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: COMPUTE-RESTARTS returns all active restarts, including those
16 with :TEST, when called without a condition. (reported by Helmut Eller for
18 * bug fix: division by zero in sb-sprof when no samples were collected
19 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
20 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
21 (thanks to Alexander Kjeldaas)
22 * new feature: ignoring the return values of destructive functions
23 like NREVERSE signals a compile-time style-warning.
24 (thanks to Kevin Reid)
25 * new feature: the alignment of alien structure fields can be
26 explicitly specified. (thanks to Cyrus Harmon)
28 ** bug fix: threads stacks belonging to dead threads are freed by the
29 next exiting thread, no need to gc to collect thread stacks anymore
30 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
31 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
32 ** bug fix: dynamic variable and thread start related gc lossage
33 * fixed some bugs revealed by Paul Dietz' test suite:
34 ** SUBTYPEP is slightly more accurate on heinously complicated
35 CONS types where some of the members have uncertain (in the
36 NIL, NIL sense) type relationships to each other.
38 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
39 * new feature: timers based on Zach Beane's excellent timer package
40 * added support for the following external formats: koi8-u,
41 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
42 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
43 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
44 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
45 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
46 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
47 * incompatible change: a threaded SBCL will no longer revert to
48 non-threaded mode on non-NPTL systems, but refuse to start entirely.
49 * bug fix: interrupts are disabled until startup is complete; no
50 more sigsegvs when receiving a signal to soon
51 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
53 * bug fix: add a workaround for the memory randomization features in
54 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
55 workaround will only be in effect on systems with the proc filesystem
57 * bug fix: printing objects of type HASH-TABLE signals a
58 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
60 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
61 have been running for over 50 days. (reported by Gilbert Baumann)
62 * bug fix: the logic for getting names of functions gets less
63 confused when confronded with alternate-metaclass
64 funcallable-instances. (reported by Cyrus Harmon)
65 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
66 consistent, even on internal alternate-metaclass objects.
67 * bug fix: flush closure information collected by physical
68 environment analysis prepass before the main pass. (bug reported
70 * bug fix: compiler pack phase does not modify a hashtable, which is
71 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
72 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
73 may be expanded. (reported by Alexander <malishev> on c.l.l)
74 * bug fix: no more style-warnings when compiling code using LAST
75 under high-SPEED low-SPACE settings. (reported by David Wallin
77 * fixed some incompatibilities between SBCL's MOP and the MOP
79 ** the METAOBJECT class is now implemented;
80 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
81 STANDARD-OBJECT, as required;
82 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
83 now compatible (as required by VALIDATE-SUPERCLASS); there
84 remains a consistency requirement over the presence or absence
85 of the FUNCTION class in the superclass list of finalized
86 classes; see the manual for more details;
87 ** the :AROUND method for COMPUTE-SLOTS on
88 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
89 requested slot ordering.
91 ** bug fix: parent thread now can be gc'ed even with a live
93 ** bug fix: binding a special with PROGV to no value is not
94 the same as that symbol not having been bound (thanks to
96 ** bug fix: binding specials is thread safe (thanks to
98 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
99 and SB-SPROF do not die with 'no handler for signal XX in
100 interrupt_handle_now(..)' anymore
101 ** bug fix: WITH-TIMEOUT works with multiple running threads
102 * fixed some bugs revealed by Paul Dietz' test suite:
103 ** ENSURE-GENERIC-FUNCTION should take a method class object for
104 the :method-class keyword argument.
106 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
107 * new port: the Solaris operating system on x86 processors is now
108 mostly supported, though some rough edges in the environment
109 remain. (thanks to Daisuke Homma)
110 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
111 space for its heap. (thanks to Thiemo Seufer)
112 * optimization: SBCL on MIPS platforms now supports dynamic-extent
113 closures. (thanks to Thiemo Seufer)
114 * minor incompatible change: eof selects abort in the debugger.
115 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
116 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
118 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
119 deprecated, and will go away in a future revision of SBCL.
120 * minor incompatible change: GC-ON and GC-OFF are no longer
121 implemented with a counter, it does not matter how many times gc
122 is switched on or off
123 * bug fix: discriminating functions for generic function classes
124 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
125 make invalid assumptions about method precedence order. (reported
127 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
128 platforms. (thanks to Thiemo Seufer)
129 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
130 * bug fix: correct pathname printing with printer escaping is on.
131 (thanks to Kevin Reid)
132 * bug fix: complex VOP definitions in "user-space" no longer trigger
133 package locks. (reported by Zach Beane)
134 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
135 for funcallable-instance functions, and is no different from
137 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
138 characters in the current readtable. (reported by Nicholas Neuss)
139 * bug fix: SUBTYPEP on various CONS types returns more a more
140 accurate acknowledgment of its certainty.
141 * optimizations: REMOVE-DUPLICATES now runs in linear time on
142 lists in some cases. This partially fixes bug 384.
143 * flush all standard streams before prompting in the REPL and the
145 * bug fix: signal handling and triggering gc do not conflict
146 directly anymore, in particular a high frequency sb-sprof does
147 not prevent gc from running
148 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
149 approximation for timezone and DST information between the
150 universal time epoch and the smallest negative 32-bit time_t.
151 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
152 year 1899 so long as the encoded time is non-negative.
153 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
154 the correct answer for non-constant rotate amounts when performing
155 an inline 32-bit rotation.
157 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
158 there is only one thread in the session
159 ** bug fix: memory leak for streams created in one thread and
160 written to in another
161 ** bug fix: lockup when compiled with gcc4
162 ** bug fix: race that allows the gc to be triggered when gc is
164 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
165 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
167 ** bug fix: finished threads are gc'ed properly
168 * fixed some bugs revealed by Paul Dietz' test suite:
169 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
170 has arguments with a different set of applicable methods from
171 the orignal arguments.
172 ** The type error thrown by MAP now has a correct expected-type
174 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
175 name a compiled function.
176 ** Three MISC tests where a large bignum was improperly coerced to
177 a float that couldn't represent that bignum during type
178 derivation were fixed.
179 ** SUBTYPEP can now handle types involving the negation of a
180 list-form FUNCTION type.
181 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
182 as (COMPLEX (AND RATIO (NOT FIXNUM))).
183 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
185 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
186 * New feature: Experimental support for bivalent streams: streams
187 opened with :element-type :default now allow character and binary
188 (unsigned-byte 8) I/O
189 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
190 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
192 * bug fix: no more highly sporadic "couldn't check whether ~S is
193 readable" when reading a stream and an interrupt hits in the middle
194 of a select system call
195 * compiler better recognizes complex arrays (reported by Eduardo
197 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
198 for variables on SBCL built with linkage-tables. (reported by Luis
200 * various error reporting improvements.
201 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
202 (Thanks to Thiemo Seufer)
203 * bug fix: nested reader invokations work correctly
204 * bug fix: it is possible to have simultaneous references to foreign
205 code and foreign data with the same name.
