed some bugs revealed by Paul Dietz' test suite:
1901 ** printing and reading of arrays with some dimensions having
1902 length 0 (thanks to Gerd Moellmann);
1903 ** BOA constructor with &AUX argument without a default value does
1904 not cause a type error;
1905 ** CONSTANTP now returns true for all self-evaluating objects.
1907 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1908 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1909 accessors that are related by inheritance, as specified in the
1910 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1911 Valtteri Vuorikoski)
1912 * The compiler is now able to inline functions that were defined in
1913 a complex lexical environment (e.g. inside a MACROLET).
1914 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1916 * Support for the upcoming FreeBSD-5.0 release has been included.
1917 (thanks to Dag-Erling Smorgrav)
1918 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1919 effect when it is not in a toplevel context.
1920 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1921 stand a better chance of being correct. (thanks to Gerd
1923 * fixed bug in COERCE, which now signals an error on coercing a
1924 rational to a bounded real type which excludes the expected
1926 * The compiler is now able to derive types more accurately from the
1927 COERCE and COMPILE functions.
1928 * fixed bug 223: functional binding is considered to be constant
1929 only for symbols in the CL package.
1930 * fixed bug 231: SETQ did not check the type of a variable being set
1931 (reported by Robert E. Brown)
1932 * A new optimization for MAKE-INSTANCE has been included, fixing
1933 various bugs (including relating to :ALLOCATION :CLASS slots and
1934 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1936 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1937 ** As required by ANSI, LOOP now disallows anonymous collection
1938 clauses such as COLLECT I in conjunction with aggregate boolean
1939 clauses such as THEREIS (= I 1);
1940 ** LOOP now signals an error when any variable is reused in the
1941 same loop (including the potentially useful construct analogous
1942 to WITH A = 1 WITH A = (1+ A);
1943 ** IT is only a special loop symbol within the first clause of a
1944 conditional loop clause;
1945 ** LOOP with a typed iteration variable over a hashtable now
1946 signals a type error iff it should.
1947 * fixed some other bugs revealed by Paul Dietz' test suite:
1948 ** FILE-STREAM now names the class previously known as FD-STREAM;
1949 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1950 argument) no longer signals an error;
1951 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1952 of the slot symbol, rather than using the current package
1953 ((:CONC-NAME "") continues to intern the slot's name in the
1955 * incremented fasl file version number, because of the incompatible
1956 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1957 of the new implementation of DEFINE-COMPILER-MACRO.
1959 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1960 * Support for building SBCL for MIPS platforms running in
1961 little-endian mode has now been checked in, and basic
1962 functionality on said platforms verified.
1963 * minor incompatible change: PCL now records the pathname of a file
1964 in which methods and the like are defined, rather than its
1966 * minor incompatible change: TRUENAME now considers the truename of
1967 a file naming a directory to be the pathname with :DIRECTORY
1968 component indicating that directory.
1969 * minor incompatible change: a NAMED clause in the extended form of
1970 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1971 reason for the previous behaviour is unclear.
1972 * more systematization and improvement of CLOS and MOP conformance
1973 in PCL (thanks to Gerd Moellman and Pierre Mai):
1974 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1976 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1977 primary methods with no specializers;
1978 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1980 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1981 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1982 FORWARD-REFERENCED-CLASSes; error reporting on
1983 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1985 ** SXHASH on CLOS instances now uses a slot internal to the
1986 instance to return different numbers on distinct instances,
1987 while preserving the same return value through invocations of
1989 ** DEFMETHOD signals errors when methods with longer incongruent
1990 lambda lists are added to generic functions;
1991 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1992 CLASS, as specified in AMOP;
1993 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1994 on the order returned by the primary method for classes of
1995 class STANDARD-CLASS;
1996 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1997 * fixed some bugs shown by Paul Dietz' test suite:
1998 ** DOLIST puts its body in TAGBODY;
1999 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2001 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2002 value producing form;
2003 ** if more variables are given to PROGV than values, extra
2004 variables are bound and made to have no value;
2005 ** NSUBSTITUTE on list arguments gets the right answer with
2007 ** ELT signals an error of type TYPE-ERROR when the index argument
2008 is not a valid sequence index;
2009 ** LOOP signals (at macroexpansion time) an error of type
2010 PROGRAM-ERROR when duplicate variable names are found;
2011 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2012 ** FUNCALL of special-operators now cause an error of type
2014 ** PSETQ now works as required in the presence of side-effecting
2015 symbol-macro places;
2016 ** NCONC accepts any object as its last argument;
2017 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2019 ** loop-for-as-package does not require a package to be explicitely
2021 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2023 * fixed bug 166: compiler preserves "there is a way to go"
2024 invariant when deleting code.
2025 * fixed bug 172: macro lambda lists with required arguments after
2026 &REST arguments now cause an error to be signalled. (thanks to
2028 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2029 bug. (thanks to Matthew Danish)
2030 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2032 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2033 arguments are lexically rebound. (thanks to Gerd Moellmann and
2035 * fixed bug 194: error messages are now more informative when there
2036 is no primary method applicable in a call to a generic
2037 function. (thanks to Gerd Moellmann)
2038 * fixed bug in command line argument checking (thanks to Julian
2040 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2041 (thanks to Matthew Danish)
2042 * incremented fasl file version number, because of the
2043 SXHASH-related changes in the layout of CLOS data structures
2045 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2046 * minor incompatible change: The runtime (the Unix executable named
2047 "sbcl") is now much pickier about the .core files it will load.
2048 Essentially it now requires .core files to descend from the same
2049 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2050 as the runtime does. (The intent is to prevent the crashes which
2051 can occur, and which can even be reported as mysterious failures,
2052 when people patch the sources or change the build parameters
2053 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2054 sbcl and .core files.)
2055 * fixed bug: VALUES-LIST is no longer optimized away.
2056 * fixed bug 142: The FFI conversion of C string values to Lisp
2057 string values no longer conses excessively. (thanks to Nathan
2058 Froyd porting Raymond Toy's fix to CMU CL)
2059 * began to systematize and improve MOP conformance in PCL (thanks to
2060 Nathan Froyd, Gerd Moellman and Pierre Mai):
2061 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2063 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2064 ** FINALIZE-INHERITANCE is now called on class finalization;
2065 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2066 argument precedence order.
2067 * fixed bug 202: The compiler no longer fails on functions whose
2068 derived types contradict their declared type.
2069 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2070 so it can be non-toplevel.
2071 * The fasl file version number has changed (because of the new
2072 implementation of DEFMACRO).
2073 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2074 safe code, that any length requirement by their type-specifier
2075 argument is valid. The exceptions to this are described in bug
2077 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2078 functions now check the types of their inputs as required by ANSI.
2079 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2080 introduced symbol is DECLAREd to be SPECIAL.
2081 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2082 * fixed bug: :COUNT argument to sequence functions may be negative
2083 * fixed bug: body of DO-SYMBOLS may contain declarations
2084 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2085 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2086 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2087 symbol macro only once
2088 * fixed printing of call frame when argument list is unavailable
2089 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2090 * compiler no longer signals WARNING on unknown keyword
2093 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2094 * A beta-quality port to the mips architecture running Linux,
2095 based on the old CMUCL backend, has been made. It has been tested
2096 on a big-endian kernel, and works sufficiently well to be able to
2097 rebuild itself; it has not been tested in little-endian mode.
