1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
3 * bug fix: add a workaround to SBCL looping infinitely at startup on
4 Linux kernels with apparently buggy implementations of personality().
5 (thanks to Svein Ove Aas)
6 * bug fix: Unicode symbols are correctly printed in LDB backtraces
7 (thanks to David Lichteblau)
8 * bug fix: local bindings shadow global symbol macros for
9 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
10 * optimization: non-open coded uses of numeric comparison operators
11 (e.g. >) no longer cons when called with more than one parameter
12 on platforms supporting dynamic-extent allocation.
13 * enhancement: saving cores with foreign code loaded is now supported
14 on MIPS/Linux in addition to the previously supported platforms.
15 * bug fix: division by zero in sb-sprof when no samples were collected
16 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
17 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
18 (thanks to Alexander Kjeldaas)
19 * new feature: ignoring the return values of destructive functions
20 like NREVERSE signals a compile-time style-warning.
21 (thanks to Kevin Reid)
23 ** bug fix: threads stacks belonging to dead threads are freed by the
24 next exiting thread, no need to gc to collect thread stacks anymore
25 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
26 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
27 ** bug fix: dynamic variable and thread start related gc lossage
28 * fixed some bugs revealed by Paul Dietz' test suite:
29 ** SUBTYPEP is slightly more accurate on heinously complicated
30 CONS types where some of the members have uncertain (in the
31 NIL, NIL sense) type relationships to each other.
33 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
34 * new feature: timers based on Zach Beane's excellent timer package
35 * added support for the following external formats: koi8-u,
36 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
37 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
38 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
39 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
40 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
41 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
42 * incompatible change: a threaded SBCL will no longer revert to
43 non-threaded mode on non-NPTL systems, but refuse to start entirely.
44 * bug fix: interrupts are disabled until startup is complete; no
45 more sigsegvs when receiving a signal to soon
46 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
48 * bug fix: add a workaround for the memory randomization features in
49 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
50 workaround will only be in effect on systems with the proc filesystem
52 * bug fix: printing objects of type HASH-TABLE signals a
53 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
55 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
56 have been running for over 50 days. (reported by Gilbert Baumann)
57 * bug fix: the logic for getting names of functions gets less
58 confused when confronded with alternate-metaclass
59 funcallable-instances. (reported by Cyrus Harmon)
60 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
61 consistent, even on internal alternate-metaclass objects.
62 * bug fix: flush closure information collected by physical
63 environment analysis prepass before the main pass. (bug reported
65 * bug fix: compiler pack phase does not modify a hashtable, which is
66 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
67 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
68 may be expanded. (reported by Alexander <malishev> on c.l.l)
69 * bug fix: no more style-warnings when compiling code using LAST
70 under high-SPEED low-SPACE settings. (reported by David Wallin
72 * fixed some incompatibilities between SBCL's MOP and the MOP
74 ** the METAOBJECT class is now implemented;
75 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
76 STANDARD-OBJECT, as required;
77 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
78 now compatible (as required by VALIDATE-SUPERCLASS); there
79 remains a consistency requirement over the presence or absence
80 of the FUNCTION class in the superclass list of finalized
81 classes; see the manual for more details;
82 ** the :AROUND method for COMPUTE-SLOTS on
83 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
84 requested slot ordering.
86 ** bug fix: parent thread now can be gc'ed even with a live
88 ** bug fix: binding a special with PROGV to no value is not
89 the same as that symbol not having been bound (thanks to
91 ** bug fix: binding specials is thread safe (thanks to
93 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
94 and SB-SPROF do not die with 'no handler for signal XX in
95 interrupt_handle_now(..)' anymore
96 ** bug fix: WITH-TIMEOUT works with multiple running threads
97 * fixed some bugs revealed by Paul Dietz' test suite:
98 ** ENSURE-GENERIC-FUNCTION should take a method class object for
99 the :method-class keyword argument.
101 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
102 * new port: the Solaris operating system on x86 processors is now
103 mostly supported, though some rough edges in the environment
104 remain. (thanks to Daisuke Homma)
105 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
106 space for its heap. (thanks to Thiemo Seufer)
107 * optimization: SBCL on MIPS platforms now supports dynamic-extent
108 closures. (thanks to Thiemo Seufer)
109 * minor incompatible change: eof selects abort in the debugger.
110 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
111 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
113 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
114 deprecated, and will go away in a future revision of SBCL.
115 * minor incompatible change: GC-ON and GC-OFF are no longer
116 implemented with a counter, it does not matter how many times gc
117 is switched on or off
118 * bug fix: discriminating functions for generic function classes
119 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
120 make invalid assumptions about method precedence order. (reported
122 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
123 platforms. (thanks to Thiemo Seufer)
124 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
125 * bug fix: correct pathname printing with printer escaping is on.
126 (thanks to Kevin Reid)
127 * bug fix: complex VOP definitions in "user-space" no longer trigger
128 package locks. (reported by Zach Beane)
129 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
130 for funcallable-instance functions, and is no different from
132 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
133 characters in the current readtable. (reported by Nicholas Neuss)
134 * bug fix: SUBTYPEP on various CONS types returns more a more
135 accurate acknowledgment of its certainty.
136 * optimizations: REMOVE-DUPLICATES now runs in linear time on
137 lists in some cases. This partially fixes bug 384.
138 * flush all standard streams before prompting in the REPL and the
140 * bug fix: signal handling and triggering gc do not conflict
141 directly anymore, in particular a high frequency sb-sprof does
142 not prevent gc from running
143 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
144 approximation for timezone and DST information between the
145 universal time epoch and the smallest negative 32-bit time_t.
146 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
147 year 1899 so long as the encoded time is non-negative.
148 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
149 the correct answer for non-constant rotate amounts when performing
150 an inline 32-bit rotation.
152 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
153 there is only one thread in the session
154 ** bug fix: memory leak for streams created in one thread and
155 written to in another
156 ** bug fix: lockup when compiled with gcc4
157 ** bug fix: race that allows the gc to be triggered when gc is
159 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
160 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
162 ** bug fix: finished threads are gc'ed properly
163 * fixed some bugs revealed by Paul Dietz' test suite:
164 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
165 has arguments with a different set of applicable methods from
166 the orignal arguments.
167 ** The type error thrown by MAP now has a correct expected-type
169 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
170 name a compiled function.
171 ** Three MISC tests where a large bignum was improperly coerced to
172 a float that couldn't represent that bignum during type
173 derivation were fixed.
174 ** SUBTYPEP can now handle types involving the negation of a
175 list-form FUNCTION type.
176 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
177 as (COMPLEX (AND RATIO (NOT FIXNUM))).
178 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
180 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
181 * New feature: Experimental support for bivalent streams: streams
182 opened with :element-type :default now allow character and binary
183 (unsigned-byte 8) I/O
184 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
185 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
187 * bug fix: no more highly sporadic "couldn't check whether ~S is
188 readable" when reading a stream and an interrupt hits in the middle
189 of a select system call
190 * compiler better recognizes complex arrays (reported by Eduardo
192 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
193 for variables on SBCL built with linkage-tables. (reported by Luis
195 * various error reporting improvements.
196 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
197 (Thanks to Thiemo Seufer)
198 * bug fix: nested reader invokations work correctly
199 * bug fix: it is possible to have simultaneous references to foreign
200 code and foreign data with the same name.
202 ** added x86-64 support
203 ** incompatible change: the threading api now works with thread
204 objects instead of thread ids
205 ** bug fix: threads are protected from signals and interruption when
206 starting up or going down
207 ** bug fix: a race where an exiting thread could lose its stack to gc
208 ** bug fix: don't halt on infinite error in threads if possible
209 ** fixed numerous gc deadlocks introduced in the pthread merge
210 ** bug fix: fixed thread safety issues in read and print
211 ** bug fix: debugger doesn't hang on session lock if interrupted at
212 an inappropriate moment
213 ** bug fix: run-program is now thread safe(r)
214 ** bug fix: inner with-recursive-lock no longer releases the mutex
215 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
216 * fixed some bugs revealed by Paul Dietz' test suite:
217 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
218 EXPECTED-TYPE slots filled.
219 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
221 ** MAKE-CONDITION accepts classes as type-designators.
222 ** COMPILE may never return NIL.
223 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
224 range before calling Unix time functions
226 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
227 * numerous signal handling fixes to increase stability
228 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
229 * minor incompatible change: we now correctly canonize default
230 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
231 as per the MOP, rather than the historical (INITARG INITFUNCTION
232 INITFORM). (reported by Bruno Haible)
233 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
234 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
235 for more information.
236 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
237 pathname is a directory pathname.
238 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
239 * dynamic space size on PPC has been increased to 768Mb. (thanks to
241 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
242 :METACLASS argument in addition to a class name. (reported by
243 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
244 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
245 (reported by Bruno Haible; patch from Gerd Moellmann)
246 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
248 * bug fix: more cleanups to the floating point exception handling on
249 x86-64 (thanks to James Knight)
250 * bug fix: the compiler does not try to constant fold calls of
251 COERCE and BIT-* functions when they return freshly constructed
252 objects. (reported by Nikodemus Siivola and Paul Dietz)
253 * optimization: improved the MIPS versions of generic (in the
254 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
255 * optimization: direct conversion of (unsigned-byte 32) to floats on
256 the PowerPC platform.
257 * optimization: structure instances with raw slots now use less
258 memory, and probably show better memory locality. (thanks to
260 * optimization: DYNAMIC-EXTENT declarations for lists and closures
261 are treated as requests for stack allocation on the x86-64,
262 Alpha-32, PPC and SPARC platforms.
263 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
264 the wrong file descriptor; implementation of SOCKET-OPEN-P.
265 (thanks to Tony Martinez)
267 ** gcing a dead thread can no longer lead to lockups
268 ** threads block signals until they are set up properly
269 ** errno is no longer shared by threads
270 ** interrupt-thread restores the eflags register on x86
271 ** fixed some lockups due to gc/thread interaction
272 * fixed some bugs revealed by Paul Dietz' test suite:
273 ** invalid dotted lists no longer raise a read error when
275 ** signal an error if a symbol that names a declaration is used
276 as the name of a type, or vice versa
277 ** allow using the (declare (typespec var*)) abbreviation for
278 (declare (type typespec var*)) with all type specifiers
279 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
280 ** always return NIL from PPRINT-POP when OBJECT is NIL
281 ** don't signal errors when pretty-printing malformed LABELS,
282 FLET or MACROLET forms
283 ** declarations in a DOLIST body are also in scope for the
285 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
287 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
290 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
291 * fixed cross-compiler leakages that prevented building a 32-bit
292 target with a 64-bit host compiler.
293 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
294 opened with :IF-EXISTS :OVERWRITE.
295 * fixed bug 281: error for an invalid qualifier in a short-form method
296 combination method is not signalled until the faulty method is called.
297 * bug fix: iteration variable type inferrer failed to deal with open
298 intervals. (reported by Alan Shields)
299 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
300 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
301 or more runtime options were provided to the sbcl binary.
302 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
303 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
304 * optimization: there's now a fast-path for fixnum arguments in the
305 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
306 * optimization: the code generated on x86-64 is more compact thanks
307 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
308 * bug fix: floating-point exception handling now partly works on
309 x86-64. (Thanks to James Knight)
310 * improvement to the MIPS backend from Thiemo Seufer: C-style
311 64-bit long long arguments and return values to alien functions
313 * fixed some bugs revealed by Paul Dietz' test suite:
314 ** the type-error signalled from WARN has a filled-in DATUM slot.
315 ** the type-error required when a stream is not associated with
316 a file has the stream as its datum.
317 ** type-errors on single-floats on x86-64 no longer have
318 :INVALID-OBJECT as the datum
319 ** the type-errors signalled for invalid function names now have
320 a correct expected type
321 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
322 for structures defined with a :TYPE. Documentation strings for
323 typed structures are no longer immediately discarded
324 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
325 broadcast streams and synonym streams. FILE-LENGTH now also works
326 on broadcast streams.
328 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
329 * incompatible change: the --noprogrammer option, deprecated since
330 version 0.7.5, has been removed. Please use the equivalent
331 --disable-debugger option instead.
332 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
334 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
335 inoperational for a while now) has been completely removed.
336 * null lexical environments are now printed as #<NULL-LEXENV>,
337 significantly reducing the amount of clutter in typical backtraces.
338 * documentation on weak pointers, finalization, and after GC hooks
339 has been added to the manual.
340 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
341 as other specialized array types, is much faster. SUBSEQ and
342 COPY-SEQ on such arrays have also been sped up.
343 * optimization: EQL is now more efficient when at least other argument
344 is known to be of type (OR FIXNUM (NOT NUMBER)).
345 * fixed bug: the runtime is now less vulnerable to changes in the
346 size of the SBCL object on OS X, and virtual memory is reserved for
347 all spaces that need to be at a fixed address.
348 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
349 * fixed bug: finalizers and after GC hooks that cause consing are now
351 * fixed bug: compiler error messages and summaries are now printed to
352 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
353 * fixed inference of the upper bound of an iteration variable.
354 (reported by Rajat Datta).
355 * fixed bug 211e: calling local functions with duplicated constant
356 keyword argument no longer causes a bogus style warning about an
358 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
359 ftype declaration to be disregarded. (reported by Dave Roberts)
360 * fixed bug 373: caused by erronous compilation of references to alien
361 variables in the runtime on ppc/darwin.
362 * fixed bug 376: CONJUGATE type deriver.
363 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
364 (reported by Baughn on #lisp)
365 * a cross-compiler bug on non-x86 platforms has been identified and
366 fixed. (thanks to Bruno Haible)
367 * improvements to the MIPS runtime code for increased stability.
368 (thanks to Thiemo Seufer)
369 * increased the maximimum compact environment size to allow
370 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
371 * partial workaround for bug 135: don't name the function we're
372 calling for hairy cases of EVAL, so as not to accumulate one environment
373 entry per call to EVAL. (reported by Kevin Reid)
374 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
375 * optimization: type testing for non-vector arrays should be faster.
376 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
378 * bug fix: setting 31st element of a bit vector to zero did not work
380 * bug fix: redefining a class definition which failed due to a
381 previous accessor / function clash now works (but see BUGS entry
382 #380 for more problems in this area). (thanks to Zach Beane)
383 * the long form of DEFINE-METHOD-COMBINATION disables method group
384 checking when given a single method group with pattern *. (thanks
386 * on x86 compiler supports stack allocation of results of simple
387 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
388 * support for building and running on Mac OS X 10.4 "Tiger" has been added
389 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
390 * fixed some bugs related to Unicode integration:
391 ** the restarts for recovering from input and output encoding
392 errors only appear when there is in fact such an error to
394 * fixed some bugs revealed by Paul Dietz' test suite:
395 ** MISC.549 and similar: late transformation of unsafe type
396 assertions into derived types caused unexpected code
398 ** SCALE-FLOAT type deriver is less wrong.
399 ** type derivers for EXP, LOG and similar functions compute result
400 types for complex arguments better.
401 ** (MISC.563) CONJUGATE type deriver works for very restricted
403 ** out-of-line type testers for character strings are available.
404 ** EQUAL compiler transform understands specialness of objects
406 ** accessing double-floats stored on the stack now works on x86-64.
407 ** debugger internals could sometimes create invalid lispobjs,
408 resulting in GC crashes.
409 ** MISC.548: type check weakening can convert required type into
411 ** initialization forms for bindings are not in scope of free special
414 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
415 * incompatible change: thread support for non-NPTL systems has
416 been removed - locking is buggy and unreliable. A threaded
417 SBCL build will now warn at startup and refuse to create
418 new threads, unless futex support is detected in the kernel
419 * incompatible change: the top level REPL now has only an ABORT
420 restart associated with it, not TOPLEVEL and ABORT as it used to.
421 TOP and TOPLEVEL are now available as debugger commands for
422 returning to the top level.
423 * incompatible change: forms evaluated in the REPL now use the
424 global optimization policy.
425 * incompatible change: user- and system-initialization files are
426 no longer processed with LOAD, but by READ and EVAL; hence the
427 global optimization policy, startup package, readtable, etc,
429 * The .fasl file format number has been incremented because of
430 various incompatible changes.
431 * internal entry point details and argument counts no longer appear
432 in backtraces unless explicitly requested by setting
433 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
434 * built-in and standard functions no longer have names like "top
435 level local call to FOO".
436 * fixed bug 32: functions defined in non-null lexical environments
437 now have more legible printed representation
438 * fixed bug 33: functions defined in non-null lexical environemnts
439 are now more amenable to inspection by INSPECT.
440 * workaround for bug 354: XEPs no longer appear in backtraces unless
441 explicitly requested.
442 * fixed bug: receiving the signal which results from attempting to
443 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
444 SIGBUS on other *BSDs) on architectures where the C stack is also
445 the Lisp stack (x86 and x86-64) from foreign code no longer leads
446 to debugger confusion or wild instability. (reported by Cheuksan
448 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
449 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
450 (reported by Lutz Euler)
451 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
452 compiled code. (reported by Paul Dietz)
453 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
454 specialized parameter to a method to something that is not TYPEP
455 the specializer is now possible.