207 ** added x86-64 support
208 ** incompatible change: the threading api now works with thread
209 objects instead of thread ids
210 ** bug fix: threads are protected from signals and interruption when
211 starting up or going down
212 ** bug fix: a race where an exiting thread could lose its stack to gc
213 ** bug fix: don't halt on infinite error in threads if possible
214 ** fixed numerous gc deadlocks introduced in the pthread merge
215 ** bug fix: fixed thread safety issues in read and print
216 ** bug fix: debugger doesn't hang on session lock if interrupted at
217 an inappropriate moment
218 ** bug fix: run-program is now thread safe(r)
219 ** bug fix: inner with-recursive-lock no longer releases the mutex
220 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
221 * fixed some bugs revealed by Paul Dietz' test suite:
222 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
223 EXPECTED-TYPE slots filled.
224 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
226 ** MAKE-CONDITION accepts classes as type-designators.
227 ** COMPILE may never return NIL.
228 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
229 range before calling Unix time functions
231 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
232 * numerous signal handling fixes to increase stability
233 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
234 * minor incompatible change: we now correctly canonize default
235 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
236 as per the MOP, rather than the historical (INITARG INITFUNCTION
237 INITFORM). (reported by Bruno Haible)
238 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
239 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
240 for more information.
241 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
242 pathname is a directory pathname.
243 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
244 * dynamic space size on PPC has been increased to 768Mb. (thanks to
246 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
247 :METACLASS argument in addition to a class name. (reported by
248 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
249 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
250 (reported by Bruno Haible; patch from Gerd Moellmann)
251 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
253 * bug fix: more cleanups to the floating point exception handling on
254 x86-64 (thanks to James Knight)
255 * bug fix: the compiler does not try to constant fold calls of
256 COERCE and BIT-* functions when they return freshly constructed
257 objects. (reported by Nikodemus Siivola and Paul Dietz)
258 * optimization: improved the MIPS versions of generic (in the
259 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
260 * optimization: direct conversion of (unsigned-byte 32) to floats on
261 the PowerPC platform.
262 * optimization: structure instances with raw slots now use less
263 memory, and probably show better memory locality. (thanks to
265 * optimization: DYNAMIC-EXTENT declarations for lists and closures
266 are treated as requests for stack allocation on the x86-64,
267 Alpha-32, PPC and SPARC platforms.
268 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
269 the wrong file descriptor; implementation of SOCKET-OPEN-P.
270 (thanks to Tony Martinez)
272 ** gcing a dead thread can no longer lead to lockups
273 ** threads block signals until they are set up properly
274 ** errno is no longer shared by threads
275 ** interrupt-thread restores the eflags register on x86
276 ** fixed some lockups due to gc/thread interaction
277 * fixed some bugs revealed by Paul Dietz' test suite:
278 ** invalid dotted lists no longer raise a read error when
280 ** signal an error if a symbol that names a declaration is used
281 as the name of a type, or vice versa
282 ** allow using the (declare (typespec var*)) abbreviation for
283 (declare (type typespec var*)) with all type specifiers
284 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
285 ** always return NIL from PPRINT-POP when OBJECT is NIL
286 ** don't signal errors when pretty-printing malformed LABELS,
287 FLET or MACROLET forms
288 ** declarations in a DOLIST body are also in scope for the
290 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
292 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
295 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
296 * fixed cross-compiler leakages that prevented building a 32-bit
297 target with a 64-bit host compiler.
298 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
299 opened with :IF-EXISTS :OVERWRITE.
300 * fixed bug 281: error for an invalid qualifier in a short-form method
301 combination method is not signalled until the faulty method is called.
302 * bug fix: iteration variable type inferrer failed to deal with open
303 intervals. (reported by Alan Shields)
304 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
305 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
306 or more runtime options were provided to the sbcl binary.
307 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
308 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
309 * optimization: there's now a fast-path for fixnum arguments in the
310 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
311 * optimization: the code generated on x86-64 is more compact thanks
312 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
313 * bug fix: floating-point exception handling now partly works on
314 x86-64. (Thanks to James Knight)
315 * improvement to the MIPS backend from Thiemo Seufer: C-style
316 64-bit long long arguments and return values to alien functions
318 * fixed some bugs revealed by Paul Dietz' test suite:
319 ** the type-error signalled from WARN has a filled-in DATUM slot.
320 ** the type-error required when a stream is not associated with
321 a file has the stream as its datum.
322 ** type-errors on single-floats on x86-64 no longer have
323 :INVALID-OBJECT as the datum
324 ** the type-errors signalled for invalid function names now have
325 a correct expected type
326 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
327 for structures defined with a :TYPE. Documentation strings for
328 typed structures are no longer immediately discarded
329 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
330 broadcast streams and synonym streams. FILE-LENGTH now also works
331 on broadcast streams.
333 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
334 * incompatible change: the --noprogrammer option, deprecated since
335 version 0.7.5, has been removed. Please use the equivalent
336 --disable-debugger option instead.
337 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
339 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
340 inoperational for a while now) has been completely removed.
341 * null lexical environments are now printed as #<NULL-LEXENV>,
342 significantly reducing the amount of clutter in typical backtraces.
343 * documentation on weak pointers, finalization, and after GC hooks
344 has been added to the manual.
345 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
346 as other specialized array types, is much faster. SUBSEQ and
347 COPY-SEQ on such arrays have also been sped up.
348 * optimization: EQL is now more efficient when at least other argument
349 is known to be of type (OR FIXNUM (NOT NUMBER)).
350 * fixed bug: the runtime is now less vulnerable to changes in the
351 size of the SBCL object on OS X, and virtual memory is reserved for
352 all spaces that need to be at a fixed address.
353 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
354 * fixed bug: finalizers and after GC hooks that cause consing are now
356 * fixed bug: compiler error messages and summaries are now printed to
357 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
358 * fixed inference of the upper bound of an iteration variable.
359 (reported by Rajat Datta).
360 * fixed bug 211e: calling local functions with duplicated constant
361 keyword argument no longer causes a bogus style warning about an
363 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
364 ftype declaration to be disregarded. (reported by Dave Roberts)
365 * fixed bug 373: caused by erronous compilation of references to alien
366 variables in the runtime on ppc/darwin.
367 * fixed bug 376: CONJUGATE type deriver.
368 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
369 (reported by Baughn on #lisp)
370 * a cross-compiler bug on non-x86 platforms has been identified and
371 fixed. (thanks to Bruno Haible)
372 * improvements to the MIPS runtime code for increased stability.
373 (thanks to Thiemo Seufer)
374 * increased the maximimum compact environment size to allow
375 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
376 * partial workaround for bug 135: don't name the function we're
377 calling for hairy cases of EVAL, so as not to accumulate one environment
378 entry per call to EVAL. (reported by Kevin Reid)
379 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
380 * optimization: type testing for non-vector arrays should be faster.