2098 * fixed an inconsistency between gencgc.c and purify.c which made
2099 dumping/loading .core files unreliable
2100 * fixed bug 120a: The compiler now deals correctly with IFs where
2101 the consequent is the same as the alternative, instead of
2102 misderiving the return type. (thanks to Alexey Dejneka)
2103 * fixed bug 113: Logical pathnames are now dumpable (the logical
2104 host is resolved at load-time, throwing an error if it is not
2106 * fixed bug 174: FORMAT's error message is slightly clearer when a
2107 non-printing character is used in a format directive.
2108 * fixed several bugs in compiler checking of type declarations, i.e.
2109 violations of the Python "declarations are assertions" principle
2110 (thanks to Alexey Dejneka)
2111 * fixed several bugs in PCL's error checking (thanks to Gerd
2113 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2115 * fixed bug in compilation of functions as first class values
2116 (thanks to Antonio Martinez-Shotton)
2117 * The compiler's handling TYPE-ERRORs which it can prove will
2118 inevitably happen at runtime has been cleaned up and corrected
2119 in several ways. (thanks to Alexey Dejneka)
2120 * improved argument type checking for various basic arithmetic
2121 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2122 lavished on them in the past that they can be compiled in many
2123 ways in different special cases
2124 * fixed bug 181: compiler checks validity of user supplied type
2126 * cleaned up code flushing in optimization: Function calls which
2127 should signal errors for safety purposes (e.g. which ANSI says
2128 should signal errors when their arguments are of incorrect type)
2129 are no longer optimized away.
2130 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2131 * incremented fasl file version number, because changes in the
2132 implementation of sequence functions like COERCE caused
2133 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2134 in old inline expansions) to become undefined. (Actually these
2135 changes were later undone, so we might very well be binary
2136 compatible with 0.7.7 after all, but leaving the version number
2137 incremented seemed like the simplest and most conservative
2140 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2141 * An alpha-quality port to the parisc architecture running Linux,
2142 based on the old CMUCL backend, has been made. This, even more so
2143 than the other backends, should be considered still a work in
2144 progress; known problems include that the Linux kernel in 64-bit
2145 mode does not propagate the correct sigcontext structure to
2146 userspace, and consequently SBCL on a parisc64 kernel will not
2148 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2149 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2150 suppressed.) Also now that the compiler is looking at declarations
2151 in the environment, it checks optimization declarations as well,
2152 and suppresses inlining when (> DEBUG SPEED).
2153 * More fixes have been made to treatment of floating point exception
2154 treatment and other Unix signals. In particular, floating point
2155 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2156 * The detection and handling of control stack exhaustion (infinite
2157 or very deeply nested recursion) has changed. Stack exhaustion
2158 detection is now done by write-protecting pages at the OS level
2159 and applies at all optimization settings; when found, a
2160 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2161 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2162 be caught using IGNORE-ERRORS.
2163 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2164 correctly when transforming recursive LABELS functions to LETs.
2165 (thanks to Alexey Dejneka)
2166 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2167 that are names of constants or global variables.
2168 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2169 alien routines with docstrings.
2170 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2171 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2173 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2174 (thanks to Eric Marsden)
2175 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2176 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2177 * Bug fix: The SPARC backend can now compile functions involving
2178 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2179 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2180 to a functions expecting a PATHNAME-DESIGNATOR.
2181 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2182 lambda lists. (thanks to Alexey Dejneka)
2183 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2184 functions so that EXPORTing the name of the class doesn't cause
2185 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2186 to Antonio Martinez for reporting this)
2187 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2188 constant-folded without failing an assertion. (thanks to Einar
2189 Floystad Dorum for reporting this)
2190 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2191 types are now dealt with more correctly by the compiler.
2192 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2193 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2194 behaviour analogous to RENAME-FILE. This puts its behaviour more
2195 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2197 * The fasl file version number has changed again. (because of the
2198 bug fix involving the names of PCL MAKE-INSTANCE functions)
2200 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2201 * bug fix: Floating point exceptions are treated much more
2202 consistently on the x86/Linux and PPC/Linux platforms.
2203 * Array initialization with :INITIAL-ELEMENT is now much faster for
2204 cases when the compiler cannot open code the array creation, but
2205 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2206 array accesses have also seen a speed increase.
2207 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2208 specified. (This was at the root of some bad interactions between
2209 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2211 * bug fix: Internal error arguments for undefined functions are now
2212 computed correctly on the PPC/Linux platform.
2213 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2214 Raymond Toy's patch for CMU CL)
2215 * Support for the Solaris 9 operating environment has been included
2216 (thanks to Daniel Merritt)
2217 * A very ugly but hopefully complete draft of the missing FFI chapter
2218 of the manual has been created by reformatting the corresponding
2219 CMU CL manual chapter into (currently very ugly and incoherent)
2220 DocBook and bringing it up to date for SBCL behavior. Thus, the
2221 manual is now essentially complete, at least by my extreme
2222 once-and-only-once standards, whereby it's acceptable to refer to
2223 the doc strings of SB-EXT functions as the primary documentation.
2224 * The fasl file version number has changed again, due to cleanup of
2225 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2226 is no longer a static symbol.)
2228 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2229 * SBCL now builds with OpenMCL (version 0.12) as the
2230 cross-compilation host; also, more progress has been made toward
2231 bootstrapping under CLISP.
2232 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2234 * bug 158 fixed: The compiler can now deal with integer loop
2235 increments different from 1; fixing this turned out also to fix
2237 * bug 169 fixed: no more bogus warnings about using lexical bindings
2238 despite the presence of perfectly good SPECIAL declarations (thanks
2239 to David Lichteblau)
2240 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2241 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2242 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2244 * bug 180 fixed: Method combination specifications no longer ignore
2245 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2246 * bug fix: Structure type predicate functions now check their argument
2247 count as they should.
2248 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2249 correctly. (thanks to Pierre Mai)
2250 * minor incompatible change: The --noprogrammer option is deprecated
2251 in favor of the new --disable-debugger option, which is very similar.
2252 (The major difference is that it takes effect at a slightly different
2253 time at startup, causing handling of errors in --sysinit and
2254 --userinit files will be affected differently.) The
2255 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2256 been added to allow this functionality to be controlled from ordinary
2257 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2258 maintainers, who might want to run non-interactive scripts to
2259 build SBCL cores which will later be used interactively.)
2260 * minor incompatible change: The LOAD function no longer, when given
2261 a wild pathname to load, loads all files matching that pathname.
2262 Instead, an error of type FILE-ERROR is signalled.
2264 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2265 * bug 147 fixed: The compiler preserves its block link/count
2266 invariants more correctly now so that it doesn't crash. (thanks
2268 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2270 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2271 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2272 no-value-for-this-character as NIL (as specified by ANSI).
2273 * HOST-NAMESTRING on physical pathnames now returns a string that is
2274 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2275 (thanks to Christophe Rhodes)
2276 * The Alpha port handles icache flushing more correctly. (thanks to
2278 * More progress has been made toward bootstrapping under CLISP. (thanks
2279 to Christophe Rhodes)
2280 * The fasl file format has changed again, because dynamic loading
2281 on OpenBSD (which has non-ELF object files) motivated some cleanups
2282 in the way that foreign symbols are transformed and passed around.
2283 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2284 is no longer treated as whitespace by the reader, but instead as
2285 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2286 |A
\7fB|, instead of A as it used to.
2288 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2289 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2290 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2291 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2292 this as of 0.7.1.45, but was left out of the previous news section
2293 (thanks to Dan Barlow)
2294 * SBCL now runs on the Solaris operating system on SPARC architectures
2295 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2296 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2297 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2298 * SPARC backend cleanups, allowing builds of cores optimized for V8
2299 and V9 SPARCS, and also emission of code targeted to a particular
2300 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2302 * SBCL is closer to bootstrapping under CLISP, thanks to various
2303 fixes by Christophe Rhodes.