456 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
457 face of package deletion.
458 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
459 pathname. (reported by David Tolpin on c.l.l)
460 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
461 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
462 STRING-STREAM and FILE-STREAM, respectively.
463 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
464 than previously; in addition, its test suite is now run on build.
465 (thanks to Luke Gorrie)
466 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
467 * optimization: added a immediate representation for single-floats
469 * fixed some bugs revealed by Paul Dietz' test suite:
470 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
471 ** attempting to create a package with a colliding nickname causes
472 correctable errors to be signalled.
473 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
474 ** several x86-64 backend bugs related to sign-extension of immediate
477 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
478 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
479 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
481 * fixed loading of multiply forward-referenced layouts.
482 (thanks to Cheuksan Wang)
483 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
484 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
485 (see COMPILE-FILE documentation for details of :PRINT option)
486 * fixed bugs 19 and 317: fixed-format floating point printing is
487 more accurate. This also fixes a bug reported by Adam Warner
488 related to the ~@F format directive.
489 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
491 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
492 dispatch table if the from-char is a dispatch macro character.
493 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
494 by the wordsize no longer ignore the last word. (reported by Lutz
496 * fixed bug in type checking of dynamic-extent variables. (reported
498 * optimization: sequence traversal functions use their freedom to
499 coerce function designators to functions.
500 * optimization: code with many calls to CLOS methods specialized on
501 CLOS classes has had redundant type checks removed.
502 * optimization: added declarations to speed up operations that access
503 the internal character database (for example STRING-UPCASE)
504 * optimization: comparison operations between floats and sufficiently small
505 fixnums no longer create extra rationals
506 * fixed some bugs related to Unicode integration:
507 ** portions of multibyte characters at the end of buffers for
508 character-based file input are correctly transferred to the
509 start of the buffer at the next read.
510 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
511 passing it through to OPEN.
512 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
513 argument given to internal calls to OPEN.
514 ** the built-in comment readers (introduced by character sequences
515 ";" and "#|") are more forgiving to encoding errors; they will
516 STYLE-WARN and then attempt to resync the stream at a character
517 boundary. (thanks to Teemu Kalvas)
518 * fixed some bugs in the x86-64 port:
519 ** Negative short int return values from c-calls are sign-extended
521 ** The stack is aligned to 16-bytes for c-calls, as required by
522 the ABI. (thanks to Cheuksan Wang)
523 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
524 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
525 for enums. (thanks to Vincent Arkesteijn)
526 ** Multiple small optimizations and bugfixes for floating point
528 * fixed some bugs revealed by Paul Dietz' test suite:
529 ** Space, Tab, Linefeed, Return and Page have the invalid
530 secondary constituent character trait.
531 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
533 ** WITH-INPUT-FROM-STRING should only update the index place on
535 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
537 ** Bit-array operations (BIT-AND and similar) worked incorrectly
538 with one-dimensional arrays with fill pointers.
539 ** TYPE-OF failed on a complex with an integer realpart and a
541 ** compiler failure during type inference for the code of form
542 (IF (EQL X (THE ...)) ...) (MISC.535).
544 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
545 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
546 platform. The port supports 61-bit fixnums, large memory spaces
547 and reloading shared object files.
548 * enhancement: saving cores with foreign code loaded is now
549 supported on ppc/Darwin in addition to the previously supported
551 * enhancement: the statistical profiler now walks deeper into the
552 call stack for more meaningful call-graphs and accrued time
553 reports (x86/x86-64 only). It also now reports time spent in
555 * enhancement: it is now possible to trace most individual methods
556 of a generic function in addition to tracing the generic function
558 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
560 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
561 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
562 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
563 produces an error. (thanks to Vincent Arkesteijn)
564 * bug fix: NAMESTRING on pathnames with :WILD components in their
565 directories works correctly. (thanks to Artem V. Andreev)
566 * fixed bug 125: compiler preserves identity of closures. (reported
568 * bug fixed: functions with &REST arguments sometimes failed with
569 "Undefined function" when compiled with (DEBUG 3). (reported by
571 * bug fix: overflow during compiling of setting element of a bit
572 vector with constant index and value. (reported by Timmy Douglas)
573 * build fix: fixed the dependence on *LOAD-PATHNAME* and
574 *COMPILE-FILE-PATHNAME* being absolute pathnames.
575 * on x86 compiler partially supports stack allocation of dynamic-extent
577 * GO and RETURN-FROM do not check the extent of their exit points
578 when compiled with SAFETY 0.
579 * fixed some bugs related to Unicode integration:
580 ** encoding and decoding errors are now much more robustly
581 handled; it should now be possible to recover even from invalid
582 input or output to the terminal. (thanks to Teemu Kalvas)
583 ** provided a first cut at implementing STRING-TO-OCTETS and
584 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
585 ** altered the SB-MD5 contributed module slightly, changing the
586 interface just enough for it to be supportable for builds where
587 lisp characters are not eight bits.
588 * fixed some bugs revealed by Paul Dietz' test suite:
589 ** the FORMATTER-generated functions for ~V[ conditionals require
590 the correct number of arguments.
591 ** READ-FROM-STRING returns the mandated second value when applied
592 to displaced strings.
593 ** the #\Rubout and #\Backspace characters are treated as invalid
594 constituent characters by the tokenizer.
596 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
597 * new feature: reloading changed shared object files with
598 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
599 * new feature: references to foreign variables and functions
600 can now be compiled and loaded before the corresponding shared
601 object file is loaded, as long as the foreign definitions are
602 available at runtime.
603 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
604 supported. (thanks to Dan Debertin)
605 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
606 just keywords. (thanks to Vincent Arkesteijn)
607 * fixed bug #331: structure-class instances corresponding to
608 DEFSTRUCT forms are now created eagerly.
609 * fixed bug #345: backtraces from calls to undefined functions work
610 on x86 as well. Related bug #61 is now also partially fixed on x86
611 (backtraces from throws to unknown catch tags.)
612 * bug fix: lambda-list parsing is now stricter vrt. order and number
613 of lambda-list keywords.
614 * bug fix: as specified by AMOP, an error is signalled if a
615 class-option appears multiple times in a DEFCLASS form. (reported
617 * bug fix: RANDOM can be compiled when the compiler derives the type
618 of its numeric argument as a disjoint set of small integers.
619 (reported by Paul Dietz)
620 * bug fix: starting a core saved with shared objects loaded when
621 those objects are not available no longer causes threaded SBCL to
622 hang. (reported by Sean Ross)
623 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
625 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
626 argument, which must always be NIL. (reported by Kalle Niemitalo)
627 * bug fix: printing 1.0d+23 no longer results in an error.
628 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
629 * bug fix: structure slot setters preserve evaluation order. (thanks
631 * fixed some bugs related to Unicode integration:
632 ** RUN-PROGRAM can allow its child to take input from a Lisp
633 stream. (reported by Stefan Scholl)
634 ** ASDF-INSTALL successfully downloads .tar.gz files in all
635 locales. (reported by Ken Causey)
636 * fixed some bugs revealed by Paul Dietz' test suite:
637 ** INCF, DECF and REMF evaluate their place form as specified in
639 ** FORMATTER expands ~{ ~} iteration directives with V or #
640 parameters correctly.
641 ** FORMATTER deals with the ~@[ ~] conditional directive where the
642 consequent uses no arguments correctly.
643 ** the system has a partial understanding of the (COMPLEX RATIO)
645 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
648 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
649 * new feature: a build-time option (controlled by the :SB-UNICODE
650 keyword feature, enabled by default) for building the system with
651 support for the entire 21-bit character space defined by the
653 * new feature: the system now has rudimentary external-format
654 support; the primary user-visible change at this time is that
655 characters with the high bit set (such as lower-case-e-acute) will
656 print correctly to a terminal in a UTF-8 environment.
657 * minor incompatible change: BASE-CHAR no longer names a class;
658 however, CHARACTER continues to do so, as required by ANSI.
659 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
660 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
661 has been moved to the SB-EXT package (temporarily re-exported from
663 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
664 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
666 * fixed bug #351: better error-handlind and reporting for malformed
668 * fixed bug #350: bignum-printing is now more memory-efficient,
669 allowing printing of very large bignums, eg. (expt 2 10000000).
670 (reported by Bruno Haible)
671 * fixed bug #302: better primitive-type selection for intersection
673 * fixed bug #308: non-graphic characters now all have names, as
674 required. (reported by Bruno Haible)
675 * bug fix: redefining a class with different superclasses now correctly
676 removes it from the direct-subclasses of its previous superclasses.
677 (reported by David Morse)
678 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
679 new value now works. (reported by Bruno Haible)
680 * bug fix: correct canonicalization of multiple non-standard slot
681 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
682 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
683 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
684 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
685 now exists, an signals an error.
686 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
687 during macroexpansion for non-iteration variables as well. (reported
688 by Bruno Haible for CMUCL)
689 * bug fix: Cyclic structures and unprintable objects in compiler
690 messages no longer cause errors. (reported by Bruno Haible)
691 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
692 and READ-FROM-STRING all now return a primary value of NIL if
693 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
694 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
695 (reported by Bruno Haible for CMUCL)
696 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
697 do not designate a condition. (reported by Bruno Haible for
699 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
700 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
701 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
702 permitting resolution in favour of any of the conflicting symbols.
703 (reported by Bruno Haible for CMUCL)
704 * bug fix: EQUAL compiler optimizations is less aggressive on
705 strings which can potentially compare true despite having distinct
706 specialized array element types.
707 * bug fix: unit enumerations can be defined without dividing by
708 zero. (thanks to Vincent Arkesteijn)
709 * FORMAT compile-time argument count checking has been enhanced.
710 (report from Bruno Haible for CMUCL)
711 * a partial workaround for the bug 262: the compiler does not try to
712 inline-expand a local function doing RETURN-FROM from a deleted
713 BLOCK. (thanks to Peter Denno for the bug report and to David
714 Wragg for the simple test case)
715 * fixed some bugs revealed by Paul Dietz' test suite:
716 ** RENAME-PACKAGE allows all package designators as new package
718 ** constraint propagation and lambda variable substitution are
719 more cautious in dealing with partially deleted code.
720 ** compiler sometimes lost reoptimization passes.
721 ** CERROR, given a condition as condition designator, passes its
722 remaining arguments to the continue format control without
724 ** Case-altering FORMAT directives work correctly on non-ASCII
726 ** The REAL type specifier handles bounds outside the floating
727 point ranges without signalling FLOATING-POINT-OVERFLOW.
728 ** Functions with IR1-transformations can create intercomponent
729 references to global functions.
730 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
732 ** FORMAT ~R treats a nil value for its first parameter correctly.
734 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
735 * enhancement: saving cores with foreign code loaded is now
736 supported on x86/NetBSD and sparc/Linux in addition to the previously
738 * bug fix: on some platforms repeated installations caused multiple
739 copies of HTML documentation to be installed -- should not happen
740 any more. (reported by Stefan Scholl)
741 * bug fix: parsing self-recursive alien record types multiple times
742 no longer causes infinite recursion. (reported by Thomas F. Burdick,
743 original patch by Helmut Eller for CMUCL)
744 * bug fix: stack-exhaustion detection works now on NetBSD as well.
745 (thanks to Richard Kreuter)
746 * bug fix: defining classes whose accessors are methods on existing
747 generic functions in other (locked) packages no longer signals
748 bogus package lock violations. (reported by François-René Rideau)
749 * bug fix: special variables as DEFMETHOD parameters no longer have
750 associated bogus type declarations. (reported by David Wragg and
752 * bug fix: read-write consistency on streams of element-type
753 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
754 * bug fix: redefiniton of the only method of a generic function with
755 no DEFGENERIC no longer emits a full WARNING. In addition,
756 redefinition of generic functions with no DEFGENERIC to an
757 incompatible lambda list now signals an error. (thanks to Zach
759 * bug fix: DEFGENERIC now works even when there's a function of the
760 same name in an enclosing lexical environment. (thanks to Zach
762 * fixed compiler failure, caused by instrumenting code during
763 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
764 * optimization: added loop analysis and improved register allocation
765 * fixed some bugs revealed by Paul Dietz' test suite:
766 ** POSITION on displaced vectors with non-zero displacement
767 returns the right answer.
768 ** (SIMPLE-STRING) is a valid type specifier for sequence
770 ** *PRINT-LEVEL* handling for slotless structures is pedantically
772 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
773 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
775 ** The justification version of the FORMAT ~< directive treats
776 non-zero minpad parameter correctly.
778 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
779 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
780 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
781 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
782 the supported interface.
783 * new feature: Single-stepping of code compiled with DEBUG 2 or
784 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
785 * new feature: saving cores with foreign code loaded is now
786 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
787 Timothy Moore's work for CMUCL)
788 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
789 parameters to * instead of NIL if no initform is supplied.
790 (reported by Johan Bockgaard)
791 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
792 correctly when FUNCALL appears as the car of the form. Note:
793 despite this FUNCALL forms are not currently subject to
794 compiler-macro expansion. (port of Raymond Toy's fix for the
795 same from CMUCL, reported by Johan Bockgaard)
796 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
797 (thanks for Teemu Kalvas)
798 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
799 case instead of terminating the iteration (thanks for Julian
800 Squires, Sean Champ and Raymond Toy)
801 * bug fix: incorrect expansion of defgeneric that caused a style
802 warning. (thanks for Zach Beane)
803 * bug fix: slot accessor effective method computation works properly
804 for classes with multiple non-standard applicable methods on
805 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
806 * on x86 compiler supports stack allocation of results of LIST and
807 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
808 CMUCL implementation by Gerd Moellmann)
809 * fixed some bugs revealed by Paul Dietz' test suite:
810 ** FORMAT strings with both the ~<~:;~> form of the justification
811 directive and pretty-printing directives cause an error.
813 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
814 * incompatible change: the internal functions
815 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
816 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
817 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
818 instead of the old functions.
819 * new feature: on platforms where "dladdr" is available foreign
820 function names now appear in backtraces. (based on Helmut Eller's
822 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
823 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
825 * documentation: ASDF manual is now distributed with SBCL.
826 * bug fix: SBCL can now load its contributed modules with REQUIRE
827 even if the system-provided entries have been removed from
828 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
830 * bug fix: backtraces involving undefined functions or assembly
831 routines are more informative. (thanks to Brian Downing)
832 * bug fix: mutually referent alien structures now work correctly.
833 (reported by Rick Taube)
834 * bug fix: structures defined by WITH-ALIEN can be referred to
835 within other definitions in the same WITH-ALIEN.
836 * bug fix: division operators (MOD, TRUNCATE and the like) with
837 constant zero divisors and integer dividends no longer generate
839 * bug fix: provide default methods for INPUT-STREAM-P and
840 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
841 * bug fix: improve the detection and resolution of MOP metacycles
842 (where effective method computation depends on the generic function
843 having its effective method computed). (reported by Bruno Haible)
844 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
845 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
846 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
847 * bug fix: pathnames with relative directory components can be
848 represented relative to default pathnames.
849 * optimization: in taking the GCD of bignums, reduce the two bignums
850 to approximately the same size (using Euclid's algorithm) before
851 applying the more sophisticated binary GCD. (thanks to Juho
853 * optimization: COUNT on bitvectors now operates word-at-a-time.
854 * optimization: ASH with a positive, but not necessarily constant,
855 (leftwards) shift, when in a modular context, is compiled to a
857 * fixed some bugs revealed by Paul Dietz' test suite:
858 ** FORMAT variable parameters ("~V<char>") are defaulted properly
859 if the corresponding argument is NIL.
860 ** FORMAT directives accepting a minpad parameter treat negative
862 ** literal commas as character parameters to FORMAT directives are
864 ** literal spaces directly after ~<Newline> directives within a
865 format-logical-block (~:< ~@:>) do not induce :FILL-style
866 conditional newlines.
867 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
868 as specified: it no longer includes conditional newlines).
869 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
871 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
872 :PER-LINE-PREFIX argument does not evaluate to a string.
874 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
875 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
876 the manual for details; for now, package locks can be disabled by
877 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
878 if no major problems are found then it is likely that they will be
879 compiled in unconditionally.
880 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
881 unsupported operators on all platforms. To load a shared library
882 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
883 object file, link it into a shared library outside of SBCL and
884 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
886 * bug workaround: ROOM T and the GC were not working together
887 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
888 reporting facility aren't preserved. That reporting has been
889 disabled, so now until and unless someone figures out how to make it
890 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
891 * minor incompatible change: as threatened around sbcl-0.8.0, the
892 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
893 The symbols are also exported from SB-PCL for backwards
894 compatibility, but more so than before SB-PCL should be treated as
895 an implementation-internal package.
896 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
898 * the SB-SPROF contrib now works on (most) non-x86 architectures.
899 It is known as of this release not to work on the Alpha, however.
900 * fixed bug #167: errors signalled due to illegal syntax in method
901 bodies are now more legible.
902 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
903 designators and can hence be used with TYPEP.