381 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
383 * bug fix: setting 31st element of a bit vector to zero did not work
385 * bug fix: redefining a class definition which failed due to a
386 previous accessor / function clash now works (but see BUGS entry
387 #380 for more problems in this area). (thanks to Zach Beane)
388 * the long form of DEFINE-METHOD-COMBINATION disables method group
389 checking when given a single method group with pattern *. (thanks
391 * on x86 compiler supports stack allocation of results of simple
392 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
393 * support for building and running on Mac OS X 10.4 "Tiger" has been added
394 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
395 * fixed some bugs related to Unicode integration:
396 ** the restarts for recovering from input and output encoding
397 errors only appear when there is in fact such an error to
399 * fixed some bugs revealed by Paul Dietz' test suite:
400 ** MISC.549 and similar: late transformation of unsafe type
401 assertions into derived types caused unexpected code
403 ** SCALE-FLOAT type deriver is less wrong.
404 ** type derivers for EXP, LOG and similar functions compute result
405 types for complex arguments better.
406 ** (MISC.563) CONJUGATE type deriver works for very restricted
408 ** out-of-line type testers for character strings are available.
409 ** EQUAL compiler transform understands specialness of objects
411 ** accessing double-floats stored on the stack now works on x86-64.
412 ** debugger internals could sometimes create invalid lispobjs,
413 resulting in GC crashes.
414 ** MISC.548: type check weakening can convert required type into
416 ** initialization forms for bindings are not in scope of free special
419 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
420 * incompatible change: thread support for non-NPTL systems has
421 been removed - locking is buggy and unreliable. A threaded
422 SBCL build will now warn at startup and refuse to create
423 new threads, unless futex support is detected in the kernel
424 * incompatible change: the top level REPL now has only an ABORT
425 restart associated with it, not TOPLEVEL and ABORT as it used to.
426 TOP and TOPLEVEL are now available as debugger commands for
427 returning to the top level.
428 * incompatible change: forms evaluated in the REPL now use the
429 global optimization policy.
430 * incompatible change: user- and system-initialization files are
431 no longer processed with LOAD, but by READ and EVAL; hence the
432 global optimization policy, startup package, readtable, etc,
434 * The .fasl file format number has been incremented because of
435 various incompatible changes.
436 * internal entry point details and argument counts no longer appear
437 in backtraces unless explicitly requested by setting
438 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
439 * built-in and standard functions no longer have names like "top
440 level local call to FOO".
441 * fixed bug 32: functions defined in non-null lexical environments
442 now have more legible printed representation
443 * fixed bug 33: functions defined in non-null lexical environemnts
444 are now more amenable to inspection by INSPECT.
445 * workaround for bug 354: XEPs no longer appear in backtraces unless
446 explicitly requested.
447 * fixed bug: receiving the signal which results from attempting to
448 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
449 SIGBUS on other *BSDs) on architectures where the C stack is also
450 the Lisp stack (x86 and x86-64) from foreign code no longer leads
451 to debugger confusion or wild instability. (reported by Cheuksan
453 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
454 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
455 (reported by Lutz Euler)
456 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
457 compiled code. (reported by Paul Dietz)
458 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
459 specialized parameter to a method to something that is not TYPEP
460 the specializer is now possible.
461 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
462 face of package deletion.
463 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
464 pathname. (reported by David Tolpin on c.l.l)
465 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
466 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
467 STRING-STREAM and FILE-STREAM, respectively.
468 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
469 than previously; in addition, its test suite is now run on build.
470 (thanks to Luke Gorrie)
471 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
472 * optimization: added a immediate representation for single-floats
474 * fixed some bugs revealed by Paul Dietz' test suite:
475 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
476 ** attempting to create a package with a colliding nickname causes
477 correctable errors to be signalled.
478 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
479 ** several x86-64 backend bugs related to sign-extension of immediate
482 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
483 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
484 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
486 * fixed loading of multiply forward-referenced layouts.
487 (thanks to Cheuksan Wang)
488 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
489 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
490 (see COMPILE-FILE documentation for details of :PRINT option)
491 * fixed bugs 19 and 317: fixed-format floating point printing is
492 more accurate. This also fixes a bug reported by Adam Warner
493 related to the ~@F format directive.
494 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
496 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
497 dispatch table if the from-char is a dispatch macro character.
498 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
499 by the wordsize no longer ignore the last word. (reported by Lutz
501 * fixed bug in type checking of dynamic-extent variables. (reported
503 * optimization: sequence traversal functions use their freedom to
504 coerce function designators to functions.
505 * optimization: code with many calls to CLOS methods specialized on
506 CLOS classes has had redundant type checks removed.
507 * optimization: added declarations to speed up operations that access
508 the internal character database (for example STRING-UPCASE)
509 * optimization: comparison operations between floats and sufficiently small
510 fixnums no longer create extra rationals
511 * fixed some bugs related to Unicode integration:
512 ** portions of multibyte characters at the end of buffers for
513 character-based file input are correctly transferred to the
514 start of the buffer at the next read.
515 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
516 passing it through to OPEN.
517 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
518 argument given to internal calls to OPEN.
519 ** the built-in comment readers (introduced by character sequences
520 ";" and "#|") are more forgiving to encoding errors; they will
521 STYLE-WARN and then attempt to resync the stream at a character
522 boundary. (thanks to Teemu Kalvas)
523 * fixed some bugs in the x86-64 port:
524 ** Negative short int return values from c-calls are sign-extended
526 ** The stack is aligned to 16-bytes for c-calls, as required by
527 the ABI. (thanks to Cheuksan Wang)
528 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
529 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
530 for enums. (thanks to Vincent Arkesteijn)
531 ** Multiple small optimizations and bugfixes for floating point
533 * fixed some bugs revealed by Paul Dietz' test suite:
534 ** Space, Tab, Linefeed, Return and Page have the invalid
535 secondary constituent character trait.
536 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
538 ** WITH-INPUT-FROM-STRING should only update the index place on
540 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
542 ** Bit-array operations (BIT-AND and similar) worked incorrectly
543 with one-dimensional arrays with fill pointers.
544 ** TYPE-OF failed on a complex with an integer realpart and a
546 ** compiler failure during type inference for the code of form
547 (IF (EQL X (THE ...)) ...) (MISC.535).
549 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
550 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
551 platform. The port supports 61-bit fixnums, large memory spaces
552 and reloading shared object files.
553 * enhancement: saving cores with foreign code loaded is now
554 supported on ppc/Darwin in addition to the previously supported
556 * enhancement: the statistical profiler now walks deeper into the
557 call stack for more meaningful call-graphs and accrued time
558 reports (x86/x86-64 only). It also now reports time spent in
560 * enhancement: it is now possible to trace most individual methods
561 of a generic function in addition to tracing the generic function
563 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
565 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
566 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
567 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
568 produces an error. (thanks to Vincent Arkesteijn)
569 * bug fix: NAMESTRING on pathnames with :WILD components in their
570 directories works correctly. (thanks to Artem V. Andreev)
571 * fixed bug 125: compiler preserves identity of closures. (reported
573 * bug fixed: functions with &REST arguments sometimes failed with
574 "Undefined function" when compiled with (DEBUG 3). (reported by
576 * bug fix: overflow during compiling of setting element of a bit
577 vector with constant index and value. (reported by Timmy Douglas)
578 * build fix: fixed the dependence on *LOAD-PATHNAME* and
579 *COMPILE-FILE-PATHNAME* being absolute pathnames.