2304 * The fasl file format has changed again, to allow the compiler's
2305 INFO database to support symbol macros.
2306 * The user manual (in doc/) is formatted into HTML more nicely.
2307 (thanks to coreythomas)
2308 * The system is smarter about SUBTYPEP relationships, especially
2309 those involving NOT types (including types such as ATOM which are
2310 represented internally using NOT types). Thus SUBTYPEP is less
2311 likely to return (VALUES NIL NIL) in general, and in particular
2312 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2314 * The fasl file format has changed again, because the internal
2315 representation of types now includes a new slot to support the new
2316 SUBTYPEP-of-NOT-types logic.
2317 * (not a change in the main branch of SBCL, but a related prototype
2318 which can hopefully be merged into the main branch of SBCL in the
2319 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2320 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2321 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2322 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2323 dispatch macro character combinations. (thanks to Alexey Dejneka)
2324 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2325 that superficially look like logical namestrings correctly.
2326 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2328 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2329 default supplied-p) for &optional and &key arguments. (thanks to
2332 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2333 * incompatible change: The compiler is now less aggressive about
2334 tail call optimization, doing it only when (> SPACE DEBUG) or
2335 (> SPEED DEBUG). (This is an incompatible change because there are
2336 programs which relied on the old CMU-CL-style behavior to optimize
2337 away their unbounded recursion which will now die of stack overflow.)
2338 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2339 for non-GENCGC systems has been increased to 20M (since that
2340 seems much closer to the likely performance optimum for modern
2341 systems than the old 4M value was)
2342 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2343 and *DEBUG-PRINT-LEVEL*
2344 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2345 of CMU CL's support for SPARC, and various endianness and other
2346 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2347 * new syntactic sugar for the Unix command line: --load foo.bar is now
2348 an alternate notation for --eval '(load "foo.bar")'.
2350 ** The system now detects stack overflow and handles it gracefully,
2351 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2352 optimization settings. (This is a good thing in general, and
2353 its introduction in this version should be particularly timely
2354 for anyone whose code fails because of suppression of tail
2356 ** The system now hunts for the C variable "environ" in a more
2357 devious way, to avoid segfaults when the C library version differs
2358 between compile time and run time. (thanks to Christophe Rhodes)
2359 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2360 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2361 ** The compiler no longer issues bogus style warnings for undefined
2362 classes in the same source file as the DEFCLASSes which defined
2363 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2365 ** fixes in CONDITION class precedence list for undefined function
2366 errors (thanks to Alexei Dejneka)
2367 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2368 correctly. (thanks to Dan Barlow)
2369 ** portability fixes aiming at bootstrapping under CLISP (thanks
2370 to Dave McDonald and Christophe Rhodes)
2371 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2372 ** fixes in type translation and and type inference (thanks to
2374 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2375 ** various fixes in the new ports (thanks to Dan Barlow)
2376 * several changes related to debugging:
2377 ** suppression of tail recursion, as noted above
2378 ** stack overflow detection, as noted above
2379 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2380 is now the default. (For some time encapsulation has been more
2381 reliable than the breakpoint-based :ENCAPSULATE NIL
2382 implementation, at least on X86 systems; and I just noticed that
2383 encapsulation also seems closer to the spirit of the ANSI
2386 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2388 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2389 up properly again. (There was a packaging bug in 0.7.0 which
2390 left their definitions in SB-SYS::LOAD-FOREIGN and
2391 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2392 vital for most things which interface to C-level interfaces,
2393 like extensions working with sockets or databases or
2394 Perl-compatible regexes or whatever, and the need to fix
2395 this bug is the main reason that 0.7.1 was released so
2397 ** DEFGENERIC is now choosier about the methods it redefines, so that
2398 reLOADing a previously-LOADed file containing DEFGENERICs does
2399 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2400 cycle now works as it should. (thanks to Alexey Dejneka)
2401 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2402 fixed by Christophe Rhodes. (He actually submitted this patch
2403 months ago, and I delayed until after 0.7.0.)
2404 ** Bug 111 (internal compiler confusion about runtime checks on
2405 FUNCTION types) was fixed by Alexey Dejneka.
2406 * Some internal cleanups (getting rid of variables which aren't
2407 needed now that the byte interpreter is gone) caused the fasl
2408 file format number to change again.
2410 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2411 * major incompatible change: The default fasl file extension, i.e. the
2412 default extension for files produced by COMPILE-FILE, has changed
2413 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2415 ** There are many changes in the implementation of the compiler.
2416 SBCL is now essentially a compiler-only implementation of ANSI
2417 Common Lisp. EVAL still "interprets" a few special cases, but
2418 almost all the interesting cases are handled by creating
2419 a LAMBDA expression, calling COMPILE on it, then calling
2420 FUNCALL on the result.
2421 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2422 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2423 Since the code is newer, there might still be some new bugs
2424 (though not as many as before Martin Atzmueller's fixes:-). But
2425 the new code is substantially simpler and clearer, and hopefully
2426 any remaining bugs will be simpler, less fundamental, and more
2427 fixable then the bugs in the old code.
2428 ** The revised compiler is still a little unsteady on its feet.
2430 *** The debugging information it produces (particularly the names
2431 of FUNCTION objects) is sometimes much less useful than what
2432 the old compiler produced.
2433 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2434 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2435 within a MACROLET) has been temporarily weakened.
2436 ** There are new compiler optimizations for various functions:
2437 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2438 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2439 *** the math functions TRUNCATE, FLOOR, and CEILING
2440 *** the function-of-all-trades COERCE
2441 Mostly these should be transparent, but there's one
2442 potentially-annoying problem (bug 117): when the compiler
2443 inline-expands a function and does type analysis on the result,
2444 it can create control paths which have type mismatches, and
2445 when it can't prove that those control paths aren't taken,
2446 it will issue WARNINGs about the type mismatches. This is
2447 a particular problem in practice for the new sequence functions.
2448 It's not clear how this should be fixed, and for now, a
2449 workaround is given in the entry for 117 in the BUGS file.
2450 ** (Because of the interaction between the two previous items --
2451 occasional inlining problems and new inline expansions -- some
2452 of the new sequence function optimizations won't really kick in
2453 completely until debugging information, and then inlining, are
2454 straightened out in some future version.)
2455 * minor incompatible changes:
2456 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2457 DIRECTORY no longer implicitly promotes NIL slots of its
2458 pathname argument to :WILD. In particular, when you ask for the
2459 contents of a directory (which you used to be able to do without
2460 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2461 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2462 ** changes in behavior that ANSI explicitly defines to be
2463 implementation dependent:
2464 *** The new compiler-only implementation still conforms with ANSI,
2465 but acts a little different than before. Besides the obvious
2466 changes in performance tradeoffs (that the cost per form passed
2467 to EVAL has gone up, and the cost per form executed by EVAL
2468 has gone down), the behavior of the system changes a little
2469 because there are no longer any interpreted function objects.
2470 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2471 e.g. doing COMPILE on the output of interactive DEFUN is
2473 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2475 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2476 has changed from (:CL) to NIL.
2477 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2478 e.g. #\Newline and #\Tab, don't have names specified in the
2479 ANSI Common Lisp standard, is now based on their ASCII symbolic
2480 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2481 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2482 are no longer used for output.
2483 ** changes in internal implementation constants:
2484 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2485 4 million. (If your application spends a lot of time GCing and
2486 you have a lot of RAM, you might want to experiment with
2487 increasing it even more.)