904 * fixed bug #333: CHECK-TYPE now ensures that the type error
905 signalled, if any, has the right object to be accessed by
906 TYPE-ERROR-DATUM. (reported by Tony Martinez)
907 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
908 5.1.2.3 for multiple-value place subforms. (reported by Kalle
910 * fixed bug #334: programmatic addition of slots using specialized
911 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
912 and :ALLOCATION :CLASS slots.
913 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
914 float exponents. (rereported by Peter Seibel)
915 * fixed a bug: #\Space (and other whitespace characters) are no
916 longer considered to be macro characters in standard syntax by
918 * fixed bug: initialization of condition class metaobjects no longer
919 causes an instance of the condition to be created. (reported by
921 * fixed bug: it is now possible to have slots such that
922 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
923 description is neither :INSTANCE nor :CLASS.
924 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
925 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
926 SLOT-VALUE-USING-CLASS), and no longer causes errors with
927 non-standard SLOT-DEFINITION-ALLOCATION values.
928 * fixed bugs: various race conditions handling exiting threads.
929 CL-PPCRE's thread torture test now passes.
930 * fixed arguably-a-bug: GC time proportional to number of threads in
931 system even when most of them are idle
932 * optimization: improved performance of BIT and SBIT on bit-vectors.
933 * .fasl file incompatibility: The fasl file version number has
934 been incremented because of changes associated with package locks.
936 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
937 * minor incompatible change: the system no longer provides
938 optimization hints (conditions which are TYPEP
939 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
940 mode; these hints are still emitted when higher SPEED optimization
942 * new contrib module: a sampling profiler (profiling by statistical
943 sampling, rather than by instrumenting functions) is available as
944 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
945 * the behaviour of the standard function ED is now customizeable by
946 third parties through a hook variable: see ED's documentation
947 string for information on the protocol.
948 * the compiler no longer emits efficiency notes for (FUNCALL X)
949 when the type of X is uncertain under default optimization
951 * fixed bug 276: mutating a binding of a specialized parameter to a
952 method to something that is not TYPEP the specializer is now
954 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
955 exhibit the required behaviour on the x86 platform. (thanks to
956 Peter van Eynde, Eric Marsden and Bruno Haible)
957 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
958 even for difficult arguments. (reported by Peter Graves)
959 * fixed bug 141a: the backquote printer now descends quoted
961 * fixed another bug in backquote printing: no more destructive
962 modification of the form's list structure. (reported by Brian
964 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
965 move between its address being taken and the call to
966 interrupt_thread, fixing a crashing race condition.
967 * the SB-POSIX contrib implementation has been adjusted so that it
968 no longer exhibits ridiculously poor performance when constructing
969 instances corresponding to C structs.
971 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
972 * minor incompatible change: the sb-grovel contrib now treats C
973 structures as alien (in the sense of SB-ALIEN) objects rather than
974 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
975 has implications for memory management of client code
976 (sb-grovel-returned objects must now be manually managed) and for
977 type safety (alien objects now have full types).
978 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
979 used to control emission of compiler diagnostics, rather than the
980 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
981 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
982 quality should be considered deprecated.
983 * install.sh now installs the user manual as well
984 * (not quite a new documentable feature, but worth considering in
985 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
986 beginnings of a semantically meaningful condition hierarchy is
987 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
989 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
990 designator as the defaults argument.
991 * fixed bug: Displaced arrays whose displaced-to array has become
992 too small now cause ARRAY-DIMENSION to signal an error, providing
993 for safer bounds-checking. (reported by Bruno Haible)
994 * fixed bug: DEFCLASS slot definitions with identical :READER and
995 :WRITER names now signal a reasonable error. (reported by Thomas
997 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
999 * fixed bug: Invalid slot specification errors now print correctly.
1000 (thanks to Zach Beane)
1001 * fixed bug 320: Shared to local slot value transfers in class
1002 redefinitions now happen corrently from superclasses as
1003 well. (reported by Bruno Haible)
1004 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1006 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1007 improper lists correctly. (reported by Bruno Haible)
1008 * fixed bug 313: source-transform for <fun-name> was erroneously
1009 applied to a call of a value of a variable with name <fun-name>.
1010 (reported by Antonio Menezes Leitao)
1011 * fixed bug 307: The obsolete instance protocol ensures that
1012 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1013 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1014 remove dead unknown-values globs from the middle of the stack.
1015 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1016 easier to resume long computations after using *BREAK-ON-SIGNALS*
1017 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1018 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1019 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1020 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1021 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1022 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1024 * fixed a (fairly theoretical) bug in string printing: if
1025 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1026 does not have array-element-type equal to the most general string
1028 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1029 function name. (thanks to Nikodemus Siivola)
1030 * fixed bug: compile-time format string checker failed on
1031 non-closed ~{. (reported by Thomas F Burdick)
1032 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1033 don't warn on legal loop code involving more than one aggregate
1034 boolean. (thanks to Nikodemus Siivola)
1035 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1036 in keyword position to MAKE-PACKAGE was not being reported as
1038 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1039 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1041 * fixed bugs 280 and 312: the checking for multiple definitions in a
1042 file is less likely to become confused by uses of inline
1044 * fixed bug: the #S reader macro performs the keyword coercion
1045 specified for slot names. (reported by Kalle Niemitalo)
1046 * fixed bug: lambda lists may contain symbols whose names start with
1047 & but are not lambda-list-keywords; their occurrence triggers a
1049 * fixed bug 321: define-method-combination argument lambda lists do
1050 not cause invalid code to be generated when &optional or &aux
1051 variables have default values. (reported by Bruno Haible)
1052 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1053 have CLOS classes; this appears also to have fixed reported
1054 instability in stack exhaustion detection.
1055 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1056 MUFFLE-WARNING when no associated restart is present is now
1058 * optimization: rearranged the expansion of various defining macros
1059 so that each expands into only one top-level form in a
1060 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1062 * optimization: used a previously little-used slot in symbols to
1063 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1065 * fixed some bugs revealed by Paul Dietz' test suite:
1066 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1068 ** Fixed an optimization bug related to inheritance of initforms
1069 from local to shared slots.
1070 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1071 ** CAST optimizer forgot to flush argument derived type.
1072 ** print/read consistency on floats is now orders of magnitude
1073 more likely. (thanks also to Bruno Haible for a similar report
1075 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1076 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1077 for objects of type REAL. Make it so.
1078 ** SXHASH is specified (infelicitously) to respect similarity,
1079 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1080 it so. (thanks to Markus Ziegler)
1081 ** on the Alpha, the compiler succeeds in compiling functions
1082 returning a known number of arguments greater than 63.
1083 ** fixed handling of invalid NIL arguments in keyword position
1085 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1087 ** provided a readably-printable representation for RANDOM-STATE
1089 ** ensured that pathnames, where they have a namestring, always
1090 print using #P"..." syntax.
1092 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1093 * Support for the forthcoming 2.0 version of the NetBSD kernel
1094 running on x86 hardware has been added. (thanks to Perry
1095 E. Metzger most immediately, and others for their past work)
1096 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1097 current release version; SBCL's OpenBSD support had been broken
1098 since about the time of OpenBSD's switch to ELF binary format.)
1099 * [placeholder for DX summary]
1100 ** user code with &REST lists declared dynamic-extent, under high
1101 speed or space and low safety and debug optimization policy.
1102 * The manual has been converted to Texinfo format and the debugger
1103 chapter from the cmucl manual has been added.
1104 * A facility has been added to extract documentation strings from
1105 sbcl and store them as Texinfo-formatted snippets for inclusion in
1106 the manual (via Texinfo's @include directive)
1107 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1108 the test case to Dave Roberts)
1109 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1110 pointers. (reported by Sean Ross)
1111 * bug fix: PROFILE output is printed nicely even for large numerical
1112 values. (thanks to Zach Beane)
1113 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1114 greater than 32 handle EOF correctly.
1115 * bug fix: on X86 an immediate argument of the IMUL instruction is
1116 correctly printed in disassembly. (thanks to Lutz Euler)
1117 * bug fix: class slots in redefined classes preserve their old
1118 values. (thanks to Bruno Haible and Nikodemus Siivola)
1119 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1120 succeeds. (reported by Marco Baringer)
1121 * fixed some bugs revealed by Paul Dietz' test suite:
1122 ** READ-SEQUENCE now works on ECHO-STREAMs.
1123 ** RATIONALIZE works more according to its specification. (thanks
1126 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1127 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1128 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1129 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1130 should matter to you only if you rebind the printer control
1131 variables and then find you want different bindings in the
1132 debugger than in the ordinary execution of your program.)
1133 * The runtime build system has been tweaked to support building
1134 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1135 assembler and linker. (thanks to Nikodemus Siivola)
1136 * Unbound, undefined, undeclared variables now trigger full
1137 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1138 more often programmer error than deliberate exploitation of undefined
1140 * optimization: the hash algorithm for strings has changed to one
1141 that is less vulnerable to spurious collisions. (thanks to Juho
1143 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1144 needless bounds checking. (thanks to Juho Snellman)
1145 * optimization: implemented multiplication as a modular
1146 (UNSIGNED-BYTE 32) operation on the PPC backend.
1147 * fixed some bugs revealed by Paul Dietz' test suite:
1148 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1150 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1151 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1152 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1154 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1156 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1158 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1159 their output stream on EOF from read.
1160 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1161 have been read to end-of-file.
1162 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1164 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1165 description of determination of which consecutive characters
1167 ** Printing the "Space" character with escaping on now yields "#\\ ",
1168 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1169 ** Reading floating-point numbers with *READ-BASE* set to a number
1170 less than 10 works correctly.
1171 ** Reading floating-point numbers with *READ-BASE* set to a number
1172 more than 10 works correctly.
1173 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1174 the readtable currently in effect.
1176 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1177 * minor incompatible change: parsing of namestrings on a physical
1178 (Unix) host has changed; numbers after the final #\. in a
1179 namestring are no longer interpreted as a version field. This is
1180 intented to be largely invisible to the user, except that the
1181 meaning of the namestring "*.*.*" has changed: it now refers to a
1182 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1183 should usually be replaced by
1184 (make-pathname :name :wild :type :wild :version :wild)
1185 with the added benefit that this is more likely to be portable.
1186 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1187 option to OPEN now signals an error if the file being opened
1188 exists; this may have an impact on existing code.
1189 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1190 signals received, on the PowerPC platforms (both Linux and
1191 Darwin). (thanks to Pierre Mai for pointing out the location of
1193 * several fixes on OS X: The system now builds and runs cleanly on
1194 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1195 (thanks to Brian Mastenbrook)
1196 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1197 second-resolution: integer multiples of 1/3600 between -24 and 24.
1198 (thanks to Vincent Arkesteijn)
1199 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1200 type when called with 1 argument; PEEK-CHAR checked type of
1201 PEEK-TYPE only after having read first character from a
1202 stream. (reported by Peter Graves)
1203 * bug fix: the garbage collector now has much better locality
1204 behaviour, and in particular no longer treats all memory as being
1205 exhausted when many small objects point to each other in a deeply
1207 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1208 recognized as being TYPEP their class.
1209 * bug fix: the PUSHNEW documentation string has been corrected.
1210 (thanks to Vincent Arkesteijn)
1211 * bug fix: defaulting of the value for the last of an atypically
1212 large number of multiple values being bound was not being
1213 performed correctly on the Alpha or PPC platforms
1214 * optimization: implemented multiplication as a modular
1215 (UNSIGNED-BYTE 32) operation on the x86 backend.
1216 * optimization: SEARCH on simple-base-strings can now be open-coded.
1217 (see also contrib/compiler-extras.lisp for inspiration for
1218 teaching the compiler about the Boyer-Moore algorithm).
1219 * value, returned by MAX (and MIN) called with several EQUALP, but
1220 not EQL, arguments now does not depend on compiler settings.
1221 * fixed some bugs revealed by Paul Dietz' test suite:
1222 ** in stack analysis liveness information is propagated from
1223 non-local entry points.
1224 ** pathwise CAST removing failed when the CAST node did not start
1226 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1227 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1229 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1230 host is already defined.
1231 ** RENAME-FILE works on streams instead of signalling an internal
1233 ** PEEK-CHAR uses the current readtable when determining whether
1234 or not a character is whitespace.
1235 ** MERGE-PATHNAMES handles the case when the pathname does not
1236 specify a name while the default-pathname specifies a version.
1237 ** Pathnames now stand a better chance of respecting print/read
1239 ** Attempting to use standardized file system operators with a
1240 pathname with invalid :DIRECTORY components signals a
1242 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1244 ** DIRECTORY on logical pathnames is more correct.
1245 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1246 signal a TYPE-ERROR if their argument is not a stream.
1247 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1248 designator argument does not designate a stream.
1249 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1250 examining the synonym.
1251 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1253 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1254 element-type larger than ([UN]SIGNED-BYTE 32).
1256 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1257 * When built with the :SB-FUTEX feature, threaded builds now take
1258 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1259 for faster/more reliable mutex and condition variable support.
1260 * Incompatible change (but one you probably shouldn't have been using
1261 anyway): the interface and code for arbitrating between multiple
1262 threads in the same user session has been redesigned.
1263 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1264 error when called without an explicit environment argument.
1265 (thanks to Nikodemus Siivola)
1266 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1267 confused about their position. (thanks to Adam Warner and Gerd
1269 * bug fix: Pretty printing backquoted forms with unquotations in the
1270 argument list position of various code constructs such as LAMBDA
1271 now works correctly. (reported by Paul Dietz)
1272 * bug fix: Pretty printing unquotations no longer loses all
1273 stream position information.
1274 * optimization: performance of string output streams is now less
1275 poor for multiple small sequence writes.
1276 * optimization: performance of CSUBTYPEP in the presence of complex
1277 expressions involving CONS and NOT many times has been improved.
1278 (reported by Paul Dietz)
1279 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1281 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1282 streams for the REPL, for the debugger, and so on. (thanks to
1284 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1285 * fixed some bugs revealed by Paul Dietz' test suite:
1286 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1287 is ignored if it is too small, rather than propagating through
1288 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1289 ** extremely complex negations of CONS types were not being
1290 sufficiently canonicalized, leading to inconsistencies in
1292 ** VALUES tranformer lost derived type.
1294 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1295 * fixed a bootstrapping bug: the build process no longer assumes
1296 that the various BOOLE-related constants have the same value in
1297 host and target lisps. (noted by Paul Dietz' test suite on an
1298 SBCL binary built from CLISP)
1299 * The system can now be dynamically linked on the MIPS platform,
1300 which enables dynamic loading of foreign code from Lisp. (thanks
1301 to Ralf Baechle for discussions on the MIPS ABI)
1302 * The system now records debugging information for its own source
1303 files in a filesystem-position-independent manner, relative to
1304 the "SYS" logical host.
1305 * fixed a compiler bug: MV-LET convertion did not check references
1306 to the "max args" entry point. (reported by Brian Downing)
1307 * tweaked disassembly notes to be less confident about proclaiming
1308 some instruction as an LRA. (thanks to Brian Downing)
1309 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1310 now each have their own history, command character, and other
1311 characteristics. (thanks to David Lichteblau)
1312 * fixed some bugs revealed by Paul Dietz' test suite:
1313 ** compiler failure in compiling LOGAND expressions including a
1315 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1316 shift greater than 32.
1317 ** FLUSH-DEST did not mark blocks for type check regeneration.
1318 ** HANDLER-CASE failed to accept declarations in handler clauses
1319 in some circumstances.
1321 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1322 * New code in contrib/sb-introspect (still probably not entirely
1323 stable yet) provides some support for smart Lisp development
1324 environments like SLIME.
1325 * The conditions signalled for errors occurring when loading .fasl
1326 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1327 in a way which should help ASDF recover gracefully.
1328 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1329 stuff has been cleaned up. If you code contrib/ stuff, this might
1330 affect you, and you can look at contrib/README, contrib/STANDARDS,
1331 and/or the 0.8.4.27 diff to check.
1332 * In full calls the compiler now does not generate checks for declared
1333 argument types for all arguments.
1334 * various threading fixes
1335 ** and some experimental patches which didn't make it into
1336 the main tree for this release, but which are shipped in
1337 contrib/experimental-thread.patch as a possible fix for some
1338 failures (deadlock, spinning...) in GC-intensive multithreaded
1340 * fixed PPC build problem (source code incompatibility of different
1341 library versions): added offsetof() hackery which attempts to divine
1342 where glibc maintainers put uc_mcontext today
1343 * fixed bug 282: compiler does not trust type assertions while passing
1344 arguments to a full call.
1345 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1346 &OPTIONAL VALUES type parameter.
1347 * fix bug 214: algorithm for noting rejected templates is now more
1348 similar to that of template seletion. (also reported by rydis on
1350 * fixed bug 141b: printing backquoted information readably and prettily
1351 inserts a space where necessary.
1352 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1353 (CHARACTER 10) are now reported as errors, rather than propagated
1354 as unknown types. (reported by piso on #lisp)
1355 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1356 with values NIL and :ERROR. (thanks to Milan Zamazal)
1357 * fixed bug 191c: CLOS now does proper keyword argument checking as
1358 described in CLHS 7.6.5 and 7.6.5.1.