580 * on x86 compiler partially supports stack allocation of dynamic-extent
582 * GO and RETURN-FROM do not check the extent of their exit points
583 when compiled with SAFETY 0.
584 * fixed some bugs related to Unicode integration:
585 ** encoding and decoding errors are now much more robustly
586 handled; it should now be possible to recover even from invalid
587 input or output to the terminal. (thanks to Teemu Kalvas)
588 ** provided a first cut at implementing STRING-TO-OCTETS and
589 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
590 ** altered the SB-MD5 contributed module slightly, changing the
591 interface just enough for it to be supportable for builds where
592 lisp characters are not eight bits.
593 * fixed some bugs revealed by Paul Dietz' test suite:
594 ** the FORMATTER-generated functions for ~V[ conditionals require
595 the correct number of arguments.
596 ** READ-FROM-STRING returns the mandated second value when applied
597 to displaced strings.
598 ** the #\Rubout and #\Backspace characters are treated as invalid
599 constituent characters by the tokenizer.
601 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
602 * new feature: reloading changed shared object files with
603 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
604 * new feature: references to foreign variables and functions
605 can now be compiled and loaded before the corresponding shared
606 object file is loaded, as long as the foreign definitions are
607 available at runtime.
608 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
609 supported. (thanks to Dan Debertin)
610 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
611 just keywords. (thanks to Vincent Arkesteijn)
612 * fixed bug #331: structure-class instances corresponding to
613 DEFSTRUCT forms are now created eagerly.
614 * fixed bug #345: backtraces from calls to undefined functions work
615 on x86 as well. Related bug #61 is now also partially fixed on x86
616 (backtraces from throws to unknown catch tags.)
617 * bug fix: lambda-list parsing is now stricter vrt. order and number
618 of lambda-list keywords.
619 * bug fix: as specified by AMOP, an error is signalled if a
620 class-option appears multiple times in a DEFCLASS form. (reported
622 * bug fix: RANDOM can be compiled when the compiler derives the type
623 of its numeric argument as a disjoint set of small integers.
624 (reported by Paul Dietz)
625 * bug fix: starting a core saved with shared objects loaded when
626 those objects are not available no longer causes threaded SBCL to
627 hang. (reported by Sean Ross)
628 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
630 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
631 argument, which must always be NIL. (reported by Kalle Niemitalo)
632 * bug fix: printing 1.0d+23 no longer results in an error.
633 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
634 * bug fix: structure slot setters preserve evaluation order. (thanks
636 * fixed some bugs related to Unicode integration:
637 ** RUN-PROGRAM can allow its child to take input from a Lisp
638 stream. (reported by Stefan Scholl)
639 ** ASDF-INSTALL successfully downloads .tar.gz files in all
640 locales. (reported by Ken Causey)
641 * fixed some bugs revealed by Paul Dietz' test suite:
642 ** INCF, DECF and REMF evaluate their place form as specified in
644 ** FORMATTER expands ~{ ~} iteration directives with V or #
645 parameters correctly.
646 ** FORMATTER deals with the ~@[ ~] conditional directive where the
647 consequent uses no arguments correctly.
648 ** the system has a partial understanding of the (COMPLEX RATIO)
650 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
653 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
654 * new feature: a build-time option (controlled by the :SB-UNICODE
655 keyword feature, enabled by default) for building the system with
656 support for the entire 21-bit character space defined by the
658 * new feature: the system now has rudimentary external-format
659 support; the primary user-visible change at this time is that
660 characters with the high bit set (such as lower-case-e-acute) will
661 print correctly to a terminal in a UTF-8 environment.
662 * minor incompatible change: BASE-CHAR no longer names a class;
663 however, CHARACTER continues to do so, as required by ANSI.
664 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
665 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
666 has been moved to the SB-EXT package (temporarily re-exported from
668 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
669 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
671 * fixed bug #351: better error-handlind and reporting for malformed
673 * fixed bug #350: bignum-printing is now more memory-efficient,
674 allowing printing of very large bignums, eg. (expt 2 10000000).
675 (reported by Bruno Haible)
676 * fixed bug #302: better primitive-type selection for intersection
678 * fixed bug #308: non-graphic characters now all have names, as
679 required. (reported by Bruno Haible)
680 * bug fix: redefining a class with different superclasses now correctly
681 removes it from the direct-subclasses of its previous superclasses.
682 (reported by David Morse)
683 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
684 new value now works. (reported by Bruno Haible)
685 * bug fix: correct canonicalization of multiple non-standard slot
686 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
687 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
688 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
689 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
690 now exists, an signals an error.
691 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
692 during macroexpansion for non-iteration variables as well. (reported
693 by Bruno Haible for CMUCL)
694 * bug fix: Cyclic structures and unprintable objects in compiler
695 messages no longer cause errors. (reported by Bruno Haible)
696 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
697 and READ-FROM-STRING all now return a primary value of NIL if
698 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
699 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
700 (reported by Bruno Haible for CMUCL)
701 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
702 do not designate a condition. (reported by Bruno Haible for
704 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
705 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
706 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
707 permitting resolution in favour of any of the conflicting symbols.
708 (reported by Bruno Haible for CMUCL)
709 * bug fix: EQUAL compiler optimizations is less aggressive on
710 strings which can potentially compare true despite having distinct
711 specialized array element types.
712 * bug fix: unit enumerations can be defined without dividing by
713 zero. (thanks to Vincent Arkesteijn)
714 * FORMAT compile-time argument count checking has been enhanced.
715 (report from Bruno Haible for CMUCL)
716 * a partial workaround for the bug 262: the compiler does not try to
717 inline-expand a local function doing RETURN-FROM from a deleted
718 BLOCK. (thanks to Peter Denno for the bug report and to David
719 Wragg for the simple test case)
720 * fixed some bugs revealed by Paul Dietz' test suite:
721 ** RENAME-PACKAGE allows all package designators as new package
723 ** constraint propagation and lambda variable substitution are
724 more cautious in dealing with partially deleted code.
725 ** compiler sometimes lost reoptimization passes.
726 ** CERROR, given a condition as condition designator, passes its
727 remaining arguments to the continue format control without
729 ** Case-altering FORMAT directives work correctly on non-ASCII
731 ** The REAL type specifier handles bounds outside the floating
732 point ranges without signalling FLOATING-POINT-OVERFLOW.
733 ** Functions with IR1-transformations can create intercomponent
734 references to global functions.
735 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
737 ** FORMAT ~R treats a nil value for its first parameter correctly.
739 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
740 * enhancement: saving cores with foreign code loaded is now
741 supported on x86/NetBSD and sparc/Linux in addition to the previously
743 * bug fix: on some platforms repeated installations caused multiple
744 copies of HTML documentation to be installed -- should not happen
745 any more. (reported by Stefan Scholl)
746 * bug fix: parsing self-recursive alien record types multiple times
747 no longer causes infinite recursion. (reported by Thomas F. Burdick,
748 original patch by Helmut Eller for CMUCL)
749 * bug fix: stack-exhaustion detection works now on NetBSD as well.