2488 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2489 However, almost all old code should still continue to work without
2490 immediate update, as SB-C-CALL is now a (deprecated) nickname
2492 ** Old operator names in the style DEF-FOO are now deprecated in
2493 favor of new corresponding names DEFINE-FOO, for consistency with
2494 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2495 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2496 internal symbols, but a few supported extensions like
2497 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2498 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2499 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2500 etc. as you get deeper into recursive calls to the debugger
2501 command loop, instead of the old "5]", "5]]", "5]]]"
2502 sequence. (I was motivated to do this when squabbles between
2503 ILISP and SBCL left me very deeply nested in the debugger. In the
2504 short term, this change will probably provoke more ILISP/SBCL
2505 squabbles, but hopefully it will be an improvement in the long run.)
2506 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2507 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2508 function, since declaiming return types involving aliens is
2509 (1) annoyingly messy to do by hand and (2) vital to efficient
2510 compilation of code which calls such functions.
2511 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2512 longer reexported by the SB-EXT package. They're solely useful
2513 for alien code, so it seems more logical that you should get
2514 them from the SB-ALIEN package, not in SB-EXT.
2515 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2516 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2517 features. Instead, the code that they used to control is always
2518 built into the system.
2519 * many other bug fixes
2520 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2521 advantage of the new EVAL-WHEN stuff and to clean them up in
2522 general, and they are now more ANSI-compliant in a number of
2523 ways. Martin Atzmueller is responsible for a lot of this.
2524 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2526 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2527 *** correct ERROR type for various file operations
2528 *** some fixes for Lisp streams
2529 *** DEFMETHOD syntax checking
2530 *** changing old weird representation of debug information as
2531 strings (which, among their other deficiencies, don't transform
2532 correctly when you rename packages, and don't change their
2533 print representation when you change things like *PACKAGE*
2534 and *PRINT-LENGTH*) to symbols and lists of symbols
2535 He also made several improvements and fixed several bugs in DESCRIBE.
2536 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2538 *** misbehavior of WRITE-STRING/WRITE-LINE
2539 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2540 and several other LOOP problems as well
2541 *** DIRECTORY when similar filenames are present
2542 *** DEFGENERIC with :METHOD options
2543 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2544 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2545 *** argument ordering in FIND with :TEST option
2546 *** mishandled package designator argument in APROPOS-LIST
2547 *** various problems in the backquote readmacro
2548 *** a bug in APROPOS
2549 *** probably some others that I'm not describing very well here,
2550 since the CVS log documents them by reference to sbcl-devel
2551 messages, and the SourceForge archives aren't working well.:-(
2552 ** Dan Barlow improved the Alpha port (and is making progress on the
2553 PPC port, for those of you who think different).
2554 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2555 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2556 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2557 continues to work on the SPARC port (for those of you in a position
2558 to look down upon our little PC-compatible boxes from a great height).
2559 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2560 and uses that copy, rather than the current dynamic value, when
2561 it's trying to decide whether to truncate output. Thus e.g.
2562 (let ((*print-lines* 50))
2563 (pprint-logical-block (stream nil)
2565 (let ((*print-lines* 8))
2566 (print (aref possiblybigthings i) stream)))))
2567 should now truncate the logical block only at 50 lines, instead of
2568 often truncating it at 8 lines, as it did before.
2569 * The doc/cmucl/ directory, containing old CMU CL documentation
2570 from the time of the fork, is no longer part of the base system.
2571 SourceForge has shut down its anonymous FTP service, and with it
2572 my original plan for distributing the old CMU CL documentation
2573 there. For now, if you need these files you can download an old
2574 SBCL source release and extract them from it.
2575 * The fasl file version number changed again, for dozens of reasons,
2576 some of which are apparent above.
2578 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2579 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2580 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2581 patch, so that DISASSEMBLE output is much nicer.
2582 * The code in the SB-PROFILE package now seems reasonably stable.
2583 I still haven't decided what the final interface should look like
2584 (I'd like PROFILE to interact cleanly with TRACE, since both
2585 facilities use function encapsulation) but if you have a need
2586 for profiling now, you can probably use it successfully with
2587 the current CMU-CL-style interface.
2588 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2589 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2590 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2591 link now returns the dangling link itself, and for similar
2592 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2593 link itself. (In these cases the old code signalled an error and
2594 looped endlessly, respectively.) Thus, DIRECTORY now works even
2595 in the presence of dangling and cyclic symbolic links.
2596 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2597 is now a supported extension again, since the consensus on
2598 sbcl-devel was that it can be useful for ordinary development
2599 work, not just for debugging SBCL itself.
2600 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2601 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2602 that currently-defined functions might be redefined later with
2603 different return types.
2604 * Hash tables can be printed readably, as inspired by CMU CL code
2605 of Eric Marsden and SBCL code of Martin Atzmueller.
2606 * better error handling in CLOS method combination, thanks to
2607 Martin Atzmueller porting Pierre Mai's CMU CL patches
2608 * more overflow fixes for >16Mbyte I/O buffers
2609 * A bug in READ has been fixed, so that now a single Ctrl-D
2610 character suffices to cause end-of-file on character streams.
2611 In particular, now you only need one Ctrl-D at the command
2612 line (not two) to exit SBCL.
2613 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2615 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2616 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2617 does the right thing.
2618 * The implementation of some type tests, especially for CONDITION
2619 types, is now tidier and maybe faster, due to CMU CL code
2620 originally by Douglas Crosher, ported by Martin Atzmueller.
2621 * Some math functions have been fixed, and there are new
2622 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2623 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2624 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2625 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2626 likely to be built into sbcl-0.7.0.)
2627 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2628 CL, ported to SBCL by Martin Atzmueller)
2629 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2630 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2631 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2632 * There's a new slam.sh hack to shorten the edit/compile/debug
2633 cycle for low-level changes to SBCL itself, and a new
2634 :SB-AFTER-XC-CORE target feature to control the generation of
2635 the after-xc.core file needed by slam.sh.
2636 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2637 COMPILE-FILE is no longer supported, so that now every function
2638 gets an entry point, so that block compilation looks a little
2639 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2640 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2641 returns the number of bytes consed since the system started,
2642 rather than the number consed since the first time the function
2643 was called. (The new definition parallels ANSI functions like
2644 CL:GET-INTERNAL-RUN-TIME.)
2645 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2646 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2647 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2648 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2649 * Fasl file version numbers are now independent of the target CPU,
2650 since historically most system changes which required version
2651 number changes have affected all CPUs equally. Similarly,
2652 the byte fasl file version is now equal to the ordinary
2655 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2656 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2657 is no longer recognized. I apologize for this, because it was
2658 listed in SB-EXT as a supported extension, but I found that
2659 its existing behavior was poorly specified, as well as incorrectly
2660 specified, and it looked like too much of a mess to straighten it
2661 out. I have enough on my hands trying to get ANSI stuff to work..
2662 * many patches ported from CMU CL by Martin Atzmueller, with
2663 half a dozen bug fixes in pretty-printing and the debugger, and
2664 half a dozen others elsewhere
2665 * fixed bug 13: Floating point infinities are now supported again.
2666 They might still be a little bit flaky, but thanks to bug reports
2667 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2668 as flaky as they were.
2669 * The --noprogrammer command line option is now supported. (Its
2670 behavior is slightly different in detail from what the old man
2671 page claimed it would do, but it's still appropriate under the
2672 same circumstances that the man page talks about.)
2673 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2674 are now supported, and enabled by default. Thus, the compiler can
2675 handle many floating point and complex operations much less
2676 inefficiently. (Thus e.g. you can implement a complex FFT
2678 * The compiler now detects type mismatches between DECLAIM FTYPE
2679 and DEFUN better, and implements CHECK-TYPE more correctly, and
2680 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2681 thanks to patches from Martin Atzmueller.