1359 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1360 longer raise an error; further, using a list as a for-as-arithmetic
1361 counter now raises a meaningful error.
1362 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1363 now understood by sequence creation functions such as MAKE-SEQUENCE
1365 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1366 a STRING-INPUT-STREAM.
1367 * compiler enhancement: SIGNUM is now better able to derive the type
1369 * type declarations inside WITH-SLOTS are checked. (reported by
1371 * fixed some bugs revealed by Paul Dietz' test suite:
1372 ** incorrect optimization of TRUNCATE for a positive first
1373 argument and negative second.
1374 ** compiler failure in let-convertion during flushing dead code.
1375 ** compiler failure while deriving type of TRUNCATE on an
1376 interval, containing 0.
1377 ** ASH of a negative bignum by a negative bignum count now returns
1379 ** intersection of CONS types now canonicalizes properly, fixing
1380 inconsistencies in SUBTYPEP.
1382 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1383 * incompatible change: The --disable-debugger command line
1384 option now clobbers the debugger at a more fundamental
1385 level, by redefining #'INVOKE-DEBUGGER instead of by
1386 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1387 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1388 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1389 enter the debugger REPL and then suffer recursive errors
1390 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1391 while under the new system, BREAK in --disable-debugger mode
1392 terminates the system just as an unhandled error would.
1393 * fixed compiler performance when processing loops with a step >1;
1394 * bug fix: DOCUMENTATION now retrieves generic function
1395 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1396 support has been systematized, and now supports the methods
1397 specified by ANSI, along with a default method and a method for
1398 slot documentation. (reported by Nathan Froyd)
1399 * bug fix: effective methods associated with a generic function are
1400 no longer cached over a change of that generic function's method
1401 combination. (reported by Andreas Fuchs)
1402 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1403 names a non-existent directory. (thanks to Andreas Fuchs)
1404 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1405 the PPC platform, lead to stack corruption; this has been fixed.
1406 (reported by Rainer Joswig)
1407 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1408 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1409 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1410 arithmetic in general, is now much more reliable on the PPC
1412 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1413 platform now returns the right answer.
1414 * optimization: restored some effective method precomputation in
1415 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1416 precomputation is now tunable.
1417 * optimization: compiler-internal data structure use has been
1418 reviewed, and changes have been made that should improve the
1419 performance of the compiler by about 20%.
1420 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1421 simple-base-strings and simple-bit-vectors is improved.
1422 * optimization: the optimization of 32-bit logical and arithmetic
1423 functions introduced in version 0.8.3 on the x86 has been
1424 implemented on the mips, ppc and sparc platforms; an
1425 implementation of the same facility, but for 64-bit arithmetic,
1426 has been added for the alpha.
1427 * microoptimization: the compiler is better able to make use of the
1428 x86 LEA instruction for multiplication by constants.
1429 * bug fix: in some situations compiler did not report usage of
1430 generic arithmetic in (SPEED 3) policy.
1431 * bug 145b fix: compiler used wrong type specifier while converting
1432 MEMBER-types to numeric.
1433 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1435 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1436 index argument now works properly on the Alpha platform.
1437 * bug fix: floating point exception treatment on the Alpha platform
1439 * bug fix: FILE-POSITION works much better on string input and
1440 output streams. (thanks to Nikodemus Siivola)
1441 * bug fix: many threading/garbage collection symptoms sorted.
1442 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1443 might be pseudo-atomic.
1444 * internal change: Stopping for GC is now done with signals not
1445 ptrace. GC is now done in whichever thread wanted it, instead of
1447 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1449 * bug fix: result form in DO is not contained in the implicit
1451 * incompatible change: ICR structure is changed; the value part of
1452 CONTINUATION is now called LVAR; corresponding functions are
1453 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1454 * added type deriver for ISQRT (thanks to Robert E. Brown).
1455 * bug fix: better support for loading from the command line when an
1456 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1458 * fixed some bugs revealed by Paul Dietz' test suite:
1459 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1460 ** ROUND and FROUND now give the right answer when given very
1461 small float arguments.
1462 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1464 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1465 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1466 ** LCM should return a non-negative integer.
1467 ** PARSE-INTEGER returned the index of a terminator instead of the
1468 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1469 ** PARSE-INTEGER returned an incorrect index being applied to a
1471 ** LCM with two arguments of 0 returns 0 rather than signalling
1473 ** unsigned addition of a 32-bit constant with the high bit set no
1474 longer causes an internal compiler error on the x86.
1475 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1476 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1477 with negative last argument.
1478 ** byte specifiers with zero size and position no longer cause
1479 an error during type derivation.
1480 ** bignum multiplication on the Alpha platform now returns the
1482 * porting: The system now builds on SuSE AMD64, although it still
1483 generates a 32-bit binary.
1484 * .fasl file incompatibility: The fasl file version number has
1485 been incremented (because of the changes to internal compiler
1486 data structures referred to above).
1488 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1489 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1490 more accurately, on the Darwin kernel running on PowerPC hardware.
1491 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1492 * Compiler code deletion notes now signal a condition of type
1493 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1494 an associated MUFFLE-WARNING restart.
1495 * The compiler now performs limited argument count validation of
1496 constant format strings in FORMAT, and where appropriate in ERROR,
1497 CERROR and WARN. (thanks to Gerd Moellmann)
1498 * New ASDF-INSTALL contrib can be used for automatic download and
1499 installation of third-party Lisp code from CCLAN or other sites
1501 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1502 which forces another thread to execute a function supplied by the
1504 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1505 now accept and act upon their :ELEMENT-TYPE keyword argument.
1506 (reported by Martin Atzmueller, Edi Weitz)
1507 * bug fix: FILE-POSITION now accepts position designators up to
1508 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1509 is the greater. (thanks to Patrik Nordebo)
1510 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1511 Kalle Olavi Niemitalo)
1512 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1513 printable. (reported by Eric Marsden)
1514 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1515 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1516 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1517 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1518 (thanks to Stig Sandoe)
1519 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1520 UNREAD-CHAR now correctly includes the unread character in the
1521 target sequence. (thanks to Gerd Moellmann)
1522 * bug fix in threaded builds: the system can now be suspended and
1523 resumed by shell job control with minimal disruption.
1524 * bug fixes in times and timezones >2038AD
1525 * better handling of "where is GNU make?" problem in build scripts
1526 (thanks to Nikodemus Siivola)
1527 * new optimization: inside a named function any reference to a
1528 function with the same name is considered to be a self-reference;
1529 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1530 optimization quality.
1531 * new optimization on x86: logical functions and + now have
1532 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1533 used when the result is truncated to 32 bits.
1534 * VALUES declaration is partially enabled.
1535 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1536 * fixed some bugs revealed by Paul Dietz' test suite:
1537 ** The system now obeys the constraint imposed by
1538 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1539 types form a lattice under type intersection.
1540 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1541 ** ASSOC now ignores NIL elements in an alist.
1542 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1543 and (1+ MOST-POSITIVE-FIXNUM) answers.
1544 ** The addition of a method with invalid qualifiers to a generic
1545 function does not cause an error to be signalled immediately;
1546 a warning is signalled, and the error is generated only on
1547 calling the generic function.
1548 * changed .fasl file version number, in part to add type codes for
1549 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1550 obscure ANSI requirements
1552 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1553 * fixed bug 148: failure to inline-expand a local function left
1554 garbage, confusing the compiler.
1555 * fixed bugs 3cd: structure slot readers perform type check if the
1556 slot can have an invalid value (i.e. it is either not initialized
1557 or can be written with a less specific slot writer).
1558 * bug fix: the compiler now traps array references to elements off
1559 the end of an array; previously, the bounds checking in some
1560 circumstances could go off-by-one.
1561 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1563 * type declarations for array element types now obey the description
1564 on the CLHS page "Declaration TYPE", as per discussions on
1565 sbcl-help around 2003-05-08. This means that a declaration
1566 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1567 declaration, all references to BAR will be asserted or assumed
1568 (with THE, so dependent on compiler policy) to involve objects of
1569 type FOO. Note that no such declaration is implied in
1570 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1571 * declared types of functions from the "Conditions"
1572 chapter. (reported by Paul Dietz)
1573 * bug fix: CERROR accepts a function as its first argument.
1574 * bug fix: NTH an NTHCDR accept a bignum as index
1575 arguments. (reported by Adam Warner)
1576 * optimization: character compare routines now optimize comparing
1577 against a constant character. (reported by Gilbert Baumann)
1578 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1579 argument now works properly on the MIPS platform.
1580 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1581 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1582 declared to have a functional type, does not cause an error
1584 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1585 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1586 * fixed bug: if last continuation of a deleted block has a
1587 destination, this destination should be deleted too. (reported by
1589 * fixed a bug in the bootstrap process: the host compiler's values
1590 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1591 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1592 test case from Patrik Nordebo)
1593 * improved the ability of the disassembler on the PPC platform to
1594 provide helpful disassembly notes.
1595 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1596 the class in more cases than previously.
1597 * bug fix: FILE-POSITION now understands :START and :END for
1598 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1599 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1600 8) data. (thanks to David Lichteblau)
1601 * bug fix: it is possible to add a method to a generic function
1602 without lambda list.
1603 * bug fix: reader failed to signal END-OF-FILE inside an
1604 object representation. (reported by Nikodemus Siivola)
1605 * fixed some bugs revealed by Paul Dietz' test suite:
1606 ** LAST and [N]BUTLAST should accept a bignum.
1607 ** condition slot accessors are methods.
1608 ** (VECTOR NIL) is a subtype of STRING.
1610 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1611 * minor incompatible change: some nonsensical specialized lambda
1612 lists (used in DEFMETHOD) which were previously ignored now signal
1614 * minor incompatible change: the system is now aware of the types of
1615 variables in the COMMON-LISP package, and will signal errors for
1616 most violations of these type constraints (where previously they
1617 were silently accepted).
1618 * minor incompatible change: COMPILE-FILE now uses the freedom
1619 afforded (ANSI 3.2.2.3) to use derived function types for
1620 functions defined in the same file. This also permits the system
1621 to warn on static type mismatches and function
1622 redefinition. (Currently it does not work with high DEBUG level.)
1623 * minor incompatible change: VALUES declaration is disabled.
1624 * When issuing notes, the compiler now signals a condition of type
1625 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1626 restart for use in user handlers. It is expected that the
1627 COMPILER-NOTE condition will eventually become a condition
1628 supertype to a hierarchy of note types, which will then be
1629 handleable in a similar fashion. However, at the moment, no such
1630 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1631 but it's an implementation detail, not a classification for the
1633 * Changes in type checking closed the following bugs:
1634 ** type checking of unused values (192b, 194d, 203);
1635 ** template selection based on unsafe type assertions (192c, 236);
1636 ** type checking in branches (194bc).
1637 * A short form of VALUES type specifier has ANSI meaning (it has
1638 increased the number of situations when SBCL cannot perform type
1640 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1641 keywords or constants is permissible.
1642 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1643 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1644 classes. (thanks to Antonio Martinez)
1645 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1646 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1647 argument to a pretty-print stream. Instead, it leaves any such
1648 operations to DESCRIBE-OBJECT methods.
1649 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1651 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1652 lambda lists. (thanks to David Lichteblau)
1653 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1654 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1655 (thanks to Gerd Moellmann)
1656 * fixed bug 246: increased compilation speed of long
1657 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1659 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1661 * DEFCONSTANT now throws a condition of type
1662 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1663 constant to a non-EQL value; CONTINUE and ABORT restarts
1664 respectively change and preserve the value.
1665 * fixed bug 63: The code walker, part of the implementation of CLOS,
1666 is now better at handling symbol macros.
1667 * bug fix: There is no longer an internal implementation type named
1668 CL:LENGTH. (reported by Raymond Toy)
1669 * bug fix: In macro-like defining macros/special operators the
1670 implicit block does not enclose the lambda list.
1671 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1672 enclosed in parentheses) are not suitable as type specifiers, and
1673 their use properly signals an error now.
1674 * bug fix: An argument count mismatch for a type specifier in code
1675 being compiled no longer causes an unhandled error at compile
1676 time, but signals a compile-time warning.
1677 * fixed simple vector readable printing
1678 * bug fix: DESCRIBE takes more care over whether the class
1679 precedence list slot of a class is bound before accessing it.
1680 (reported by Markus Krummenacker)
1681 * bug fix: FORMATTER can successfully compile pretty-printer format
1682 strings which use variants of the ~* directive inside.
1683 * bug fix: SEARCH now applies its TEST predicate to the elements of
1684 the arguments in the correct order. (thanks to Wolfhard Buss)
1685 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1686 in type inference. (thanks to Robert E. Brown)
1687 * bug fix: Reading in symbols with an explicit package name of ""
1688 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1689 * fixed some bugs revealed by Paul Dietz' test suite:
1690 ** NIL is now allowed as a structure slot name.
1691 ** Arbitrary numbers, not just REALs, are allowed in certain
1692 circumstances in LOOP for-as-arithmetic clauses.
1693 ** Multiple class redefinitions before slot access no longer
1694 causes a type error.
1695 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1696 association between the name and a class.
1697 ** Generic functions with non-standard method-combination and over
1698 five methods all of which return constants no longer return NIL
1699 after the first few invocations. (thanks to Gerd Moellmann)
1700 ** CALL-NEXT-METHOD with no arguments now passes the original
1701 values of the arguments, even in the presence of assignment.
1702 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1704 ** DISASSEMBLE works with closures and funcallable instances.
1705 ** ADD-METHOD now returns the generic function, not the new method.
1706 ** FIND-METHOD signals an error if the lengths of the specializers
1707 is incompatible with the generic function, even if the ERRORP
1709 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1710 which its argument is a member.
1711 ** DEFCLASS only redefines the class named by its class-name
1712 argument if that name is the proper name of the class;
1713 otherwise, it creates a new class.
1714 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1715 of the UNBOUND-SLOT condition to the name of the slot.
1716 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1717 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1718 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1719 SLOT-MAKUNBOUND in the specified fashion.
1721 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1722 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1723 cross-compilation host. As a consequence, we can now bootstrap our
1724 way up to SBCL starting with a bare gcc toolchain and human-readable
1725 source code (first the source to CLISP, then the source to SBCL).
1726 * A contributed module containing a partial implementation of the
1727 simple-streams interface has been included. (thanks to Rudi
1729 * A contributed module implementing the RFC1321 Message Digest
1730 Algorithm, known as MD5, has been included.
1731 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1732 no longer has any effect, as the code controlled by this feature
1733 has been deleted. (As far as we know, no-one has ever built using
1734 this feature, and its semantics were confused in any case).
1735 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1736 work on conditions (as required by the ANSI specification),
1737 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1738 expected behaviour on conditions. Users should note, however,
1739 that such behaviour is not required by the ANSI specification,
1740 and so use of this behaviour may render their code unportable.
1741 * fixed some bugs revealed by Paul Dietz' test suite:
1742 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1744 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1745 ** SXHASH on condition objects no longer returns NIL.
1746 ** :ALLOCATION :CLASS slots are better treated; their values are
1747 updated on class redefinition, and initforms inherited from
1748 superclasses are applied.
1749 ** REMOVE-METHOD returns its generic function argument even when
1750 no method was removed.
1751 ** SHARED-INITIALIZE now initializes the values of the requested
1752 slots, including those with :ALLOCATION :CLASS.
1753 ** ALLOCATE-INSTANCE now works on structure classes defined via
1754 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1756 ** SLOT-EXISTS-P now works on conditions, as well as structures
1758 ** MAKE-LOAD-FORM now has the required methods on
1759 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1760 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1761 keyword, and now implements the SLOT-NAMES argument.
1762 ** methods with &OPTIONAL arguments no longer allow too many
1763 arguments to be passed in the call without error.
1764 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1765 option is consistent with the required arguments of the generic
1766 function lambda list.
1767 * bug fix: REQUIRE accepts a string designator. (Thanks to
1769 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1770 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1771 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1772 * bug fix: The compiler now checks for duplicated variables in macro
1774 * bug fix: SETQ on globals returns the correct value.
1775 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1776 not a COMPILER-ERROR (followed by some other strange error on
1777 choosing the CONTINUE restart).
1778 * bug fix: make.sh and friends are now more consistent in the way that
1779 they look for GNU "make".
1781 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1782 * experimental native threads support (on x86 Linux >=2.4 only).
1783 This is not compiled in by default: you need to add :SB-THREAD to
1784 the target features. See the "Beyond ANSI" chapter of the manual
1786 * fix for longstanding nonANSIism: The old distinction between
1787 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1788 The return value from CL:FIND-CLASS is now a CLOS class, and
1789 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1790 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1791 * An interface to the MetaObject Protocol, as described in Kiczales,
1792 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1793 MIT Press, 1991, is available from the SB-MOP package.