750 (thanks to Richard Kreuter)
751 * bug fix: defining classes whose accessors are methods on existing
752 generic functions in other (locked) packages no longer signals
753 bogus package lock violations. (reported by François-René Rideau)
754 * bug fix: special variables as DEFMETHOD parameters no longer have
755 associated bogus type declarations. (reported by David Wragg and
757 * bug fix: read-write consistency on streams of element-type
758 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
759 * bug fix: redefiniton of the only method of a generic function with
760 no DEFGENERIC no longer emits a full WARNING. In addition,
761 redefinition of generic functions with no DEFGENERIC to an
762 incompatible lambda list now signals an error. (thanks to Zach
764 * bug fix: DEFGENERIC now works even when there's a function of the
765 same name in an enclosing lexical environment. (thanks to Zach
767 * fixed compiler failure, caused by instrumenting code during
768 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
769 * optimization: added loop analysis and improved register allocation
770 * fixed some bugs revealed by Paul Dietz' test suite:
771 ** POSITION on displaced vectors with non-zero displacement
772 returns the right answer.
773 ** (SIMPLE-STRING) is a valid type specifier for sequence
775 ** *PRINT-LEVEL* handling for slotless structures is pedantically
777 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
778 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
780 ** The justification version of the FORMAT ~< directive treats
781 non-zero minpad parameter correctly.
783 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
784 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
785 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
786 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
787 the supported interface.
788 * new feature: Single-stepping of code compiled with DEBUG 2 or
789 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
790 * new feature: saving cores with foreign code loaded is now
791 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
792 Timothy Moore's work for CMUCL)
793 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
794 parameters to * instead of NIL if no initform is supplied.
795 (reported by Johan Bockgaard)
796 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
797 correctly when FUNCALL appears as the car of the form. Note:
798 despite this FUNCALL forms are not currently subject to
799 compiler-macro expansion. (port of Raymond Toy's fix for the
800 same from CMUCL, reported by Johan Bockgaard)
801 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
802 (thanks for Teemu Kalvas)
803 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
804 case instead of terminating the iteration (thanks for Julian
805 Squires, Sean Champ and Raymond Toy)
806 * bug fix: incorrect expansion of defgeneric that caused a style
807 warning. (thanks for Zach Beane)
808 * bug fix: slot accessor effective method computation works properly
809 for classes with multiple non-standard applicable methods on
810 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
811 * on x86 compiler supports stack allocation of results of LIST and
812 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
813 CMUCL implementation by Gerd Moellmann)
814 * fixed some bugs revealed by Paul Dietz' test suite:
815 ** FORMAT strings with both the ~<~:;~> form of the justification
816 directive and pretty-printing directives cause an error.
818 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
819 * incompatible change: the internal functions
820 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
821 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
822 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
823 instead of the old functions.
824 * new feature: on platforms where "dladdr" is available foreign
825 function names now appear in backtraces. (based on Helmut Eller's
827 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
828 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
830 * documentation: ASDF manual is now distributed with SBCL.
831 * bug fix: SBCL can now load its contributed modules with REQUIRE
832 even if the system-provided entries have been removed from
833 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
835 * bug fix: backtraces involving undefined functions or assembly
836 routines are more informative. (thanks to Brian Downing)
837 * bug fix: mutually referent alien structures now work correctly.
838 (reported by Rick Taube)
839 * bug fix: structures defined by WITH-ALIEN can be referred to
840 within other definitions in the same WITH-ALIEN.
841 * bug fix: division operators (MOD, TRUNCATE and the like) with
842 constant zero divisors and integer dividends no longer generate
844 * bug fix: provide default methods for INPUT-STREAM-P and
845 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
846 * bug fix: improve the detection and resolution of MOP metacycles
847 (where effective method computation depends on the generic function
848 having its effective method computed). (reported by Bruno Haible)
849 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
850 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
851 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
852 * bug fix: pathnames with relative directory components can be
853 represented relative to default pathnames.
854 * optimization: in taking the GCD of bignums, reduce the two bignums
855 to approximately the same size (using Euclid's algorithm) before
856 applying the more sophisticated binary GCD. (thanks to Juho
858 * optimization: COUNT on bitvectors now operates word-at-a-time.
859 * optimization: ASH with a positive, but not necessarily constant,
860 (leftwards) shift, when in a modular context, is compiled to a
862 * fixed some bugs revealed by Paul Dietz' test suite:
863 ** FORMAT variable parameters ("~V<char>") are defaulted properly
864 if the corresponding argument is NIL.
865 ** FORMAT directives accepting a minpad parameter treat negative
867 ** literal commas as character parameters to FORMAT directives are
869 ** literal spaces directly after ~<Newline> directives within a
870 format-logical-block (~:< ~@:>) do not induce :FILL-style
871 conditional newlines.
872 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
873 as specified: it no longer includes conditional newlines).
874 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
876 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
877 :PER-LINE-PREFIX argument does not evaluate to a string.
879 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
880 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
881 the manual for details; for now, package locks can be disabled by
882 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
883 if no major problems are found then it is likely that they will be
884 compiled in unconditionally.
885 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
886 unsupported operators on all platforms. To load a shared library
887 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
888 object file, link it into a shared library outside of SBCL and
889 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
891 * bug workaround: ROOM T and the GC were not working together
892 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
893 reporting facility aren't preserved. That reporting has been
894 disabled, so now until and unless someone figures out how to make it
895 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
896 * minor incompatible change: as threatened around sbcl-0.8.0, the
897 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
898 The symbols are also exported from SB-PCL for backwards
899 compatibility, but more so than before SB-PCL should be treated as
900 an implementation-internal package.
901 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
903 * the SB-SPROF contrib now works on (most) non-x86 architectures.
904 It is known as of this release not to work on the Alpha, however.
905 * fixed bug #167: errors signalled due to illegal syntax in method
906 bodies are now more legible.
907 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
908 designators and can hence be used with TYPEP.
909 * fixed bug #333: CHECK-TYPE now ensures that the type error
910 signalled, if any, has the right object to be accessed by
911 TYPE-ERROR-DATUM. (reported by Tony Martinez)
912 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
913 5.1.2.3 for multiple-value place subforms. (reported by Kalle
915 * fixed bug #334: programmatic addition of slots using specialized
916 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
917 and :ALLOCATION :CLASS slots.
918 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
919 float exponents. (rereported by Peter Seibel)
920 * fixed a bug: #\Space (and other whitespace characters) are no
921 longer considered to be macro characters in standard syntax by
923 * fixed bug: initialization of condition class metaobjects no longer
924 causes an instance of the condition to be created. (reported by
926 * fixed bug: it is now possible to have slots such that
927 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
928 description is neither :INSTANCE nor :CLASS.
929 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
930 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
931 SLOT-VALUE-USING-CLASS), and no longer causes errors with
932 non-standard SLOT-DEFINITION-ALLOCATION values.