2682 * various fixes to make the cross-compiler more portable to
2683 ANSI-conforming-but-different cross-compilation hosts (notably
2684 Lispworks for Windows, following bug reports from Arthur Lemmens)
2685 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2686 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2687 fixed thanks to Pierre Mai's CMU CL patches.
2688 * fixes to keep the system from overflowing internal counters when
2689 it tries to use i/o buffers larger than 16M bytes
2690 * fixed bug 45a: Various internal functions required to support
2691 complex special functions have been merged from CMU CL sources.
2692 (When I was first setting up SBCL, I misunderstood a compile-time
2693 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2694 * improved support for type intersection and union, fixing bug 12
2695 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2696 more obscure bugs as well
2697 * some steps toward byte-compiling non-performance-critical
2698 parts of the system, courtesy of patches from Martin Atzmueller
2699 * Christophe Rhodes has made some debian packages of sbcl at
2700 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2701 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2702 stable, but are nonetheless usable. When he's ready, I'd be happy
2703 to add them to the SourceForge "File Releases" section. (And if
2704 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2705 * new fasl file format version number (because of changes in
2706 internal representation of (OR ..) types to accommodate the new
2707 support for (AND ..) types, among other things)
2709 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2710 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2712 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2714 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2715 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2716 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2717 * DESCRIBE now works on structure objects again.
2718 * Most function call argument type mismatches are now handled as
2719 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2720 know whether the function will be redefined before the call is
2721 executed. (The compiler could flag local calls with full WARNINGs,
2722 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2723 it doesn't keep track of enough information to know whether calls
2724 are local in this sense.)
2725 * Compiler output is now more verbose, with messages truncated
2726 later than before. (There should be some supported way for users
2727 to override the default verbosity, but I haven't decided how to
2728 provide it yet, so this behavior is still controlled by the internal
2729 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2730 src/compiler/ir1util.lisp.)
2731 * Fasl file format version numbers have increased again, because
2732 support for the Gray streams extension changes the layout of the
2733 system's STREAM objects.
2734 * The Gray subclassable streams extension now works, thanks to a
2735 patch from Martin Atzmueller.
2736 * The full LOAD-FOREIGN extension (not just the primitive
2737 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2738 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2739 but like most other programs, it defaults to copying the Unix
2740 environment from the original process instead of starting the
2741 new process in an empty environment.
2742 * Extensions which manipulate the Unix environment now support
2743 an :ENVIRONMENT keyword option which doesn't smash case or
2744 do other bad things. The CMU-CL-style :ENV option is retained
2745 for porting convenience.
2746 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2747 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2749 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2751 * A patch from Martin Atzmueller seems to have solved the SIGINT
2752 problem, and as far as we know, signal-handling now works cleanly.
2753 (If you find any new bugs, please report them!)
2754 * The system no longer defaults Lisp source file names to types
2755 ".l", ".cl", or ".lsp", but only to ".lisp".
2756 * The compiler no longer uses special default file extensions for
2757 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2758 seems to expect a single default extension for all compiled code,
2759 and there's no compelling reason to try to stretch the standard
2760 to allow two different extensions.) Instead, byte-compiled files
2761 default to the same extension as native-compiled files.
2762 * Fasl file format version numbers have increased again, because
2763 a rearrangement of internal implementation packages made some
2764 dumped symbols in old fasl files unreadable in new cores.
2765 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2766 many fewer weird special cases.
2767 * Bug #17 (differing COMPILE-FILE behavior between logical and
2768 physical pathnames) has been fixed, and some related misbehavior too,
2769 thanks to a patch from Martin Atzmueller.
2770 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2771 by Tim Moore, ported to SBCL by Martin Atzmueller.
2772 * Martin Atzmueller fixed several filesystem-related problems,
2773 including bug #36, in part by porting CMU CL patches, which were
2774 written in part by Paul Werkowski.
2775 * More compiler warnings in src/runtime/ are gone, thanks to
2776 more patches from Martin Atzmueller.
2777 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2780 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2782 * DESCRIBE now works on CONDITION objects.
2783 * The debugger now handles errors which arise when trying to print
2784 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2786 * The build system now uses an additional file, customize-target-features.lisp,
2787 to allow local modifications to the target *FEATURES* list. (The point of
2788 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2789 debugging features enabled, without having to worry about propagating it
2790 into everyone's system when I do a "cvs update".) When no
2791 customize-target-features.lisp file exists, the target *FEATURES* list
2792 should be constructed the same way as before.
2793 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2794 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2795 warn about behavior which is undefined under the ANSI spec. Note: This
2796 is specified by ANSI, but it's not very popular with programmers.
2797 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2798 macro in the SBCL sources for an example of a workaround which you
2799 might use to make portable ANSI-standard code which does what you want.
2800 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2801 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2802 when it's used as a non-toplevel form. (This is required in order
2803 to implement the DEFCONSTANT-EQX macro.)
2804 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2805 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2806 funny things, due to bugs in the implementation of EVAL-WHEN
2807 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2808 non-toplevel DEFCONSTANTs are uncommon.)
2809 * The core file version number and fasl file version number have been
2810 incremented, because the old noncompliant DEFCONSTANT behavior involved
2811 calling functions which no longer exist, and because I also took the
2812 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2813 * fixed bug 1 (error handling before read-eval-print loop starts), and
2814 redid debugger restarts and related debugger commands somewhat while
2816 ** The QUIT debugger command is gone, since it did something
2817 rather different than the SB-EXT:QUIT command, and since it never
2818 worked properly outside the main toplevel read/eval/print loop.
2819 Invoking the new TOPLEVEL restart provides the same functionality.
2820 ** The GO debugger command is also gone, since you can just invoke
2821 the CONTINUE restart directly instead.
2822 ** The TOP debugger command is also gone, since it's redundant with the
2823 FRAME 0 command, and since it interfered with abbreviations for the
2825 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2827 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2828 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2829 some porting work by Martin Atzmueller)
2830 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2831 with a boatload of patches which he ported to SBCL. Now that those
2833 ** The system tries to make sure that its low-priority messages
2834 are prefixed by semicolons, to help people who like to use
2835 syntax highlighting in their ILISP buffer. (This patch
2836 was originally due to Raymond Toy.)
2837 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2838 patches originally written by Raymond Toy.
2839 ** The compiler understands coercion between single-value and
2840 multiple-VALUES type expressions better, getting rid of some very
2841 weird behavior, thanks to patches originally by Robert MacLachlan
2842 and Douglas Crosher.
2843 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2844 lambda lists, thanks to a patch originally by Pierre Mai.
2845 ** The system no longer bogusly warns about "abbreviated type
2847 ** The compiler gets less confused by inlining and RETURN-FROM,
2848 thanks to some patches originally by Tim Moore.
2849 ** The system no longer hangs when dumping circular lists to fasl
2850 files, thanks to a patch originally from Douglas Crosher.
2851 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2852 undefined function error.
2853 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2854 in structure slot initforms) for now, documented workaround instead:-|
2855 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2856 * fixed bug 5: added stubs for various Gray stream functions called
2857 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2858 installed, at least appropriate type errors are generated
2859 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2860 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2861 consistently in DEFMETHOD forms.
2862 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2863 it doesn't seem to affect SBCL after all
2864 * The C runtime system now builds with better optimization and many
2865 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2867 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2869 * The system is now under CVS at SourceForge (instead of the
2870 CVS repository on my home machine).
2871 * The new signal handling code has been tweaked to treat register
2872 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2873 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2874 downstream, e.g. in debug-int.lisp, has implicit dependencies
2875 on the unsignedness of integer representation of machine words,
2876 and that caused the system to bomb out with infinite regress
2877 when trying to recover from type errors involving signed values,
2878 e.g. (BUTLAST '(1 2 3) -1).)