1794 * incompatible change: the SB-PCL package should now be considered
1795 a private implementation detail, and no longer a semi-private MOP
1797 * minor incompatible change: due to rearrangement for threads, the
1798 control stack and binding stack are now allocated at arbitrary
1799 addresses instead of being hardcoded per-port. Users affected by
1800 this probably have to be doing advanced things with shared
1801 libraries, and will know who they are.
1802 * minor incompatible change: Previously, all --eval forms used were
1803 processed with READ before any of them were processed with EVAL.
1804 Now each --eval form is processed with both READ and EVAL before
1805 the next --eval form is processed. (Thus package operations like
1806 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1807 work as the user might reasonably expect.)
1808 * minor incompatible change: *STANDARD-INPUT* is now only an
1809 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1811 * minor incompatible change: Y-OR-N-P is now character-oriented,
1812 not line oriented. Also, YES-OR-NO-P now works without errors.
1813 (thanks to Antonio Martinez)
1814 * sb-aclrepl module improvements: an integrated inspector, added
1815 repl features, and a bug fix to :trace command.
1816 * Known functions, which cannot be open coded by the backend, are
1817 considered to be able to check types of their arguments. (fixing
1818 a bug report by Nathan J. Froyd)
1819 * fixed a bug in computing method discriminating functions: It is
1820 now possible to define methods specialized on classes which have
1821 forward-referenced superclasses. (thanks to Gerd Moellmann)
1822 * fixed evaluation order in optional entries (reported by Gilbert
1824 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1825 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1826 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1827 required-by-AMOP NAME argument, as well as CLASS and
1828 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1829 * fixed bug 20: DEFMETHOD can define methods using names that are
1830 not the proper names of classes to designate class specializers.
1831 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1832 Unix file descriptors, instead of blowing up. (thanks to Antonio
1834 * Garbage collection refactoring: user-visible change is that a
1835 call to the GC function during WITHOUT-GCING will not do garbage
1836 collection until the end of the WITHOUT-GCING. If you were doing
1837 this you were probably losing anyway.
1838 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1839 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1840 TYPEP the latter but not the former.
1841 * The compiler issues a full WARNING for calls to undefined functions
1842 with names from the CL package.
1843 * MAP-INTO for a vector destination is open coded. (reported by
1844 Brian Downing on c.l.l)
1845 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1846 documentation string.
1847 * fixed some bugs revealed by Paul Dietz' test suite:
1848 ** COPY-ALIST now signals an error if its argument is a dotted
1850 ** Condition slots are now accessed more correctly in the presence
1851 of multiple initargs for a given slot.
1852 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1853 exclude restarts of the same name associated with a different
1855 ** DEFCLASS of forward-referenced classes with another
1856 forward-referenced class in the superclasses list no longer
1858 ** Condition slots are now initialized once each, not multiple
1859 times. (thanks to Gerd Moellmann)
1860 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1862 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1863 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1864 arguments contain duplicated elements.
1865 ** RESTART-CASE understands local macros.
1866 ** RESTART-CASE associates exactly its own restarts with a condition.
1867 ** ENDP in safe mode checks its argument to be of type LIST.
1868 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1869 in question is unbound.
1870 ** Optimized MAKE-INSTANCE functions no longer cause internal
1871 assertion failures in the presence of duplicate initargs.
1872 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1873 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1874 propagated even in the ctor optimized implementation of
1876 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1878 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1879 * a better implementation of SXHASH on (simple) bit vectors,
1880 measured both in execution speed and in distribution of results
1881 over the positive fixnums, has been installed. Likewise, a better
1882 implementation of EQUAL for simple bit vectors is now available.
1883 * fixed CEILING optimization for a divisor of form 2^k.
1884 * fixed bug 240 (emitting extra style warnings "using the lexical
1885 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1886 by Antonio Martinez)
1887 * fixed SXHASH, giving different results for NIL depending on type
1888 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1889 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1890 variable. (found by Rolf Wester)
1891 * SBCL does not ignore type declarations for special
1892 variables. (reported by rif on c.l.l 2003-03-05)
1893 * some bug fixes in contrib/sb-aclrepl/
1894 * fixed some bugs revealed by Paul Dietz' test suite:
1895 ** a bug in the CONS type specifier, whereby the CAR and CDR
1896 types got intertwined, has been fixed;
1897 ** the type system is now able to reason about the interaction
1898 between INTEGER and RATIO types more completely;
1899 ** APPEND, [N]REVERSE and NRECONC check that those their
1900 arguments, which must be proper lists, are really so;
1901 ** An array specialized to be unable to hold elements has been
1902 implemented, as required -- yes, really -- by ANSI;
1903 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1904 on malformed property lists;
1906 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1907 * incompatible packaging change: in line with Unix convention,
1908 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1909 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1910 * REQUIRE and PROVIDE are now optionally capable of doing something
1911 useful. See the documentation string for REQUIRE.
1912 * infrastructure for a managed SBCL contrib system: contributed
1913 modules in this release include:
1914 ** the ASDF system definition facility;
1915 ** an interface to the BSD Sockets API;
1916 ** an ACL-like convenience interface to the repl;
1917 (thanks to Kevin Rosenberg)
1918 ** an implementation of ROTATE-BYTE, with efficient implementation
1920 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1921 gives a closer approximation to the right answer than previously.
1922 (thanks to Raymond Toy)
1923 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1924 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1925 optional environment arguments, as required by ANSI.
1926 * fixed bugs in other functions taking environment objects, allowing
1927 calls with an explicit NIL environment argument to be compiled
1929 * fixed bug 228: primary return values from
1930 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1931 COMPILE or FUNCTION.
1932 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1933 :TYPE will no longer signal errors on innocuous objects.
1934 * fixed bug 231b: SETQ is better at respecting type declarations in
1935 the lexical environment.
1936 * fixed a bug in DEFCLASS: classes named by symbols with no or
1937 unprintable packages can now be defined.
1938 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1939 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1940 * fixed compiler failure related to checking types of functions.
1941 (reported by Robert E. Brown)
1942 * the compiler is now much more consistent in its error-checking
1943 treatment of bounding index arguments to sequence functions: in
1944 (SAFETY 3) code, errors will be signalled in almost all cases if
1945 invalid sequence bounding indices are passed to functions defined
1946 by ANSI to operate on sequences.
1947 * fixed a bug in the build procedure: documentation of SBCL-specific
1948 packages is now preserved and available in the final Lisp image.
1949 * lifted FDEFINITION lookup out of loops in the implementation of
1950 many list operations. (thanks to Robert E. Brown)
1951 * fixed a bug in the reader: the #n# reader macro now works for
1952 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1953 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1954 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1955 E. Brown and Tony Martinez respectively)
1956 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1957 SLOT-BOUNDP: in particular, it is now able to optimize them much
1958 better, and is now not vulnerable to having packages renamed.
1959 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1960 always called when a slot is not present in an instance. (thanks
1962 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1963 longer signal an error on unreading a character following EOF on
1964 the previous constituent stream. (thanks to Tony Martinez)
1965 * fixed some bugs revealed by Paul Dietz' test suite:
1966 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1967 not just nonnegative fixnums;
1968 ** the logical bit-array operators such as BIT-AND now accept an
1969 explicit NIL for their "opt-arg" argument (to indicate a
1970 freshly-consed result bit-array);
1971 ** ELT now signals an error on an invalid sequence index in safe
1973 ** the type system is now cleverer about negations of numeric
1974 types, and consequently understands the BIGNUM and RATIO types
1976 ** the type system is now cleverer about the interaction between
1977 INTEGER and RATIO types: while bugs still remain, many more
1978 cases are accurately computed;
1979 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1980 if it is in the last clause;
1981 ** CONSTANTLY now correctly returns a side-effect-free function in
1983 ** DECLARE is no longer treated as a special-operator; in
1984 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1985 * incremented fasl file version number due to the change in the
1986 DEFSTRUCT-SLOT-DESCRIPTION structure.
1988 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1989 * minor incompatible change: code processed by the "interpreter" or
1990 EVAL now has a compilation optimization policy of (DEBUG 2)
1991 (changed from (DEBUG 1)) to improve debuggability of interactive
1992 development, and to allow the use of the debug RETURN command in
1994 * an experimental implementation of the RETURN command for the
1995 debugger has been included. (thanks to Frederik Kuivinen)
1996 * fixed bug 62: constraints were not propagated into a loop.
1997 * fixed bug in embedded calls of SORT (reported and investigated by
1999 * fixed some bugs revealed by Paul Dietz' test suite:
2000 ** printing and reading of arrays with some dimensions having
2001 length 0 (thanks to Gerd Moellmann);
2002 ** BOA constructor with &AUX argument without a default value does
2003 not cause a type error;
2004 ** CONSTANTP now returns true for all self-evaluating objects.
2006 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2007 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2008 accessors that are related by inheritance, as specified in the
2009 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2010 Valtteri Vuorikoski)
2011 * The compiler is now able to inline functions that were defined in
2012 a complex lexical environment (e.g. inside a MACROLET).
2013 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2015 * Support for the upcoming FreeBSD-5.0 release has been included.
2016 (thanks to Dag-Erling Smorgrav)
2017 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2018 effect when it is not in a toplevel context.
2019 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2020 stand a better chance of being correct. (thanks to Gerd
2022 * fixed bug in COERCE, which now signals an error on coercing a
2023 rational to a bounded real type which excludes the expected
2025 * The compiler is now able to derive types more accurately from the
2026 COERCE and COMPILE functions.
2027 * fixed bug 223: functional binding is considered to be constant
2028 only for symbols in the CL package.
2029 * fixed bug 231: SETQ did not check the type of a variable being set
2030 (reported by Robert E. Brown)
2031 * A new optimization for MAKE-INSTANCE has been included, fixing
2032 various bugs (including relating to :ALLOCATION :CLASS slots and
2033 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2035 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2036 ** As required by ANSI, LOOP now disallows anonymous collection
2037 clauses such as COLLECT I in conjunction with aggregate boolean
2038 clauses such as THEREIS (= I 1);
2039 ** LOOP now signals an error when any variable is reused in the
2040 same loop (including the potentially useful construct analogous
2041 to WITH A = 1 WITH A = (1+ A);
2042 ** IT is only a special loop symbol within the first clause of a
2043 conditional loop clause;
2044 ** LOOP with a typed iteration variable over a hashtable now
2045 signals a type error iff it should.
2046 * fixed some other bugs revealed by Paul Dietz' test suite:
2047 ** FILE-STREAM now names the class previously known as FD-STREAM;
2048 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2049 argument) no longer signals an error;
2050 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2051 of the slot symbol, rather than using the current package
2052 ((:CONC-NAME "") continues to intern the slot's name in the
2054 * incremented fasl file version number, because of the incompatible
2055 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2056 of the new implementation of DEFINE-COMPILER-MACRO.
2058 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2059 * Support for building SBCL for MIPS platforms running in
2060 little-endian mode has now been checked in, and basic
2061 functionality on said platforms verified.
2062 * minor incompatible change: PCL now records the pathname of a file
2063 in which methods and the like are defined, rather than its
2065 * minor incompatible change: TRUENAME now considers the truename of
2066 a file naming a directory to be the pathname with :DIRECTORY
2067 component indicating that directory.
2068 * minor incompatible change: a NAMED clause in the extended form of
2069 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2070 reason for the previous behaviour is unclear.
2071 * more systematization and improvement of CLOS and MOP conformance
2072 in PCL (thanks to Gerd Moellman and Pierre Mai):
2073 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2075 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2076 primary methods with no specializers;
2077 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2079 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2080 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2081 FORWARD-REFERENCED-CLASSes; error reporting on
2082 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2084 ** SXHASH on CLOS instances now uses a slot internal to the
2085 instance to return different numbers on distinct instances,
2086 while preserving the same return value through invocations of
2088 ** DEFMETHOD signals errors when methods with longer incongruent
2089 lambda lists are added to generic functions;
2090 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2091 CLASS, as specified in AMOP;
2092 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2093 on the order returned by the primary method for classes of
2094 class STANDARD-CLASS;
2095 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2096 * fixed some bugs shown by Paul Dietz' test suite:
2097 ** DOLIST puts its body in TAGBODY;
2098 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2100 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2101 value producing form;
2102 ** if more variables are given to PROGV than values, extra
2103 variables are bound and made to have no value;
2104 ** NSUBSTITUTE on list arguments gets the right answer with
2106 ** ELT signals an error of type TYPE-ERROR when the index argument
2107 is not a valid sequence index;
2108 ** LOOP signals (at macroexpansion time) an error of type
2109 PROGRAM-ERROR when duplicate variable names are found;
2110 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2111 ** FUNCALL of special-operators now cause an error of type
2113 ** PSETQ now works as required in the presence of side-effecting
2114 symbol-macro places;
2115 ** NCONC accepts any object as its last argument;
2116 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2118 ** loop-for-as-package does not require a package to be explicitely
2120 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2122 * fixed bug 166: compiler preserves "there is a way to go"
2123 invariant when deleting code.
2124 * fixed bug 172: macro lambda lists with required arguments after
2125 &REST arguments now cause an error to be signalled. (thanks to
2127 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2128 bug. (thanks to Matthew Danish)
2129 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2131 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2132 arguments are lexically rebound. (thanks to Gerd Moellmann and
2134 * fixed bug 194: error messages are now more informative when there
2135 is no primary method applicable in a call to a generic
2136 function. (thanks to Gerd Moellmann)
2137 * fixed bug in command line argument checking (thanks to Julian
2139 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2140 (thanks to Matthew Danish)
2141 * incremented fasl file version number, because of the
2142 SXHASH-related changes in the layout of CLOS data structures
2144 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2145 * minor incompatible change: The runtime (the Unix executable named
2146 "sbcl") is now much pickier about the .core files it will load.
2147 Essentially it now requires .core files to descend from the same
2148 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2149 as the runtime does. (The intent is to prevent the crashes which
2150 can occur, and which can even be reported as mysterious failures,
2151 when people patch the sources or change the build parameters
2152 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2153 sbcl and .core files.)
2154 * fixed bug: VALUES-LIST is no longer optimized away.
2155 * fixed bug 142: The FFI conversion of C string values to Lisp
2156 string values no longer conses excessively. (thanks to Nathan
2157 Froyd porting Raymond Toy's fix to CMU CL)
2158 * began to systematize and improve MOP conformance in PCL (thanks to
2159 Nathan Froyd, Gerd Moellman and Pierre Mai):
2160 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2162 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2163 ** FINALIZE-INHERITANCE is now called on class finalization;
2164 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2165 argument precedence order.
2166 * fixed bug 202: The compiler no longer fails on functions whose
2167 derived types contradict their declared type.
2168 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2169 so it can be non-toplevel.
2170 * The fasl file version number has changed (because of the new
2171 implementation of DEFMACRO).
2172 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2173 safe code, that any length requirement by their type-specifier
2174 argument is valid. The exceptions to this are described in bug
2176 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2177 functions now check the types of their inputs as required by ANSI.
2178 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2179 introduced symbol is DECLAREd to be SPECIAL.
2180 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2181 * fixed bug: :COUNT argument to sequence functions may be negative
2182 * fixed bug: body of DO-SYMBOLS may contain declarations
2183 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2184 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2185 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2186 symbol macro only once
2187 * fixed printing of call frame when argument list is unavailable
2188 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2189 * compiler no longer signals WARNING on unknown keyword
2192 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2193 * A beta-quality port to the mips architecture running Linux,
2194 based on the old CMUCL backend, has been made. It has been tested
2195 on a big-endian kernel, and works sufficiently well to be able to
2196 rebuild itself; it has not been tested in little-endian mode.
2197 * fixed an inconsistency between gencgc.c and purify.c which made
2198 dumping/loading .core files unreliable
2199 * fixed bug 120a: The compiler now deals correctly with IFs where
2200 the consequent is the same as the alternative, instead of
2201 misderiving the return type. (thanks to Alexey Dejneka)
2202 * fixed bug 113: Logical pathnames are now dumpable (the logical
2203 host is resolved at load-time, throwing an error if it is not
2205 * fixed bug 174: FORMAT's error message is slightly clearer when a
2206 non-printing character is used in a format directive.
2207 * fixed several bugs in compiler checking of type declarations, i.e.
2208 violations of the Python "declarations are assertions" principle
2209 (thanks to Alexey Dejneka)
2210 * fixed several bugs in PCL's error checking (thanks to Gerd
2212 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2214 * fixed bug in compilation of functions as first class values
2215 (thanks to Antonio Martinez-Shotton)
2216 * The compiler's handling TYPE-ERRORs which it can prove will
2217 inevitably happen at runtime has been cleaned up and corrected
2218 in several ways. (thanks to Alexey Dejneka)
2219 * improved argument type checking for various basic arithmetic
2220 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2221 lavished on them in the past that they can be compiled in many
2222 ways in different special cases
2223 * fixed bug 181: compiler checks validity of user supplied type
2225 * cleaned up code flushing in optimization: Function calls which
2226 should signal errors for safety purposes (e.g. which ANSI says
2227 should signal errors when their arguments are of incorrect type)
2228 are no longer optimized away.