933 * fixed bugs: various race conditions handling exiting threads.
934 CL-PPCRE's thread torture test now passes.
935 * fixed arguably-a-bug: GC time proportional to number of threads in
936 system even when most of them are idle
937 * optimization: improved performance of BIT and SBIT on bit-vectors.
938 * .fasl file incompatibility: The fasl file version number has
939 been incremented because of changes associated with package locks.
941 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
942 * minor incompatible change: the system no longer provides
943 optimization hints (conditions which are TYPEP
944 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
945 mode; these hints are still emitted when higher SPEED optimization
947 * new contrib module: a sampling profiler (profiling by statistical
948 sampling, rather than by instrumenting functions) is available as
949 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
950 * the behaviour of the standard function ED is now customizeable by
951 third parties through a hook variable: see ED's documentation
952 string for information on the protocol.
953 * the compiler no longer emits efficiency notes for (FUNCALL X)
954 when the type of X is uncertain under default optimization
956 * fixed bug 276: mutating a binding of a specialized parameter to a
957 method to something that is not TYPEP the specializer is now
959 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
960 exhibit the required behaviour on the x86 platform. (thanks to
961 Peter van Eynde, Eric Marsden and Bruno Haible)
962 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
963 even for difficult arguments. (reported by Peter Graves)
964 * fixed bug 141a: the backquote printer now descends quoted
966 * fixed another bug in backquote printing: no more destructive
967 modification of the form's list structure. (reported by Brian
969 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
970 move between its address being taken and the call to
971 interrupt_thread, fixing a crashing race condition.
972 * the SB-POSIX contrib implementation has been adjusted so that it
973 no longer exhibits ridiculously poor performance when constructing
974 instances corresponding to C structs.
976 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
977 * minor incompatible change: the sb-grovel contrib now treats C
978 structures as alien (in the sense of SB-ALIEN) objects rather than
979 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
980 has implications for memory management of client code
981 (sb-grovel-returned objects must now be manually managed) and for
982 type safety (alien objects now have full types).
983 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
984 used to control emission of compiler diagnostics, rather than the
985 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
986 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
987 quality should be considered deprecated.
988 * install.sh now installs the user manual as well
989 * (not quite a new documentable feature, but worth considering in
990 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
991 beginnings of a semantically meaningful condition hierarchy is
992 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
994 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
995 designator as the defaults argument.
996 * fixed bug: Displaced arrays whose displaced-to array has become
997 too small now cause ARRAY-DIMENSION to signal an error, providing
998 for safer bounds-checking. (reported by Bruno Haible)
999 * fixed bug: DEFCLASS slot definitions with identical :READER and
1000 :WRITER names now signal a reasonable error. (reported by Thomas
1002 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1004 * fixed bug: Invalid slot specification errors now print correctly.
1005 (thanks to Zach Beane)
1006 * fixed bug 320: Shared to local slot value transfers in class
1007 redefinitions now happen corrently from superclasses as
1008 well. (reported by Bruno Haible)
1009 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1011 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1012 improper lists correctly. (reported by Bruno Haible)
1013 * fixed bug 313: source-transform for <fun-name> was erroneously
1014 applied to a call of a value of a variable with name <fun-name>.
1015 (reported by Antonio Menezes Leitao)
1016 * fixed bug 307: The obsolete instance protocol ensures that
1017 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1018 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1019 remove dead unknown-values globs from the middle of the stack.
1020 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1021 easier to resume long computations after using *BREAK-ON-SIGNALS*
1022 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1023 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1024 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1025 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1026 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1027 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1029 * fixed a (fairly theoretical) bug in string printing: if
1030 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1031 does not have array-element-type equal to the most general string
1033 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1034 function name. (thanks to Nikodemus Siivola)
1035 * fixed bug: compile-time format string checker failed on
1036 non-closed ~{. (reported by Thomas F Burdick)
1037 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1038 don't warn on legal loop code involving more than one aggregate
1039 boolean. (thanks to Nikodemus Siivola)
1040 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1041 in keyword position to MAKE-PACKAGE was not being reported as
1043 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1044 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1046 * fixed bugs 280 and 312: the checking for multiple definitions in a
1047 file is less likely to become confused by uses of inline
1049 * fixed bug: the #S reader macro performs the keyword coercion
1050 specified for slot names. (reported by Kalle Niemitalo)
1051 * fixed bug: lambda lists may contain symbols whose names start with
1052 & but are not lambda-list-keywords; their occurrence triggers a
1054 * fixed bug 321: define-method-combination argument lambda lists do
1055 not cause invalid code to be generated when &optional or &aux
1056 variables have default values. (reported by Bruno Haible)
1057 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1058 have CLOS classes; this appears also to have fixed reported
1059 instability in stack exhaustion detection.
1060 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1061 MUFFLE-WARNING when no associated restart is present is now
1063 * optimization: rearranged the expansion of various defining macros
1064 so that each expands into only one top-level form in a
1065 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1067 * optimization: used a previously little-used slot in symbols to
1068 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1070 * fixed some bugs revealed by Paul Dietz' test suite:
1071 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1073 ** Fixed an optimization bug related to inheritance of initforms
1074 from local to shared slots.
1075 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1076 ** CAST optimizer forgot to flush argument derived type.
1077 ** print/read consistency on floats is now orders of magnitude
1078 more likely. (thanks also to Bruno Haible for a similar report
1080 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1081 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1082 for objects of type REAL. Make it so.
1083 ** SXHASH is specified (infelicitously) to respect similarity,
1084 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1085 it so. (thanks to Markus Ziegler)
1086 ** on the Alpha, the compiler succeeds in compiling functions
1087 returning a known number of arguments greater than 63.
1088 ** fixed handling of invalid NIL arguments in keyword position
1090 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1092 ** provided a readably-printable representation for RANDOM-STATE
1094 ** ensured that pathnames, where they have a namestring, always
1095 print using #P"..." syntax.
1097 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1098 * Support for the forthcoming 2.0 version of the NetBSD kernel
1099 running on x86 hardware has been added. (thanks to Perry
1100 E. Metzger most immediately, and others for their past work)
1101 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1102 current release version; SBCL's OpenBSD support had been broken
1103 since about the time of OpenBSD's switch to ELF binary format.)
1104 * [placeholder for DX summary]
1105 ** user code with &REST lists declared dynamic-extent, under high
1106 speed or space and low safety and debug optimization policy.
1107 * The manual has been converted to Texinfo format and the debugger
1108 chapter from the cmucl manual has been added.
1109 * A facility has been added to extract documentation strings from
1110 sbcl and store them as Texinfo-formatted snippets for inclusion in
1111 the manual (via Texinfo's @include directive)
1112 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1113 the test case to Dave Roberts)
1114 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1115 pointers. (reported by Sean Ross)
1116 * bug fix: PROFILE output is printed nicely even for large numerical
1117 values. (thanks to Zach Beane)
1118 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1119 greater than 32 handle EOF correctly.