2879 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2880 (This was one of the bugs Peter Van Eynde reported back in July.)
2881 * The system now uses code inspired by Colin Walters' O(N)
2882 implementation of MAP (from the cmucl-imp@cons.org mailing
2883 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2884 inline the MAP operation, and there is more than one
2885 sequence argument to the MAP call (so that it can't just
2886 do ETYPECASE once and for all based on the type of the
2887 single sequence argument). (The old non-inline implementation
2888 of the general M-argument sequence-of-length-N case required
2889 O(M*N*N) time when any of the sequence arguments were LISTs.)
2890 * The QUIT :UNIX-CODE keyword argument has been renamed to
2891 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2892 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2894 * Raymond Wiker's patches to port dynamic loading from Linux to
2895 FreeBSD have been added.
2896 * The BUGS file is now more nearly up to date, thanks in large part
2897 to Martin Atzmueller's review of it.
2898 * The debugger now flushes standard output streams before it begins
2899 its output ("debugger invoked" and so forth).
2900 * The core version number and fasl file version number have both
2901 been incremented, because of incompatible changes in the layout
2903 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2904 used to not be called for a saved Lisp image.)
2905 * Martin Atzmueller's version of a patch to fix a compiler crash,
2906 as posted on sbcl-devel 13 September 2000, has been installed.
2907 * Instead of installing Martin Atzmueller's patch for the
2908 compiler transform for SUBSEQ, I deleted the compiler transform,
2909 and transforms for some similar consing operations.
2910 * A bug in signal handling which kept TRACE from working on OpenBSD
2912 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2913 away by constant folding
2914 * The system now defines its address space constants in one place
2915 (in the Lisp sources), and propagates them automatically elsewhere
2916 (through GENESIS and the sbcl.h file). Therefore, patching the
2917 address map is less unnecessarily tedious and error-prone. The
2918 Lisp names of address space constants have also been systematized.
2919 * CVS tags like dollar-Header-dollar have been removed from
2920 the sources, because they have never saved me trouble and
2921 they've been source of trouble working with patches and other
2922 diff-related operations.
2923 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2924 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2926 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2928 * The system has been ported to OpenBSD.
2929 * The system now compiles with a simple "sh make.sh" on the systems
2930 that it's supported on. I.e., now you no longer need to tweak
2931 text in the target-features.lisp-expr and symlinks in src/runtime/
2932 by hand, the make.sh takes care of it for you.
2933 * The system is no longer so grossly inefficient when compiling code
2934 involving vectors implemented as general (not simple) vectors (VECTOR T),
2935 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2936 which dares to use the various sequence functions on non-simple
2937 vectors, takes less of a performance hit.
2938 * There is now a primitive type predicate VECTOR-T-P
2939 to test for the (VECTOR T) type, so that e.g.
2940 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2941 can now be compiled with some semblance of efficiency. (The old code
2942 turned the type declaration into a full call to %TYPEP at runtime!)
2943 * AREF on (VECTOR T) is still not fast, since it's still compiled
2944 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2945 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2946 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2947 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2948 without making full calls to SUBTYPEP at runtime.
2949 (Some analogous efficiency issues for non-simple vectors specialized to
2950 element types other than T, or for non-simple multidimensional arrays,
2951 have not been addressed. They could almost certainly be handled the
2952 same way if anyone is motivated to do so.)
2953 * The changes in array handling break binary compatibility, so
2954 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2955 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2956 instead of (VALUES T T).
2957 * By following the instructions that Dan Barlow posted to sbcl-devel
2958 on 2 July 2000, I was able to enable primitive dynamic object
2959 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2960 functionality is not implemented (since it calls ld to resolve
2961 library references automatically, requiring RUN-PROGRAM for its
2962 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2963 try to resolve library references) is now supported.
2964 * The system now flushes the standard output streams when it terminates,
2965 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2966 them at several other probably-convenient times, e.g. in each pass of
2967 the toplevel read-eval-print loop, and after evaluating a form given
2968 as an "--eval" command-line option. (These changes were motivated by a
2969 discussion of stream flushing issues on cmucl-imp in August 2000.)
2970 * The source transform for TYPEP of array types no longer assumes
2971 that an array whose element type is a not-yet-defined type
2972 is implemented as an array of T, but instead punts, so that the
2973 type will be interpreted at runtime.
2974 * There is now some support for cross-compiling in make.sh: each of
2975 the phases of make.sh has its own script. (This should be transparent
2976 to people doing ordinary, non-cross-compile builds.)
2977 * Since my laptop doesn't have hundreds of megabytes of memory like
2978 my desktop machine, I became more motivated to do some items on
2979 my to-do list in order to reduce the size of the system a little:
2980 ** Arrange for various needed-only-at-cold-init things to be
2981 uninterned after cold init. To support this, those things have
2982 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2983 symbols with such names are now uninterned after cold init).
2984 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2985 fasl files for cold load.
2986 ** Remove the old compiler structure pooling code (which used to
2987 be conditional on the target feature :SB-ALLOC) completely.
2988 ** Redo the representation of some data in cold init to be more compact.
2989 (I also looked into supporting byte compiled code at bootstrap time,
2990 which would probably reduce the size of the system a lot, but that
2991 looked too complicated, so I punted for now.)
2992 * The maximum signal nesting depth in the src/runtime/ support code has
2993 been reduced from 4096 to 256. (I don't know any reason for the very
2994 large old value. If the new smaller value turns out to break something,
2995 I'll probably just bump it back up.)
2996 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2998 * Many, many bugs reported by Peter Van Eynde have been added to
2999 the BUGS list; some have even been fixed.
3000 * While enabling dynamic object file loading, I tried to make the
3001 code easier to understand, renaming various functions and variables
3002 with less ambiguous names, and changing some function calling
3003 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3004 * While trying to figure out how to do the OpenBSD port, I tried to
3005 clean up some of the code in src/runtime/. In particular, I dropped
3006 support for non-POSIX signal handling, added various comments,
3007 tweaked the code to reduce the number of compilation warnings, and
3008 renamed some files to increase consistency.
3009 * To support the new automatic configuration functionality in make.sh,
3010 the source file target-features.lisp-expr has been replaced with the
3011 source file base-target-features.lisp-expr and the machine-generated
3012 file local-target-features.lisp-expr.
3013 * fixed a stupid quoting error in make.sh so that using CMU CL
3014 "lisp -batch" as cross-compilation host works again
3016 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3018 * DESCRIBE no longer tries to call itself recursively to describe
3019 bound/fbound values, so that it no longer fails on symbols which are
3020 bound to themselves (like keywords, T, and NIL).
3021 * DESCRIBE now works on generic functions.
3022 * The printer now prints less-screwed-up representations of closures
3023 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3024 * A private symbol is used instead of the :EMPTY keyword previously
3025 used to mark empty slots in hash tables. Thus
3026 (DEFVAR *HT* (MAKE-HASH-TABLE))
3027 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3028 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3029 now does what ANSI says that it should. (You can still get
3030 similar noncompliant behavior if bang on the hash table
3031 implementation with all the symbols you get back from
3032 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3033 This breaks binary compatibility, since tests for equality to
3034 :EMPTY are wired into things like the macroexpansion of
3035 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3037 * There's now a minimal placeholder implementation for CL:STEP,
3038 as required by ANSI.