2229 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2230 * incremented fasl file version number, because changes in the
2231 implementation of sequence functions like COERCE caused
2232 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2233 in old inline expansions) to become undefined. (Actually these
2234 changes were later undone, so we might very well be binary
2235 compatible with 0.7.7 after all, but leaving the version number
2236 incremented seemed like the simplest and most conservative
2239 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2240 * An alpha-quality port to the parisc architecture running Linux,
2241 based on the old CMUCL backend, has been made. This, even more so
2242 than the other backends, should be considered still a work in
2243 progress; known problems include that the Linux kernel in 64-bit
2244 mode does not propagate the correct sigcontext structure to
2245 userspace, and consequently SBCL on a parisc64 kernel will not
2247 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2248 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2249 suppressed.) Also now that the compiler is looking at declarations
2250 in the environment, it checks optimization declarations as well,
2251 and suppresses inlining when (> DEBUG SPEED).
2252 * More fixes have been made to treatment of floating point exception
2253 treatment and other Unix signals. In particular, floating point
2254 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2255 * The detection and handling of control stack exhaustion (infinite
2256 or very deeply nested recursion) has changed. Stack exhaustion
2257 detection is now done by write-protecting pages at the OS level
2258 and applies at all optimization settings; when found, a
2259 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2260 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2261 be caught using IGNORE-ERRORS.
2262 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2263 correctly when transforming recursive LABELS functions to LETs.
2264 (thanks to Alexey Dejneka)
2265 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2266 that are names of constants or global variables.
2267 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2268 alien routines with docstrings.
2269 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2270 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2272 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2273 (thanks to Eric Marsden)
2274 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2275 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2276 * Bug fix: The SPARC backend can now compile functions involving
2277 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2278 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2279 to a functions expecting a PATHNAME-DESIGNATOR.
2280 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2281 lambda lists. (thanks to Alexey Dejneka)
2282 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2283 functions so that EXPORTing the name of the class doesn't cause
2284 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2285 to Antonio Martinez for reporting this)
2286 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2287 constant-folded without failing an assertion. (thanks to Einar
2288 Floystad Dorum for reporting this)
2289 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2290 types are now dealt with more correctly by the compiler.
2291 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2292 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2293 behaviour analogous to RENAME-FILE. This puts its behaviour more
2294 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2296 * The fasl file version number has changed again. (because of the
2297 bug fix involving the names of PCL MAKE-INSTANCE functions)
2299 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2300 * bug fix: Floating point exceptions are treated much more
2301 consistently on the x86/Linux and PPC/Linux platforms.
2302 * Array initialization with :INITIAL-ELEMENT is now much faster for
2303 cases when the compiler cannot open code the array creation, but
2304 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2305 array accesses have also seen a speed increase.
2306 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2307 specified. (This was at the root of some bad interactions between
2308 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2310 * bug fix: Internal error arguments for undefined functions are now
2311 computed correctly on the PPC/Linux platform.
2312 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2313 Raymond Toy's patch for CMU CL)
2314 * Support for the Solaris 9 operating environment has been included
2315 (thanks to Daniel Merritt)
2316 * A very ugly but hopefully complete draft of the missing FFI chapter
2317 of the manual has been created by reformatting the corresponding
2318 CMU CL manual chapter into (currently very ugly and incoherent)
2319 DocBook and bringing it up to date for SBCL behavior. Thus, the
2320 manual is now essentially complete, at least by my extreme
2321 once-and-only-once standards, whereby it's acceptable to refer to
2322 the doc strings of SB-EXT functions as the primary documentation.
2323 * The fasl file version number has changed again, due to cleanup of
2324 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2325 is no longer a static symbol.)
2327 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2328 * SBCL now builds with OpenMCL (version 0.12) as the
2329 cross-compilation host; also, more progress has been made toward
2330 bootstrapping under CLISP.
2331 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2333 * bug 158 fixed: The compiler can now deal with integer loop
2334 increments different from 1; fixing this turned out also to fix
2336 * bug 169 fixed: no more bogus warnings about using lexical bindings
2337 despite the presence of perfectly good SPECIAL declarations (thanks
2338 to David Lichteblau)
2339 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2340 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2341 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2343 * bug 180 fixed: Method combination specifications no longer ignore
2344 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2345 * bug fix: Structure type predicate functions now check their argument
2346 count as they should.
2347 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2348 correctly. (thanks to Pierre Mai)
2349 * minor incompatible change: The --noprogrammer option is deprecated
2350 in favor of the new --disable-debugger option, which is very similar.
2351 (The major difference is that it takes effect at a slightly different
2352 time at startup, causing handling of errors in --sysinit and
2353 --userinit files will be affected differently.) The
2354 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2355 been added to allow this functionality to be controlled from ordinary
2356 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2357 maintainers, who might want to run non-interactive scripts to
2358 build SBCL cores which will later be used interactively.)
2359 * minor incompatible change: The LOAD function no longer, when given
2360 a wild pathname to load, loads all files matching that pathname.
2361 Instead, an error of type FILE-ERROR is signalled.
2363 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2364 * bug 147 fixed: The compiler preserves its block link/count
2365 invariants more correctly now so that it doesn't crash. (thanks
2367 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2369 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2370 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2371 no-value-for-this-character as NIL (as specified by ANSI).
2372 * HOST-NAMESTRING on physical pathnames now returns a string that is
2373 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2374 (thanks to Christophe Rhodes)
2375 * The Alpha port handles icache flushing more correctly. (thanks to
2377 * More progress has been made toward bootstrapping under CLISP. (thanks
2378 to Christophe Rhodes)
2379 * The fasl file format has changed again, because dynamic loading
2380 on OpenBSD (which has non-ELF object files) motivated some cleanups
2381 in the way that foreign symbols are transformed and passed around.
2382 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2383 is no longer treated as whitespace by the reader, but instead as
2384 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2385 |A
\7fB|, instead of A as it used to.
2387 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2388 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2389 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2390 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2391 this as of 0.7.1.45, but was left out of the previous news section
2392 (thanks to Dan Barlow)
2393 * SBCL now runs on the Solaris operating system on SPARC architectures
2394 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2395 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2396 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2397 * SPARC backend cleanups, allowing builds of cores optimized for V8
2398 and V9 SPARCS, and also emission of code targeted to a particular
2399 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2401 * SBCL is closer to bootstrapping under CLISP, thanks to various
2402 fixes by Christophe Rhodes.
2403 * The fasl file format has changed again, to allow the compiler's
2404 INFO database to support symbol macros.
2405 * The user manual (in doc/) is formatted into HTML more nicely.
2406 (thanks to coreythomas)
2407 * The system is smarter about SUBTYPEP relationships, especially
2408 those involving NOT types (including types such as ATOM which are
2409 represented internally using NOT types). Thus SUBTYPEP is less
2410 likely to return (VALUES NIL NIL) in general, and in particular
2411 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2413 * The fasl file format has changed again, because the internal
2414 representation of types now includes a new slot to support the new
2415 SUBTYPEP-of-NOT-types logic.
2416 * (not a change in the main branch of SBCL, but a related prototype
2417 which can hopefully be merged into the main branch of SBCL in the
2418 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2419 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2420 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2421 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2422 dispatch macro character combinations. (thanks to Alexey Dejneka)
2423 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2424 that superficially look like logical namestrings correctly.
2425 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2427 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2428 default supplied-p) for &optional and &key arguments. (thanks to
2431 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2432 * incompatible change: The compiler is now less aggressive about
2433 tail call optimization, doing it only when (> SPACE DEBUG) or
2434 (> SPEED DEBUG). (This is an incompatible change because there are
2435 programs which relied on the old CMU-CL-style behavior to optimize
2436 away their unbounded recursion which will now die of stack overflow.)
2437 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2438 for non-GENCGC systems has been increased to 20M (since that
2439 seems much closer to the likely performance optimum for modern
2440 systems than the old 4M value was)
2441 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2442 and *DEBUG-PRINT-LEVEL*
2443 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2444 of CMU CL's support for SPARC, and various endianness and other
2445 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2446 * new syntactic sugar for the Unix command line: --load foo.bar is now
2447 an alternate notation for --eval '(load "foo.bar")'.
2449 ** The system now detects stack overflow and handles it gracefully,
2450 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2451 optimization settings. (This is a good thing in general, and
2452 its introduction in this version should be particularly timely
2453 for anyone whose code fails because of suppression of tail
2455 ** The system now hunts for the C variable "environ" in a more
2456 devious way, to avoid segfaults when the C library version differs
2457 between compile time and run time. (thanks to Christophe Rhodes)
2458 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2459 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2460 ** The compiler no longer issues bogus style warnings for undefined
2461 classes in the same source file as the DEFCLASSes which defined
2462 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2464 ** fixes in CONDITION class precedence list for undefined function
2465 errors (thanks to Alexei Dejneka)
2466 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2467 correctly. (thanks to Dan Barlow)
2468 ** portability fixes aiming at bootstrapping under CLISP (thanks
2469 to Dave McDonald and Christophe Rhodes)
2470 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2471 ** fixes in type translation and and type inference (thanks to
2473 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2474 ** various fixes in the new ports (thanks to Dan Barlow)
2475 * several changes related to debugging:
2476 ** suppression of tail recursion, as noted above
2477 ** stack overflow detection, as noted above
2478 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2479 is now the default. (For some time encapsulation has been more
2480 reliable than the breakpoint-based :ENCAPSULATE NIL
2481 implementation, at least on X86 systems; and I just noticed that
2482 encapsulation also seems closer to the spirit of the ANSI
2485 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2487 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2488 up properly again. (There was a packaging bug in 0.7.0 which
2489 left their definitions in SB-SYS::LOAD-FOREIGN and
2490 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2491 vital for most things which interface to C-level interfaces,
2492 like extensions working with sockets or databases or
2493 Perl-compatible regexes or whatever, and the need to fix
2494 this bug is the main reason that 0.7.1 was released so
2496 ** DEFGENERIC is now choosier about the methods it redefines, so that
2497 reLOADing a previously-LOADed file containing DEFGENERICs does
2498 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2499 cycle now works as it should. (thanks to Alexey Dejneka)
2500 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2501 fixed by Christophe Rhodes. (He actually submitted this patch
2502 months ago, and I delayed until after 0.7.0.)
2503 ** Bug 111 (internal compiler confusion about runtime checks on
2504 FUNCTION types) was fixed by Alexey Dejneka.
2505 * Some internal cleanups (getting rid of variables which aren't
2506 needed now that the byte interpreter is gone) caused the fasl
2507 file format number to change again.
2509 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2510 * major incompatible change: The default fasl file extension, i.e. the
2511 default extension for files produced by COMPILE-FILE, has changed
2512 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2514 ** There are many changes in the implementation of the compiler.
2515 SBCL is now essentially a compiler-only implementation of ANSI
2516 Common Lisp. EVAL still "interprets" a few special cases, but
2517 almost all the interesting cases are handled by creating
2518 a LAMBDA expression, calling COMPILE on it, then calling
2519 FUNCALL on the result.
2520 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2521 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2522 Since the code is newer, there might still be some new bugs
2523 (though not as many as before Martin Atzmueller's fixes:-). But
2524 the new code is substantially simpler and clearer, and hopefully
2525 any remaining bugs will be simpler, less fundamental, and more
2526 fixable then the bugs in the old code.
2527 ** The revised compiler is still a little unsteady on its feet.
2529 *** The debugging information it produces (particularly the names
2530 of FUNCTION objects) is sometimes much less useful than what
2531 the old compiler produced.
2532 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2533 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2534 within a MACROLET) has been temporarily weakened.
2535 ** There are new compiler optimizations for various functions:
2536 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2537 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2538 *** the math functions TRUNCATE, FLOOR, and CEILING
2539 *** the function-of-all-trades COERCE
2540 Mostly these should be transparent, but there's one
2541 potentially-annoying problem (bug 117): when the compiler
2542 inline-expands a function and does type analysis on the result,
2543 it can create control paths which have type mismatches, and
2544 when it can't prove that those control paths aren't taken,
2545 it will issue WARNINGs about the type mismatches. This is
2546 a particular problem in practice for the new sequence functions.
2547 It's not clear how this should be fixed, and for now, a
2548 workaround is given in the entry for 117 in the BUGS file.
2549 ** (Because of the interaction between the two previous items --
2550 occasional inlining problems and new inline expansions -- some
2551 of the new sequence function optimizations won't really kick in
2552 completely until debugging information, and then inlining, are
2553 straightened out in some future version.)
2554 * minor incompatible changes:
2555 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2556 DIRECTORY no longer implicitly promotes NIL slots of its
2557 pathname argument to :WILD. In particular, when you ask for the
2558 contents of a directory (which you used to be able to do without
2559 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2560 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2561 ** changes in behavior that ANSI explicitly defines to be
2562 implementation dependent:
2563 *** The new compiler-only implementation still conforms with ANSI,
2564 but acts a little different than before. Besides the obvious
2565 changes in performance tradeoffs (that the cost per form passed
2566 to EVAL has gone up, and the cost per form executed by EVAL
2567 has gone down), the behavior of the system changes a little
2568 because there are no longer any interpreted function objects.
2569 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2570 e.g. doing COMPILE on the output of interactive DEFUN is
2572 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2574 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2575 has changed from (:CL) to NIL.
2576 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2577 e.g. #\Newline and #\Tab, don't have names specified in the
2578 ANSI Common Lisp standard, is now based on their ASCII symbolic
2579 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2580 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2581 are no longer used for output.
2582 ** changes in internal implementation constants:
2583 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2584 4 million. (If your application spends a lot of time GCing and
2585 you have a lot of RAM, you might want to experiment with
2586 increasing it even more.)
2587 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2588 However, almost all old code should still continue to work without
2589 immediate update, as SB-C-CALL is now a (deprecated) nickname
2591 ** Old operator names in the style DEF-FOO are now deprecated in
2592 favor of new corresponding names DEFINE-FOO, for consistency with
2593 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2594 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2595 internal symbols, but a few supported extensions like
2596 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2597 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2598 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2599 etc. as you get deeper into recursive calls to the debugger
2600 command loop, instead of the old "5]", "5]]", "5]]]"
2601 sequence. (I was motivated to do this when squabbles between
2602 ILISP and SBCL left me very deeply nested in the debugger. In the
2603 short term, this change will probably provoke more ILISP/SBCL
2604 squabbles, but hopefully it will be an improvement in the long run.)
2605 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2606 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2607 function, since declaiming return types involving aliens is
2608 (1) annoyingly messy to do by hand and (2) vital to efficient
2609 compilation of code which calls such functions.
2610 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2611 longer reexported by the SB-EXT package. They're solely useful
2612 for alien code, so it seems more logical that you should get
2613 them from the SB-ALIEN package, not in SB-EXT.
2614 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2615 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2616 features. Instead, the code that they used to control is always
2617 built into the system.
2618 * many other bug fixes
2619 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2620 advantage of the new EVAL-WHEN stuff and to clean them up in
2621 general, and they are now more ANSI-compliant in a number of
2622 ways. Martin Atzmueller is responsible for a lot of this.
2623 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2625 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2626 *** correct ERROR type for various file operations
2627 *** some fixes for Lisp streams
2628 *** DEFMETHOD syntax checking
2629 *** changing old weird representation of debug information as
2630 strings (which, among their other deficiencies, don't transform
2631 correctly when you rename packages, and don't change their
2632 print representation when you change things like *PACKAGE*
2633 and *PRINT-LENGTH*) to symbols and lists of symbols
2634 He also made several improvements and fixed several bugs in DESCRIBE.
2635 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2637 *** misbehavior of WRITE-STRING/WRITE-LINE
2638 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2639 and several other LOOP problems as well
2640 *** DIRECTORY when similar filenames are present
2641 *** DEFGENERIC with :METHOD options
2642 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2643 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2644 *** argument ordering in FIND with :TEST option
2645 *** mishandled package designator argument in APROPOS-LIST
2646 *** various problems in the backquote readmacro
2647 *** a bug in APROPOS
2648 *** probably some others that I'm not describing very well here,
2649 since the CVS log documents them by reference to sbcl-devel
2650 messages, and the SourceForge archives aren't working well.:-(
2651 ** Dan Barlow improved the Alpha port (and is making progress on the
2652 PPC port, for those of you who think different).
2653 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2654 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2655 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2656 continues to work on the SPARC port (for those of you in a position
2657 to look down upon our little PC-compatible boxes from a great height).
2658 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2659 and uses that copy, rather than the current dynamic value, when
2660 it's trying to decide whether to truncate output. Thus e.g.
2661 (let ((*print-lines* 50))
2662 (pprint-logical-block (stream nil)
2664 (let ((*print-lines* 8))
2665 (print (aref possiblybigthings i) stream)))))
2666 should now truncate the logical block only at 50 lines, instead of
2667 often truncating it at 8 lines, as it did before.
2668 * The doc/cmucl/ directory, containing old CMU CL documentation
2669 from the time of the fork, is no longer part of the base system.
2670 SourceForge has shut down its anonymous FTP service, and with it
2671 my original plan for distributing the old CMU CL documentation
2672 there. For now, if you need these files you can download an old
2673 SBCL source release and extract them from it.
2674 * The fasl file version number changed again, for dozens of reasons,
2675 some of which are apparent above.