1120 * bug fix: on X86 an immediate argument of the IMUL instruction is
1121 correctly printed in disassembly. (thanks to Lutz Euler)
1122 * bug fix: class slots in redefined classes preserve their old
1123 values. (thanks to Bruno Haible and Nikodemus Siivola)
1124 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1125 succeeds. (reported by Marco Baringer)
1126 * fixed some bugs revealed by Paul Dietz' test suite:
1127 ** READ-SEQUENCE now works on ECHO-STREAMs.
1128 ** RATIONALIZE works more according to its specification. (thanks
1131 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1132 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1133 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1134 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1135 should matter to you only if you rebind the printer control
1136 variables and then find you want different bindings in the
1137 debugger than in the ordinary execution of your program.)
1138 * The runtime build system has been tweaked to support building
1139 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1140 assembler and linker. (thanks to Nikodemus Siivola)
1141 * Unbound, undefined, undeclared variables now trigger full
1142 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1143 more often programmer error than deliberate exploitation of undefined
1145 * optimization: the hash algorithm for strings has changed to one
1146 that is less vulnerable to spurious collisions. (thanks to Juho
1148 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1149 needless bounds checking. (thanks to Juho Snellman)
1150 * optimization: implemented multiplication as a modular
1151 (UNSIGNED-BYTE 32) operation on the PPC backend.
1152 * fixed some bugs revealed by Paul Dietz' test suite:
1153 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1155 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1156 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1157 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1159 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1161 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1163 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1164 their output stream on EOF from read.
1165 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1166 have been read to end-of-file.
1167 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1169 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1170 description of determination of which consecutive characters
1172 ** Printing the "Space" character with escaping on now yields "#\\ ",
1173 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1174 ** Reading floating-point numbers with *READ-BASE* set to a number
1175 less than 10 works correctly.
1176 ** Reading floating-point numbers with *READ-BASE* set to a number
1177 more than 10 works correctly.
1178 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1179 the readtable currently in effect.
1181 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1182 * minor incompatible change: parsing of namestrings on a physical
1183 (Unix) host has changed; numbers after the final #\. in a
1184 namestring are no longer interpreted as a version field. This is
1185 intented to be largely invisible to the user, except that the
1186 meaning of the namestring "*.*.*" has changed: it now refers to a
1187 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1188 should usually be replaced by
1189 (make-pathname :name :wild :type :wild :version :wild)
1190 with the added benefit that this is more likely to be portable.
1191 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1192 option to OPEN now signals an error if the file being opened
1193 exists; this may have an impact on existing code.
1194 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1195 signals received, on the PowerPC platforms (both Linux and
1196 Darwin). (thanks to Pierre Mai for pointing out the location of
1198 * several fixes on OS X: The system now builds and runs cleanly on
1199 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1200 (thanks to Brian Mastenbrook)
1201 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1202 second-resolution: integer multiples of 1/3600 between -24 and 24.
1203 (thanks to Vincent Arkesteijn)
1204 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1205 type when called with 1 argument; PEEK-CHAR checked type of
1206 PEEK-TYPE only after having read first character from a
1207 stream. (reported by Peter Graves)
1208 * bug fix: the garbage collector now has much better locality
1209 behaviour, and in particular no longer treats all memory as being
1210 exhausted when many small objects point to each other in a deeply
1212 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1213 recognized as being TYPEP their class.
1214 * bug fix: the PUSHNEW documentation string has been corrected.
1215 (thanks to Vincent Arkesteijn)
1216 * bug fix: defaulting of the value for the last of an atypically
1217 large number of multiple values being bound was not being
1218 performed correctly on the Alpha or PPC platforms
1219 * optimization: implemented multiplication as a modular
1220 (UNSIGNED-BYTE 32) operation on the x86 backend.
1221 * optimization: SEARCH on simple-base-strings can now be open-coded.
1222 (see also contrib/compiler-extras.lisp for inspiration for
1223 teaching the compiler about the Boyer-Moore algorithm).
1224 * value, returned by MAX (and MIN) called with several EQUALP, but
1225 not EQL, arguments now does not depend on compiler settings.
1226 * fixed some bugs revealed by Paul Dietz' test suite:
1227 ** in stack analysis liveness information is propagated from
1228 non-local entry points.
1229 ** pathwise CAST removing failed when the CAST node did not start
1231 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1232 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1234 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1235 host is already defined.
1236 ** RENAME-FILE works on streams instead of signalling an internal
1238 ** PEEK-CHAR uses the current readtable when determining whether
1239 or not a character is whitespace.
1240 ** MERGE-PATHNAMES handles the case when the pathname does not
1241 specify a name while the default-pathname specifies a version.
1242 ** Pathnames now stand a better chance of respecting print/read
1244 ** Attempting to use standardized file system operators with a
1245 pathname with invalid :DIRECTORY components signals a
1247 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1249 ** DIRECTORY on logical pathnames is more correct.
1250 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1251 signal a TYPE-ERROR if their argument is not a stream.
1252 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1253 designator argument does not designate a stream.
1254 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1255 examining the synonym.
1256 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1258 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1259 element-type larger than ([UN]SIGNED-BYTE 32).
1261 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1262 * When built with the :SB-FUTEX feature, threaded builds now take
1263 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1264 for faster/more reliable mutex and condition variable support.
1265 * Incompatible change (but one you probably shouldn't have been using
1266 anyway): the interface and code for arbitrating between multiple
1267 threads in the same user session has been redesigned.
1268 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1269 error when called without an explicit environment argument.
1270 (thanks to Nikodemus Siivola)
1271 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1272 confused about their position. (thanks to Adam Warner and Gerd
1274 * bug fix: Pretty printing backquoted forms with unquotations in the
1275 argument list position of various code constructs such as LAMBDA
1276 now works correctly. (reported by Paul Dietz)
1277 * bug fix: Pretty printing unquotations no longer loses all
1278 stream position information.
1279 * optimization: performance of string output streams is now less
1280 poor for multiple small sequence writes.
1281 * optimization: performance of CSUBTYPEP in the presence of complex
1282 expressions involving CONS and NOT many times has been improved.
1283 (reported by Paul Dietz)
1284 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1286 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1287 streams for the REPL, for the debugger, and so on. (thanks to
1289 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1290 * fixed some bugs revealed by Paul Dietz' test suite:
1291 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1292 is ignored if it is too small, rather than propagating through
1293 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1294 ** extremely complex negations of CONS types were not being
1295 sufficiently canonicalized, leading to inconsistencies in
1297 ** VALUES tranformer lost derived type.
1299 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1300 * fixed a bootstrapping bug: the build process no longer assumes
1301 that the various BOOLE-related constants have the same value in
1302 host and target lisps. (noted by Paul Dietz' test suite on an
1303 SBCL binary built from CLISP)
1304 * The system can now be dynamically linked on the MIPS platform,
1305 which enables dynamic loading of foreign code from Lisp. (thanks
1306 to Ralf Baechle for discussions on the MIPS ABI)
1307 * The system now records debugging information for its own source
1308 files in a filesystem-position-independent manner, relative to
1309 the "SYS" logical host.