3039 * An obscure bug in the interaction of the normal compiler, the byte
3040 compiler, inlining, and structure predicates has been patched
3041 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3042 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3044 * Missing ordinary arguments in a macro call are now detected even
3045 when the macro lambda list contains &KEY or &REST.
3046 * The debugger no longer complains about encountering the top of the
3047 stack when you type "FRAME 0" to explicitly instruct it to go to
3048 the top of the stack. And it now prints the frame you request even
3049 if it's the current frame (instead of saying "You are here.").
3050 * As specified by ANSI, the system now always prints keywords
3051 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3053 * The default initial SIZE of HASH-TABLEs is now smaller.
3054 * Type information from CLOS class dispatch is now propagated
3055 into DEFMETHOD bodies, so that e.g.
3056 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3058 is now basically equivalent to
3059 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3060 (DECLARE (TYPE SINGLE-FLOAT X))
3062 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3063 operation, without having to do run-time type dispatch.
3064 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3065 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3066 * Since I don't seem to be making big file reorganizations very often
3067 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3068 is overflowing my ability to conveniently back them up), I've finally
3069 checked the system into CVS. (The CVS repository is on my home system,
3070 not at SourceForge -- putting it on SourceForge might come later.)
3071 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3072 high-level GC-NOTIFY-FOO functions send their output. (There's
3073 still very little control of where low-level verbose GC functions
3074 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3075 less than it used to -- the GC-NOTIFY-FOO functions are now under
3076 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3077 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3078 in only one place in the source code, and propagates it automatically
3079 everywhere that it's needed. Thus e.g. when I bump the version from
3080 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3081 * The C source files now include boilerplate legalese and documentation
3082 at the head of each file (just as the Lisp source files already did).
3083 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3084 to his page will be replaced with a link to his new CLiki service.
3086 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3088 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3089 * The build process now looks for GNU make under the default name "gmake",
3090 instead of "make" as it used to. If GNU make is not available as "gmake"
3091 on your system, you can change this default behavior by setting the
3092 GNUMAKE environment variable.
3093 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3094 can build without error under CMU CL.
3096 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3098 * There is now a partial SBCL user manual (with some new text and some
3099 text cribbed from the CMU CL manual).
3100 * The beginnings of a profiler have been added (starting with the
3101 CMU CL profiler and simplifying and cleaning up). Eventually the
3102 main interface should be through the TRACE macro, but for now,
3103 it's still accessed through vaguely CMU-CL-style functions and macros
3104 exported from the package SB-PROFILE.
3105 * Some problems left over from porting CMU CL to the new
3106 cross-compilation bootstrap process have been cleaned up:
3107 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3108 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3109 part when getting rid of the compiler *BACKEND* stuff.)
3110 ** The value of *NULL-TYPE* was screwed up, because it was
3111 being initialized before the type system knew the final
3112 definition of the 'NULL type. This screwed up several key
3113 optimizations in the compiler, causing inefficiency in all sorts
3114 of places. (I found it because I wanted to understand why
3115 GET-INTERNAL-RUN-TIME was consing.)
3116 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3117 PROCLAIM FTYPE information. Unfortunately this broke binary
3118 compatibility again, since now the forms output by DEFGENERIC
3119 to refer to functions which didn't exist in 0.6.3.
3120 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3121 can use the new (as of 0.6.3) transform for SOME into MAP into
3123 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3124 values to (INTEGER 0 1000000), so that the time code will no longer
3125 occasionally get blown up by Linux returning 1000000 microseconds
3126 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3127 its output conform to the ANSI spec. (Alas, this makes its output
3128 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3129 * A full call to MAP NIL with a single sequence argument no longer conses.
3130 * fixes to problems pointed out by Martin Atzmueller:
3131 * The manual page no longer talks about multiprocessing as though
3132 it were currently supported.
3133 * The ILISP support patches have been removed from the distribution,
3134 because as of version 5.10.1, ILISP now supports SBCL without us
3135 having to maintain patches.
3136 * added a modified version of Raymond Toy's recent CMU CL patch for
3137 EQUALP comparison of HASH-TABLE
3139 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3141 * The system still can't cross-compile itself with
3142 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3143 entails), but at least it can compile more of itself that way
3144 than it used to be able to, and various buglets which were uncovered
3145 by trying to cross-compile itself that way have now been fixed.
3146 * This release breaks binary compatibility again. This time
3147 at least I've incremented the FASL file format version to 2, so that the
3148 problem can be detected reliably instead of just causing weird errors.
3149 * various new style warnings:
3150 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3151 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3152 ** using the lexical binding of a variable named in the *FOO* style
3153 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3154 as specified by ANSI.
3155 * *RANDOM-STATE* is no longer automatically initialized from
3156 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3157 default behavior of the system is to repeat its behavior every time
3158 it's run. If you'd like to change this behavior, you can always
3159 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3160 old convention there was no comparably easy way to get the system to
3161 repeat its behavior every time it was run.
3162 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3163 FTYPE declarations has been removed, in favor of their ANSI
3164 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3165 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3166 the types of their sequence arguments can be determined at compile time.
3167 This is done through a new open code expansion for MAP which eliminates
3168 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3169 argument types can be determined at compile time.
3170 * The optimizer now transforms COERCE into an identity operation when it
3171 can prove that the coerced object is already of the correct type. (This
3172 can be a win for machine generated code, including the output of other
3173 optimization transforms, such as the MAP transform above.)
3174 * Credit information has been moved from source file headers into CREDITS.
3175 * Source file headers have been made more standard.
3176 * The CASE macro now compiles without complaining even when it has
3179 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3181 * (Note that the way that the PCL macroexpansions were rewritten
3182 to accommodate the change in DEFGENERIC below breaks binary
3183 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3184 not run under sbcl-0.6.2. Once we get out of alpha releases,
3185 i.e. hit release 1.0.0, we'll probably try to maintain binary
3186 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3187 and sbcl-1.4.4. Until then, however, it might be fairly common
3188 for maintenance releases to break binary compatibility.)
3189 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3191 * The reporting of unhandled signals has been changed to print some
3192 explanatory text as well as the report form. (Previously only
3193 the report form was printed.)
3194 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3195 it defines, so that the compiler no longer issues undefined function
3196 warnings for compiled-but-not-yet-loaded generic functions.
3197 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3198 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3199 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3200 section of the ANSI spec are supported.
3201 * The "" nickname for the "KEYWORD" package has been removed.
3202 The reader still handles symbol tokens which begin with a package marker
3203 as keywords, but it doesn't expose its mechanism for doing so in the
3204 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3205 * The system now issues STYLE-WARNINGs for contradictory TYPE
3206 proclamations. (Warnings for contradictory FTYPE proclamations would
3207 be nice too, but those can't be done usefully unless the type system
3208 is made smarter about FUNCTION types.)
3209 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3210 systematized, so that "*target-*.lisp is supposed to exist only on the
3211 target and imply that there's a related file which exists on the
3212 host, and *host-*.lisp is supposed to exist only on the host and imply
3213 that there's a related file which exists on the target. This involves a
3214 lot of renaming. Hopefully the acute confusion caused by the renaming
3215 will be justified by the reduction in chronic confusion..