2677 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2678 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2679 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2680 patch, so that DISASSEMBLE output is much nicer.
2681 * The code in the SB-PROFILE package now seems reasonably stable.
2682 I still haven't decided what the final interface should look like
2683 (I'd like PROFILE to interact cleanly with TRACE, since both
2684 facilities use function encapsulation) but if you have a need
2685 for profiling now, you can probably use it successfully with
2686 the current CMU-CL-style interface.
2687 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2688 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2689 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2690 link now returns the dangling link itself, and for similar
2691 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2692 link itself. (In these cases the old code signalled an error and
2693 looped endlessly, respectively.) Thus, DIRECTORY now works even
2694 in the presence of dangling and cyclic symbolic links.
2695 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2696 is now a supported extension again, since the consensus on
2697 sbcl-devel was that it can be useful for ordinary development
2698 work, not just for debugging SBCL itself.
2699 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2700 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2701 that currently-defined functions might be redefined later with
2702 different return types.
2703 * Hash tables can be printed readably, as inspired by CMU CL code
2704 of Eric Marsden and SBCL code of Martin Atzmueller.
2705 * better error handling in CLOS method combination, thanks to
2706 Martin Atzmueller porting Pierre Mai's CMU CL patches
2707 * more overflow fixes for >16Mbyte I/O buffers
2708 * A bug in READ has been fixed, so that now a single Ctrl-D
2709 character suffices to cause end-of-file on character streams.
2710 In particular, now you only need one Ctrl-D at the command
2711 line (not two) to exit SBCL.
2712 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2714 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2715 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2716 does the right thing.
2717 * The implementation of some type tests, especially for CONDITION
2718 types, is now tidier and maybe faster, due to CMU CL code
2719 originally by Douglas Crosher, ported by Martin Atzmueller.
2720 * Some math functions have been fixed, and there are new
2721 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2722 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2723 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2724 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2725 likely to be built into sbcl-0.7.0.)
2726 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2727 CL, ported to SBCL by Martin Atzmueller)
2728 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2729 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2730 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2731 * There's a new slam.sh hack to shorten the edit/compile/debug
2732 cycle for low-level changes to SBCL itself, and a new
2733 :SB-AFTER-XC-CORE target feature to control the generation of
2734 the after-xc.core file needed by slam.sh.
2735 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2736 COMPILE-FILE is no longer supported, so that now every function
2737 gets an entry point, so that block compilation looks a little
2738 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2739 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2740 returns the number of bytes consed since the system started,
2741 rather than the number consed since the first time the function
2742 was called. (The new definition parallels ANSI functions like
2743 CL:GET-INTERNAL-RUN-TIME.)
2744 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2745 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2746 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2747 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2748 * Fasl file version numbers are now independent of the target CPU,
2749 since historically most system changes which required version
2750 number changes have affected all CPUs equally. Similarly,
2751 the byte fasl file version is now equal to the ordinary
2754 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2755 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2756 is no longer recognized. I apologize for this, because it was
2757 listed in SB-EXT as a supported extension, but I found that
2758 its existing behavior was poorly specified, as well as incorrectly
2759 specified, and it looked like too much of a mess to straighten it
2760 out. I have enough on my hands trying to get ANSI stuff to work..
2761 * many patches ported from CMU CL by Martin Atzmueller, with
2762 half a dozen bug fixes in pretty-printing and the debugger, and
2763 half a dozen others elsewhere
2764 * fixed bug 13: Floating point infinities are now supported again.
2765 They might still be a little bit flaky, but thanks to bug reports
2766 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2767 as flaky as they were.
2768 * The --noprogrammer command line option is now supported. (Its
2769 behavior is slightly different in detail from what the old man
2770 page claimed it would do, but it's still appropriate under the
2771 same circumstances that the man page talks about.)
2772 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2773 are now supported, and enabled by default. Thus, the compiler can
2774 handle many floating point and complex operations much less
2775 inefficiently. (Thus e.g. you can implement a complex FFT
2777 * The compiler now detects type mismatches between DECLAIM FTYPE
2778 and DEFUN better, and implements CHECK-TYPE more correctly, and
2779 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2780 thanks to patches from Martin Atzmueller.
2781 * various fixes to make the cross-compiler more portable to
2782 ANSI-conforming-but-different cross-compilation hosts (notably
2783 Lispworks for Windows, following bug reports from Arthur Lemmens)
2784 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2785 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2786 fixed thanks to Pierre Mai's CMU CL patches.
2787 * fixes to keep the system from overflowing internal counters when
2788 it tries to use i/o buffers larger than 16M bytes
2789 * fixed bug 45a: Various internal functions required to support
2790 complex special functions have been merged from CMU CL sources.
2791 (When I was first setting up SBCL, I misunderstood a compile-time
2792 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2793 * improved support for type intersection and union, fixing bug 12
2794 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2795 more obscure bugs as well
2796 * some steps toward byte-compiling non-performance-critical
2797 parts of the system, courtesy of patches from Martin Atzmueller
2798 * Christophe Rhodes has made some debian packages of sbcl at
2799 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2800 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2801 stable, but are nonetheless usable. When he's ready, I'd be happy
2802 to add them to the SourceForge "File Releases" section. (And if
2803 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2804 * new fasl file format version number (because of changes in
2805 internal representation of (OR ..) types to accommodate the new
2806 support for (AND ..) types, among other things)
2808 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2809 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2811 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2813 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2814 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2815 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2816 * DESCRIBE now works on structure objects again.
2817 * Most function call argument type mismatches are now handled as
2818 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2819 know whether the function will be redefined before the call is
2820 executed. (The compiler could flag local calls with full WARNINGs,
2821 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2822 it doesn't keep track of enough information to know whether calls
2823 are local in this sense.)
2824 * Compiler output is now more verbose, with messages truncated
2825 later than before. (There should be some supported way for users
2826 to override the default verbosity, but I haven't decided how to
2827 provide it yet, so this behavior is still controlled by the internal
2828 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2829 src/compiler/ir1util.lisp.)
2830 * Fasl file format version numbers have increased again, because
2831 support for the Gray streams extension changes the layout of the
2832 system's STREAM objects.
2833 * The Gray subclassable streams extension now works, thanks to a
2834 patch from Martin Atzmueller.
2835 * The full LOAD-FOREIGN extension (not just the primitive
2836 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2837 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2838 but like most other programs, it defaults to copying the Unix
2839 environment from the original process instead of starting the
2840 new process in an empty environment.
2841 * Extensions which manipulate the Unix environment now support
2842 an :ENVIRONMENT keyword option which doesn't smash case or
2843 do other bad things. The CMU-CL-style :ENV option is retained
2844 for porting convenience.
2845 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2846 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2848 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2850 * A patch from Martin Atzmueller seems to have solved the SIGINT
2851 problem, and as far as we know, signal-handling now works cleanly.
2852 (If you find any new bugs, please report them!)
2853 * The system no longer defaults Lisp source file names to types
2854 ".l", ".cl", or ".lsp", but only to ".lisp".
2855 * The compiler no longer uses special default file extensions for
2856 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2857 seems to expect a single default extension for all compiled code,
2858 and there's no compelling reason to try to stretch the standard
2859 to allow two different extensions.) Instead, byte-compiled files
2860 default to the same extension as native-compiled files.
2861 * Fasl file format version numbers have increased again, because
2862 a rearrangement of internal implementation packages made some
2863 dumped symbols in old fasl files unreadable in new cores.
2864 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2865 many fewer weird special cases.
2866 * Bug #17 (differing COMPILE-FILE behavior between logical and
2867 physical pathnames) has been fixed, and some related misbehavior too,
2868 thanks to a patch from Martin Atzmueller.
2869 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2870 by Tim Moore, ported to SBCL by Martin Atzmueller.
2871 * Martin Atzmueller fixed several filesystem-related problems,
2872 including bug #36, in part by porting CMU CL patches, which were
2873 written in part by Paul Werkowski.
2874 * More compiler warnings in src/runtime/ are gone, thanks to
2875 more patches from Martin Atzmueller.
2876 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2879 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2881 * DESCRIBE now works on CONDITION objects.
2882 * The debugger now handles errors which arise when trying to print
2883 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2885 * The build system now uses an additional file, customize-target-features.lisp,
2886 to allow local modifications to the target *FEATURES* list. (The point of
2887 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2888 debugging features enabled, without having to worry about propagating it
2889 into everyone's system when I do a "cvs update".) When no
2890 customize-target-features.lisp file exists, the target *FEATURES* list
2891 should be constructed the same way as before.
2892 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2893 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2894 warn about behavior which is undefined under the ANSI spec. Note: This
2895 is specified by ANSI, but it's not very popular with programmers.
2896 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2897 macro in the SBCL sources for an example of a workaround which you
2898 might use to make portable ANSI-standard code which does what you want.
2899 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2900 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2901 when it's used as a non-toplevel form. (This is required in order
2902 to implement the DEFCONSTANT-EQX macro.)
2903 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2904 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2905 funny things, due to bugs in the implementation of EVAL-WHEN
2906 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2907 non-toplevel DEFCONSTANTs are uncommon.)
2908 * The core file version number and fasl file version number have been
2909 incremented, because the old noncompliant DEFCONSTANT behavior involved
2910 calling functions which no longer exist, and because I also took the
2911 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2912 * fixed bug 1 (error handling before read-eval-print loop starts), and
2913 redid debugger restarts and related debugger commands somewhat while
2915 ** The QUIT debugger command is gone, since it did something
2916 rather different than the SB-EXT:QUIT command, and since it never
2917 worked properly outside the main toplevel read/eval/print loop.
2918 Invoking the new TOPLEVEL restart provides the same functionality.
2919 ** The GO debugger command is also gone, since you can just invoke
2920 the CONTINUE restart directly instead.
2921 ** The TOP debugger command is also gone, since it's redundant with the
2922 FRAME 0 command, and since it interfered with abbreviations for the
2924 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2926 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2927 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2928 some porting work by Martin Atzmueller)
2929 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2930 with a boatload of patches which he ported to SBCL. Now that those
2932 ** The system tries to make sure that its low-priority messages
2933 are prefixed by semicolons, to help people who like to use
2934 syntax highlighting in their ILISP buffer. (This patch
2935 was originally due to Raymond Toy.)
2936 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2937 patches originally written by Raymond Toy.
2938 ** The compiler understands coercion between single-value and
2939 multiple-VALUES type expressions better, getting rid of some very
2940 weird behavior, thanks to patches originally by Robert MacLachlan
2941 and Douglas Crosher.
2942 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2943 lambda lists, thanks to a patch originally by Pierre Mai.
2944 ** The system no longer bogusly warns about "abbreviated type
2946 ** The compiler gets less confused by inlining and RETURN-FROM,
2947 thanks to some patches originally by Tim Moore.
2948 ** The system no longer hangs when dumping circular lists to fasl
2949 files, thanks to a patch originally from Douglas Crosher.
2950 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2951 undefined function error.
2952 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2953 in structure slot initforms) for now, documented workaround instead:-|
2954 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2955 * fixed bug 5: added stubs for various Gray stream functions called
2956 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2957 installed, at least appropriate type errors are generated
2958 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2959 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2960 consistently in DEFMETHOD forms.
2961 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2962 it doesn't seem to affect SBCL after all
2963 * The C runtime system now builds with better optimization and many
2964 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2966 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2968 * The system is now under CVS at SourceForge (instead of the
2969 CVS repository on my home machine).
2970 * The new signal handling code has been tweaked to treat register
2971 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2972 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2973 downstream, e.g. in debug-int.lisp, has implicit dependencies
2974 on the unsignedness of integer representation of machine words,
2975 and that caused the system to bomb out with infinite regress
2976 when trying to recover from type errors involving signed values,
2977 e.g. (BUTLAST '(1 2 3) -1).)
2978 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2979 (This was one of the bugs Peter Van Eynde reported back in July.)
2980 * The system now uses code inspired by Colin Walters' O(N)
2981 implementation of MAP (from the cmucl-imp@cons.org mailing
2982 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2983 inline the MAP operation, and there is more than one
2984 sequence argument to the MAP call (so that it can't just
2985 do ETYPECASE once and for all based on the type of the
2986 single sequence argument). (The old non-inline implementation
2987 of the general M-argument sequence-of-length-N case required
2988 O(M*N*N) time when any of the sequence arguments were LISTs.)
2989 * The QUIT :UNIX-CODE keyword argument has been renamed to
2990 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2991 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2993 * Raymond Wiker's patches to port dynamic loading from Linux to
2994 FreeBSD have been added.
2995 * The BUGS file is now more nearly up to date, thanks in large part
2996 to Martin Atzmueller's review of it.
2997 * The debugger now flushes standard output streams before it begins
2998 its output ("debugger invoked" and so forth).
2999 * The core version number and fasl file version number have both
3000 been incremented, because of incompatible changes in the layout
3002 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3003 used to not be called for a saved Lisp image.)
3004 * Martin Atzmueller's version of a patch to fix a compiler crash,
3005 as posted on sbcl-devel 13 September 2000, has been installed.
3006 * Instead of installing Martin Atzmueller's patch for the
3007 compiler transform for SUBSEQ, I deleted the compiler transform,
3008 and transforms for some similar consing operations.
3009 * A bug in signal handling which kept TRACE from working on OpenBSD
3011 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3012 away by constant folding
3013 * The system now defines its address space constants in one place
3014 (in the Lisp sources), and propagates them automatically elsewhere
3015 (through GENESIS and the sbcl.h file). Therefore, patching the
3016 address map is less unnecessarily tedious and error-prone. The
3017 Lisp names of address space constants have also been systematized.
3018 * CVS tags like dollar-Header-dollar have been removed from
3019 the sources, because they have never saved me trouble and
3020 they've been source of trouble working with patches and other
3021 diff-related operations.
3022 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3023 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3025 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3027 * The system has been ported to OpenBSD.
3028 * The system now compiles with a simple "sh make.sh" on the systems
3029 that it's supported on. I.e., now you no longer need to tweak
3030 text in the target-features.lisp-expr and symlinks in src/runtime/
3031 by hand, the make.sh takes care of it for you.
3032 * The system is no longer so grossly inefficient when compiling code
3033 involving vectors implemented as general (not simple) vectors (VECTOR T),
3034 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3035 which dares to use the various sequence functions on non-simple
3036 vectors, takes less of a performance hit.
3037 * There is now a primitive type predicate VECTOR-T-P
3038 to test for the (VECTOR T) type, so that e.g.
3039 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3040 can now be compiled with some semblance of efficiency. (The old code
3041 turned the type declaration into a full call to %TYPEP at runtime!)
3042 * AREF on (VECTOR T) is still not fast, since it's still compiled
3043 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3044 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3045 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3046 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3047 without making full calls to SUBTYPEP at runtime.
3048 (Some analogous efficiency issues for non-simple vectors specialized to
3049 element types other than T, or for non-simple multidimensional arrays,
3050 have not been addressed. They could almost certainly be handled the
3051 same way if anyone is motivated to do so.)
3052 * The changes in array handling break binary compatibility, so
3053 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3054 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3055 instead of (VALUES T T).
3056 * By following the instructions that Dan Barlow posted to sbcl-devel
3057 on 2 July 2000, I was able to enable primitive dynamic object
3058 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3059 functionality is not implemented (since it calls ld to resolve
3060 library references automatically, requiring RUN-PROGRAM for its
3061 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3062 try to resolve library references) is now supported.
3063 * The system now flushes the standard output streams when it terminates,
3064 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3065 them at several other probably-convenient times, e.g. in each pass of
3066 the toplevel read-eval-print loop, and after evaluating a form given
3067 as an "--eval" command-line option. (These changes were motivated by a
3068 discussion of stream flushing issues on cmucl-imp in August 2000.)
3069 * The source transform for TYPEP of array types no longer assumes
3070 that an array whose element type is a not-yet-defined type
3071 is implemented as an array of T, but instead punts, so that the
3072 type will be interpreted at runtime.
3073 * There is now some support for cross-compiling in make.sh: each of
3074 the phases of make.sh has its own script. (This should be transparent
3075 to people doing ordinary, non-cross-compile builds.)
3076 * Since my laptop doesn't have hundreds of megabytes of memory like
3077 my desktop machine, I became more motivated to do some items on
3078 my to-do list in order to reduce the size of the system a little:
3079 ** Arrange for various needed-only-at-cold-init things to be
3080 uninterned after cold init. To support this, those things have
3081 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3082 symbols with such names are now uninterned after cold init).
3083 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3084 fasl files for cold load.
3085 ** Remove the old compiler structure pooling code (which used to
3086 be conditional on the target feature :SB-ALLOC) completely.
3087 ** Redo the representation of some data in cold init to be more compact.
3088 (I also looked into supporting byte compiled code at bootstrap time,
3089 which would probably reduce the size of the system a lot, but that
3090 looked too complicated, so I punted for now.)