3216 ** runtime-type.lisp -> early-target-type.lisp
3217 ** target-type.lisp -> late-target-type.lisp
3218 ** early-host-format.lisp -> early-format.lisp
3219 ** late-host-format.lisp -> late-format.lisp
3220 ** host-error.lisp -> misc-error.lisp
3221 ** early-error.lisp -> early-target-error.lisp
3222 ** late-error.lisp -> late-target-error.lisp
3223 ** host-defboot.lisp -> early-defboot.lisp
3224 ** code/misc.lisp -> code/target-misc.lisp
3225 ** code/host-misc.lisp -> code/misc.lisp
3226 ** code/numbers.lisp -> code/target-numbers.lisp
3227 ** code/early-numbers.lisp -> numbers.lisp
3228 ** early-host-type.lisp -> early-type.lisp
3229 ** late-host-type.lisp -> late-type.lisp
3230 ** host-typep.lisp -> typep.lisp
3231 ** load.lisp -> target-load.lisp
3232 ** host-load.lisp -> load.lisp
3233 ** host-disassem.lisp -> disassem.lisp
3234 ** host-insts.lisp -> insts.lisp
3235 ** byte-comp.lisp -> target-byte-comp.lisp
3236 ** host-byte-comp.lisp -> byte-comp.lisp
3237 ** host-signal.lisp -> signal.lisp
3238 ** host-defstruct.lisp -> defstruct.lisp
3239 ** late-target-type.lisp -> deftypes-for-target.lisp
3240 Furthermore, several other previously target-only files foo.lisp (e.g.
3241 hash-table.lisp and random.lisp) have been split into a target-and-host
3242 foo.lisp file and a target-only target-foo.lisp file, with their key type
3243 definitions in the target-and-host part, so that the cross-compiler will
3244 know more about target types.
3245 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3246 gone away. In their place are various *BACKEND-FOO* variables
3247 corresponding to the slots of the old structure.
3248 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3249 into the target SBCL has been fixed.
3250 * The chill.lisp system for loading cold code into a running SBCL
3252 * Support for the CMU CL "scavenger hook" extension has been removed.
3253 (It was undocumented and unused in the CMU CL sources that SBCL was
3254 derived from, and stale in sbcl-0.6.1.)
3255 * Various errors in the cross-compiler type system were detected
3256 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3257 (enabling various consistency checks). Many of them were fixed,
3258 but some hard problems remain, so the compiler is back to
3259 running without *TYPE-SYSTEM-INITIALIZED* for now.
3260 * As part of the cross-compiler type system cleanup, I implemented
3261 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3262 * I have started adding UNCROSS calls throughout the type system
3263 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3264 on cross-compiler input files will be able to go away, and various
3266 * CONSTANTP now returns true for quoted forms (as explicitly required
3269 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3271 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3272 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3273 (MAKE-LIST -1). In the longer term, it ought to become true
3274 that declarations are assertions even at SAFETY 1. For now, it's not
3275 quite true even at SAFETY 3, but it's at least more nearly true..
3276 (Note that this change seems to increases the size of the system by
3277 O(5%) and to decrease the speed of the compiler by 20% or more.)
3278 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3279 for the problem of printing dozens of lines of distracting information
3280 about low-level system machinery as part of the top stack frame
3281 on entry to the debugger when an undefined function was called.
3282 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3283 is not reset to COMMON-LISP-USER.
3284 * Compilation of stuff related to dyncount.lisp has been made conditional
3285 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3286 smaller. The various dyncount-related symbols have been moved into
3287 a new "SB-DYNCOUNT" package.
3288 * tty-inspect.lisp has been renamed to inspect.lisp.
3289 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3290 feature has gone away. (When we eventually port to other flavors of
3291 libc and/or Unix, we'll try to make the differences between flavors
3292 invisible at the user level.)
3293 * Various other *FEATURES* tags, and/or their associated conditionals,
3294 have been removed if obsolescent, or given better documentation, or
3295 sometimes given more-mnemonic names.
3297 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3299 * tidied up "make.sh" script
3300 * tidied up system directory structure
3301 * better "clean.sh" behavior
3302 * added doc/FOR-CMUCL-DEVELOPERS
3303 * many many small tweaks to output format, e.g. removing possibly-confusing
3304 trailing #\. character in DESCRIBE-INSTANCE
3305 * (EQUALP #\A 'A) no longer signals an error.
3306 * new hashing code, including EQUALP hashing
3307 * tidied up Lisp initialization and toplevel
3308 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3309 * command line argument processing
3310 * added POSIX-GETENV function to deal with Unix-ish environment variables
3311 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3312 terminating SBCL on EOF
3313 * non-verbose GC by default
3314 * There is no more "sbcl" shell script; the sbcl file is now the C
3315 runtime executable (just like CMU CL).
3316 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3318 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3319 symbol and package information as Lisp native symbol and package objects
3320 instead of strings naming symbols and strings naming packages. This way,
3321 whenever packages are renamed (as in warm init), debug information is
3322 transformed along with everything else.
3323 * tweaked the optimization policy declarations which control the building
3324 of SBCL itself. Now, among other things, the system no longer saves
3325 source location debugging information. (This helps two problems at once
3326 by reducing SBCL size and by keeping SBCL from trying to look for its
3327 sources -- which may not exist -- when reporting errors.)
3328 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3329 debugging and testing purposes
3330 * cleaned up printing, making the printer call PRINT-OBJECT for
3331 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3332 methods, giving nearly-ANSI behavior
3333 * converted almost all special variables to use *FOO* naming convention
3334 * deleted PARSE-TIME functionality, since it can be done portably
3335 * moved some files out of cold init into warm init
3336 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3338 * regularized formatting of source files
3339 * added an install.sh script
3340 * fixed ridiculous memory usage of cross-compiler by making
3341 compiler/alloc.lisp not try to do pooling unless it can hook
3342 itself into the GC of the cross-compilation host. Now the system
3343 builds nicely on my old laptop.
3344 * added :SB-ALLOC in target-features.lisp-expr
3345 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3346 was not implemented)
3347 * re-did condition handling and note reporting in the compiler. Notes
3348 are no longer handled by signalling conditions. Style warnings
3349 and warnings are handled more correctly and reported in such a way
3350 that it's easy to find one or the other in your output (so that you
3351 can e.g. figure out which of many problems caused COMPILE-FILE to
3353 * changed the severity of several compiler warnings from full WARNING
3354 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3355 compiler note reporting so that it doesn't use the condition system
3356 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3357 COMPILE-FILE command)
3358 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3359 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3360 no longer supported, since their implementation was deeply intertwingled
3361 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3362 * removed bogus "support" for compiler macros named (SETF FOO), and
3363 removed the compiler macro for SETF INFO (but only after making a fool
3364 of myself on the cmucl-imp mailing list by posting a bogus patch for
3365 DEFINE-COMPILER-MACRO..)
3366 * Compiled files containing forms which have side effects on the Lisp
3367 reader (such as DEFPACKAGE forms) are now handled more correctly.
3368 (Compiler queuing of top level lambdas has been suppressed by setting
3369 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3370 * deleted various currently-unused source files, e.g. gengc.lisp. They
3371 may be added back at some point e.g. when porting to other architectures,
3372 but until they are it's distracting to distribute them and to try to
3374 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3375 there were so many of them they're just distractions, and UNCROSS is
3376 known to be able to handle the current sources
3377 * moved PROFILE functionality into TRACE, so that it will be clear
3378 how the wrapping and unwrapping of functions when you profile them
3379 interacts with the wrapping and unwrapping of functions when you
3380 trace them. (Actually, the functionality isn't there yet, but at least
3381 the interface specification is there. Hopefully, the functionality will
3382 arrive with some future maintenance release.)
3383 * removed host-oops.lisp
3384 * changed signature of QUIT function to allow UNIX-CODE argument
3385 * fixed READ-SEQUENCE bug
3386 * tweaked verbose GC output so that it looks more like the progress
3387 output that ANSI specifies for functions like LOAD
3388 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3389 * added <http://sbcl.sourceforge.com> to the banner information printed by