3091 * The maximum signal nesting depth in the src/runtime/ support code has
3092 been reduced from 4096 to 256. (I don't know any reason for the very
3093 large old value. If the new smaller value turns out to break something,
3094 I'll probably just bump it back up.)
3095 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3097 * Many, many bugs reported by Peter Van Eynde have been added to
3098 the BUGS list; some have even been fixed.
3099 * While enabling dynamic object file loading, I tried to make the
3100 code easier to understand, renaming various functions and variables
3101 with less ambiguous names, and changing some function calling
3102 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3103 * While trying to figure out how to do the OpenBSD port, I tried to
3104 clean up some of the code in src/runtime/. In particular, I dropped
3105 support for non-POSIX signal handling, added various comments,
3106 tweaked the code to reduce the number of compilation warnings, and
3107 renamed some files to increase consistency.
3108 * To support the new automatic configuration functionality in make.sh,
3109 the source file target-features.lisp-expr has been replaced with the
3110 source file base-target-features.lisp-expr and the machine-generated
3111 file local-target-features.lisp-expr.
3112 * fixed a stupid quoting error in make.sh so that using CMU CL
3113 "lisp -batch" as cross-compilation host works again
3115 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3117 * DESCRIBE no longer tries to call itself recursively to describe
3118 bound/fbound values, so that it no longer fails on symbols which are
3119 bound to themselves (like keywords, T, and NIL).
3120 * DESCRIBE now works on generic functions.
3121 * The printer now prints less-screwed-up representations of closures
3122 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3123 * A private symbol is used instead of the :EMPTY keyword previously
3124 used to mark empty slots in hash tables. Thus
3125 (DEFVAR *HT* (MAKE-HASH-TABLE))
3126 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3127 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3128 now does what ANSI says that it should. (You can still get
3129 similar noncompliant behavior if bang on the hash table
3130 implementation with all the symbols you get back from
3131 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3132 This breaks binary compatibility, since tests for equality to
3133 :EMPTY are wired into things like the macroexpansion of
3134 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3136 * There's now a minimal placeholder implementation for CL:STEP,
3137 as required by ANSI.
3138 * An obscure bug in the interaction of the normal compiler, the byte
3139 compiler, inlining, and structure predicates has been patched
3140 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3141 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3143 * Missing ordinary arguments in a macro call are now detected even
3144 when the macro lambda list contains &KEY or &REST.
3145 * The debugger no longer complains about encountering the top of the
3146 stack when you type "FRAME 0" to explicitly instruct it to go to
3147 the top of the stack. And it now prints the frame you request even
3148 if it's the current frame (instead of saying "You are here.").
3149 * As specified by ANSI, the system now always prints keywords
3150 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3152 * The default initial SIZE of HASH-TABLEs is now smaller.
3153 * Type information from CLOS class dispatch is now propagated
3154 into DEFMETHOD bodies, so that e.g.
3155 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3157 is now basically equivalent to
3158 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3159 (DECLARE (TYPE SINGLE-FLOAT X))
3161 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3162 operation, without having to do run-time type dispatch.
3163 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3164 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3165 * Since I don't seem to be making big file reorganizations very often
3166 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3167 is overflowing my ability to conveniently back them up), I've finally
3168 checked the system into CVS. (The CVS repository is on my home system,
3169 not at SourceForge -- putting it on SourceForge might come later.)
3170 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3171 high-level GC-NOTIFY-FOO functions send their output. (There's
3172 still very little control of where low-level verbose GC functions
3173 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3174 less than it used to -- the GC-NOTIFY-FOO functions are now under
3175 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3176 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3177 in only one place in the source code, and propagates it automatically
3178 everywhere that it's needed. Thus e.g. when I bump the version from
3179 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3180 * The C source files now include boilerplate legalese and documentation
3181 at the head of each file (just as the Lisp source files already did).
3182 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3183 to his page will be replaced with a link to his new CLiki service.
3185 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3187 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3188 * The build process now looks for GNU make under the default name "gmake",
3189 instead of "make" as it used to. If GNU make is not available as "gmake"
3190 on your system, you can change this default behavior by setting the
3191 GNUMAKE environment variable.
3192 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3193 can build without error under CMU CL.
3195 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3197 * There is now a partial SBCL user manual (with some new text and some
3198 text cribbed from the CMU CL manual).
3199 * The beginnings of a profiler have been added (starting with the
3200 CMU CL profiler and simplifying and cleaning up). Eventually the
3201 main interface should be through the TRACE macro, but for now,
3202 it's still accessed through vaguely CMU-CL-style functions and macros
3203 exported from the package SB-PROFILE.
3204 * Some problems left over from porting CMU CL to the new
3205 cross-compilation bootstrap process have been cleaned up:
3206 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3207 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3208 part when getting rid of the compiler *BACKEND* stuff.)
3209 ** The value of *NULL-TYPE* was screwed up, because it was
3210 being initialized before the type system knew the final
3211 definition of the 'NULL type. This screwed up several key
3212 optimizations in the compiler, causing inefficiency in all sorts
3213 of places. (I found it because I wanted to understand why
3214 GET-INTERNAL-RUN-TIME was consing.)
3215 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3216 PROCLAIM FTYPE information. Unfortunately this broke binary
3217 compatibility again, since now the forms output by DEFGENERIC
3218 to refer to functions which didn't exist in 0.6.3.
3219 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3220 can use the new (as of 0.6.3) transform for SOME into MAP into
3222 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3223 values to (INTEGER 0 1000000), so that the time code will no longer
3224 occasionally get blown up by Linux returning 1000000 microseconds
3225 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3226 its output conform to the ANSI spec. (Alas, this makes its output
3227 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3228 * A full call to MAP NIL with a single sequence argument no longer conses.
3229 * fixes to problems pointed out by Martin Atzmueller:
3230 * The manual page no longer talks about multiprocessing as though
3231 it were currently supported.
3232 * The ILISP support patches have been removed from the distribution,
3233 because as of version 5.10.1, ILISP now supports SBCL without us
3234 having to maintain patches.
3235 * added a modified version of Raymond Toy's recent CMU CL patch for
3236 EQUALP comparison of HASH-TABLE
3238 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3240 * The system still can't cross-compile itself with
3241 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3242 entails), but at least it can compile more of itself that way
3243 than it used to be able to, and various buglets which were uncovered
3244 by trying to cross-compile itself that way have now been fixed.
3245 * This release breaks binary compatibility again. This time
3246 at least I've incremented the FASL file format version to 2, so that the
3247 problem can be detected reliably instead of just causing weird errors.
3248 * various new style warnings:
3249 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3250 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3251 ** using the lexical binding of a variable named in the *FOO* style
3252 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3253 as specified by ANSI.
3254 * *RANDOM-STATE* is no longer automatically initialized from
3255 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3256 default behavior of the system is to repeat its behavior every time
3257 it's run. If you'd like to change this behavior, you can always
3258 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3259 old convention there was no comparably easy way to get the system to
3260 repeat its behavior every time it was run.
3261 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3262 FTYPE declarations has been removed, in favor of their ANSI
3263 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3264 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3265 the types of their sequence arguments can be determined at compile time.
3266 This is done through a new open code expansion for MAP which eliminates
3267 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3268 argument types can be determined at compile time.
3269 * The optimizer now transforms COERCE into an identity operation when it
3270 can prove that the coerced object is already of the correct type. (This
3271 can be a win for machine generated code, including the output of other
3272 optimization transforms, such as the MAP transform above.)
3273 * Credit information has been moved from source file headers into CREDITS.
3274 * Source file headers have been made more standard.
3275 * The CASE macro now compiles without complaining even when it has
3278 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3280 * (Note that the way that the PCL macroexpansions were rewritten
3281 to accommodate the change in DEFGENERIC below breaks binary
3282 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3283 not run under sbcl-0.6.2. Once we get out of alpha releases,
3284 i.e. hit release 1.0.0, we'll probably try to maintain binary
3285 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3286 and sbcl-1.4.4. Until then, however, it might be fairly common
3287 for maintenance releases to break binary compatibility.)
3288 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3290 * The reporting of unhandled signals has been changed to print some
3291 explanatory text as well as the report form. (Previously only
3292 the report form was printed.)
3293 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3294 it defines, so that the compiler no longer issues undefined function
3295 warnings for compiled-but-not-yet-loaded generic functions.
3296 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3297 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3298 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3299 section of the ANSI spec are supported.
3300 * The "" nickname for the "KEYWORD" package has been removed.
3301 The reader still handles symbol tokens which begin with a package marker
3302 as keywords, but it doesn't expose its mechanism for doing so in the
3303 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3304 * The system now issues STYLE-WARNINGs for contradictory TYPE
3305 proclamations. (Warnings for contradictory FTYPE proclamations would
3306 be nice too, but those can't be done usefully unless the type system
3307 is made smarter about FUNCTION types.)
3308 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3309 systematized, so that "*target-*.lisp is supposed to exist only on the
3310 target and imply that there's a related file which exists on the
3311 host, and *host-*.lisp is supposed to exist only on the host and imply
3312 that there's a related file which exists on the target. This involves a
3313 lot of renaming. Hopefully the acute confusion caused by the renaming
3314 will be justified by the reduction in chronic confusion..
3315 ** runtime-type.lisp -> early-target-type.lisp
3316 ** target-type.lisp -> late-target-type.lisp
3317 ** early-host-format.lisp -> early-format.lisp
3318 ** late-host-format.lisp -> late-format.lisp
3319 ** host-error.lisp -> misc-error.lisp
3320 ** early-error.lisp -> early-target-error.lisp
3321 ** late-error.lisp -> late-target-error.lisp
3322 ** host-defboot.lisp -> early-defboot.lisp
3323 ** code/misc.lisp -> code/target-misc.lisp
3324 ** code/host-misc.lisp -> code/misc.lisp
3325 ** code/numbers.lisp -> code/target-numbers.lisp
3326 ** code/early-numbers.lisp -> numbers.lisp
3327 ** early-host-type.lisp -> early-type.lisp
3328 ** late-host-type.lisp -> late-type.lisp
3329 ** host-typep.lisp -> typep.lisp
3330 ** load.lisp -> target-load.lisp
3331 ** host-load.lisp -> load.lisp
3332 ** host-disassem.lisp -> disassem.lisp
3333 ** host-insts.lisp -> insts.lisp
3334 ** byte-comp.lisp -> target-byte-comp.lisp
3335 ** host-byte-comp.lisp -> byte-comp.lisp
3336 ** host-signal.lisp -> signal.lisp
3337 ** host-defstruct.lisp -> defstruct.lisp
3338 ** late-target-type.lisp -> deftypes-for-target.lisp
3339 Furthermore, several other previously target-only files foo.lisp (e.g.
3340 hash-table.lisp and random.lisp) have been split into a target-and-host
3341 foo.lisp file and a target-only target-foo.lisp file, with their key type
3342 definitions in the target-and-host part, so that the cross-compiler will
3343 know more about target types.
3344 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3345 gone away. In their place are various *BACKEND-FOO* variables
3346 corresponding to the slots of the old structure.
3347 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3348 into the target SBCL has been fixed.
3349 * The chill.lisp system for loading cold code into a running SBCL
3351 * Support for the CMU CL "scavenger hook" extension has been removed.
3352 (It was undocumented and unused in the CMU CL sources that SBCL was
3353 derived from, and stale in sbcl-0.6.1.)
3354 * Various errors in the cross-compiler type system were detected
3355 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3356 (enabling various consistency checks). Many of them were fixed,
3357 but some hard problems remain, so the compiler is back to
3358 running without *TYPE-SYSTEM-INITIALIZED* for now.
3359 * As part of the cross-compiler type system cleanup, I implemented
3360 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3361 * I have started adding UNCROSS calls throughout the type system
3362 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3363 on cross-compiler input files will be able to go away, and various
3365 * CONSTANTP now returns true for quoted forms (as explicitly required
3368 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3370 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3371 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3372 (MAKE-LIST -1). In the longer term, it ought to become true
3373 that declarations are assertions even at SAFETY 1. For now, it's not
3374 quite true even at SAFETY 3, but it's at least more nearly true..
3375 (Note that this change seems to increases the size of the system by
3376 O(5%) and to decrease the speed of the compiler by 20% or more.)
3377 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3378 for the problem of printing dozens of lines of distracting information
3379 about low-level system machinery as part of the top stack frame
3380 on entry to the debugger when an undefined function was called.
3381 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3382 is not reset to COMMON-LISP-USER.
3383 * Compilation of stuff related to dyncount.lisp has been made conditional
3384 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3385 smaller. The various dyncount-related symbols have been moved into
3386 a new "SB-DYNCOUNT" package.
3387 * tty-inspect.lisp has been renamed to inspect.lisp.
3388 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3389 feature has gone away. (When we eventually port to other flavors of
3390 libc and/or Unix, we'll try to make the differences between flavors
3391 invisible at the user level.)
3392 * Various other *FEATURES* tags, and/or their associated conditionals,
3393 have been removed if obsolescent, or given better documentation, or
3394 sometimes given more-mnemonic names.
3396 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3398 * tidied up "make.sh" script
3399 * tidied up system directory structure
3400 * better "clean.sh" behavior
3401 * added doc/FOR-CMUCL-DEVELOPERS
3402 * many many small tweaks to output format, e.g. removing possibly-confusing
3403 trailing #\. character in DESCRIBE-INSTANCE
3404 * (EQUALP #\A 'A) no longer signals an error.
3405 * new hashing code, including EQUALP hashing
3406 * tidied up Lisp initialization and toplevel
3407 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3408 * command line argument processing
3409 * added POSIX-GETENV function to deal with Unix-ish environment variables
3410 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3411 terminating SBCL on EOF
3412 * non-verbose GC by default
3413 * There is no more "sbcl" shell script; the sbcl file is now the C
3414 runtime executable (just like CMU CL).
3415 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3417 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3418 symbol and package information as Lisp native symbol and package objects
3419 instead of strings naming symbols and strings naming packages. This way,
3420 whenever packages are renamed (as in warm init), debug information is
3421 transformed along with everything else.
3422 * tweaked the optimization policy declarations which control the building
3423 of SBCL itself. Now, among other things, the system no longer saves
3424 source location debugging information. (This helps two problems at once
3425 by reducing SBCL size and by keeping SBCL from trying to look for its
3426 sources -- which may not exist -- when reporting errors.)
3427 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3428 debugging and testing purposes
3429 * cleaned up printing, making the printer call PRINT-OBJECT for
3430 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3431 methods, giving nearly-ANSI behavior
3432 * converted almost all special variables to use *FOO* naming convention
3433 * deleted PARSE-TIME functionality, since it can be done portably
3434 * moved some files out of cold init into warm init
3435 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3437 * regularized formatting of source files
3438 * added an install.sh script
3439 * fixed ridiculous memory usage of cross-compiler by making
3440 compiler/alloc.lisp not try to do pooling unless it can hook
3441 itself into the GC of the cross-compilation host. Now the system
3442 builds nicely on my old laptop.
3443 * added :SB-ALLOC in target-features.lisp-expr
3444 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3445 was not implemented)
3446 * re-did condition handling and note reporting in the compiler. Notes
3447 are no longer handled by signalling conditions. Style warnings
3448 and warnings are handled more correctly and reported in such a way
3449 that it's easy to find one or the other in your output (so that you
3450 can e.g. figure out which of many problems caused COMPILE-FILE to
3452 * changed the severity of several compiler warnings from full WARNING
3453 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3454 compiler note reporting so that it doesn't use the condition system
3455 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3456 COMPILE-FILE command)
3457 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3458 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3459 no longer supported, since their implementation was deeply intertwingled
3460 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3461 * removed bogus "support" for compiler macros named (SETF FOO), and
3462 removed the compiler macro for SETF INFO (but only after making a fool
3463 of myself on the cmucl-imp mailing list by posting a bogus patch for
3464 DEFINE-COMPILER-MACRO..)
3465 * Compiled files containing forms which have side effects on the Lisp
3466 reader (such as DEFPACKAGE forms) are now handled more correctly.
3467 (Compiler queuing of top level lambdas has been suppressed by setting
3468 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3469 * deleted various currently-unused source files, e.g. gengc.lisp. They
3470 may be added back at some point e.g. when porting to other architectures,
3471 but until they are it's distracting to distribute them and to try to
3473 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3474 there were so many of them they're just distractions, and UNCROSS is
3475 known to be able to handle the current sources
3476 * moved PROFILE functionality into TRACE, so that it will be clear
3477 how the wrapping and unwrapping of functions when you profile them
3478 interacts with the wrapping and unwrapping of functions when you
3479 trace them. (Actually, the functionality isn't there yet, but at least
3480 the interface specification is there. Hopefully, the functionality will
3481 arrive with some future maintenance release.)
3482 * removed host-oops.lisp
3483 * changed signature of QUIT function to allow UNIX-CODE argument
3484 * fixed READ-SEQUENCE bug
3485 * tweaked verbose GC output so that it looks more like the progress
3486 output that ANSI specifies for functions like LOAD
3487 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3488 * added <http://sbcl.sourceforge.com> to the banner information printed by