1310 * fixed a compiler bug: MV-LET convertion did not check references
1311 to the "max args" entry point. (reported by Brian Downing)
1312 * tweaked disassembly notes to be less confident about proclaiming
1313 some instruction as an LRA. (thanks to Brian Downing)
1314 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1315 now each have their own history, command character, and other
1316 characteristics. (thanks to David Lichteblau)
1317 * fixed some bugs revealed by Paul Dietz' test suite:
1318 ** compiler failure in compiling LOGAND expressions including a
1320 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1321 shift greater than 32.
1322 ** FLUSH-DEST did not mark blocks for type check regeneration.
1323 ** HANDLER-CASE failed to accept declarations in handler clauses
1324 in some circumstances.
1326 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1327 * New code in contrib/sb-introspect (still probably not entirely
1328 stable yet) provides some support for smart Lisp development
1329 environments like SLIME.
1330 * The conditions signalled for errors occurring when loading .fasl
1331 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1332 in a way which should help ASDF recover gracefully.
1333 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1334 stuff has been cleaned up. If you code contrib/ stuff, this might
1335 affect you, and you can look at contrib/README, contrib/STANDARDS,
1336 and/or the 0.8.4.27 diff to check.
1337 * In full calls the compiler now does not generate checks for declared
1338 argument types for all arguments.
1339 * various threading fixes
1340 ** and some experimental patches which didn't make it into
1341 the main tree for this release, but which are shipped in
1342 contrib/experimental-thread.patch as a possible fix for some
1343 failures (deadlock, spinning...) in GC-intensive multithreaded
1345 * fixed PPC build problem (source code incompatibility of different
1346 library versions): added offsetof() hackery which attempts to divine
1347 where glibc maintainers put uc_mcontext today
1348 * fixed bug 282: compiler does not trust type assertions while passing
1349 arguments to a full call.
1350 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1351 &OPTIONAL VALUES type parameter.
1352 * fix bug 214: algorithm for noting rejected templates is now more
1353 similar to that of template seletion. (also reported by rydis on
1355 * fixed bug 141b: printing backquoted information readably and prettily
1356 inserts a space where necessary.
1357 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1358 (CHARACTER 10) are now reported as errors, rather than propagated
1359 as unknown types. (reported by piso on #lisp)
1360 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1361 with values NIL and :ERROR. (thanks to Milan Zamazal)
1362 * fixed bug 191c: CLOS now does proper keyword argument checking as
1363 described in CLHS 7.6.5 and 7.6.5.1.
1364 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1365 longer raise an error; further, using a list as a for-as-arithmetic
1366 counter now raises a meaningful error.
1367 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1368 now understood by sequence creation functions such as MAKE-SEQUENCE
1370 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1371 a STRING-INPUT-STREAM.
1372 * compiler enhancement: SIGNUM is now better able to derive the type
1374 * type declarations inside WITH-SLOTS are checked. (reported by
1376 * fixed some bugs revealed by Paul Dietz' test suite:
1377 ** incorrect optimization of TRUNCATE for a positive first
1378 argument and negative second.
1379 ** compiler failure in let-convertion during flushing dead code.
1380 ** compiler failure while deriving type of TRUNCATE on an
1381 interval, containing 0.
1382 ** ASH of a negative bignum by a negative bignum count now returns
1384 ** intersection of CONS types now canonicalizes properly, fixing
1385 inconsistencies in SUBTYPEP.
1387 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1388 * incompatible change: The --disable-debugger command line
1389 option now clobbers the debugger at a more fundamental
1390 level, by redefining #'INVOKE-DEBUGGER instead of by
1391 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1392 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1393 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1394 enter the debugger REPL and then suffer recursive errors
1395 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1396 while under the new system, BREAK in --disable-debugger mode
1397 terminates the system just as an unhandled error would.
1398 * fixed compiler performance when processing loops with a step >1;
1399 * bug fix: DOCUMENTATION now retrieves generic function
1400 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1401 support has been systematized, and now supports the methods
1402 specified by ANSI, along with a default method and a method for
1403 slot documentation. (reported by Nathan Froyd)
1404 * bug fix: effective methods associated with a generic function are
1405 no longer cached over a change of that generic function's method
1406 combination. (reported by Andreas Fuchs)
1407 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1408 names a non-existent directory. (thanks to Andreas Fuchs)
1409 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1410 the PPC platform, lead to stack corruption; this has been fixed.
1411 (reported by Rainer Joswig)
1412 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1413 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1414 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1415 arithmetic in general, is now much more reliable on the PPC
1417 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1418 platform now returns the right answer.
1419 * optimization: restored some effective method precomputation in
1420 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1421 precomputation is now tunable.
1422 * optimization: compiler-internal data structure use has been
1423 reviewed, and changes have been made that should improve the
1424 performance of the compiler by about 20%.
1425 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1426 simple-base-strings and simple-bit-vectors is improved.
1427 * optimization: the optimization of 32-bit logical and arithmetic
1428 functions introduced in version 0.8.3 on the x86 has been
1429 implemented on the mips, ppc and sparc platforms; an
1430 implementation of the same facility, but for 64-bit arithmetic,
1431 has been added for the alpha.
1432 * microoptimization: the compiler is better able to make use of the
1433 x86 LEA instruction for multiplication by constants.
1434 * bug fix: in some situations compiler did not report usage of
1435 generic arithmetic in (SPEED 3) policy.
1436 * bug 145b fix: compiler used wrong type specifier while converting
1437 MEMBER-types to numeric.
1438 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1440 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1441 index argument now works properly on the Alpha platform.
1442 * bug fix: floating point exception treatment on the Alpha platform
1444 * bug fix: FILE-POSITION works much better on string input and
1445 output streams. (thanks to Nikodemus Siivola)
1446 * bug fix: many threading/garbage collection symptoms sorted.
1447 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1448 might be pseudo-atomic.
1449 * internal change: Stopping for GC is now done with signals not
1450 ptrace. GC is now done in whichever thread wanted it, instead of
1452 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1454 * bug fix: result form in DO is not contained in the implicit
1456 * incompatible change: ICR structure is changed; the value part of
1457 CONTINUATION is now called LVAR; corresponding functions are
1458 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1459 * added type deriver for ISQRT (thanks to Robert E. Brown).
1460 * bug fix: better support for loading from the command line when an
1461 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1463 * fixed some bugs revealed by Paul Dietz' test suite:
1464 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1465 ** ROUND and FROUND now give the right answer when given very
1466 small float arguments.
1467 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1469 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1470 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1471 ** LCM should return a non-negative integer.
1472 ** PARSE-INTEGER returned the index of a terminator instead of the
1473 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1474 ** PARSE-INTEGER returned an incorrect index being applied to a
1476 ** LCM with two arguments of 0 returns 0 rather than signalling
1478 ** unsigned addition of a 32-bit constant with the high bit set no
1479 longer causes an internal compiler error on the x86.
1480 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1481 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1482 with negative last argument.
1483 ** byte specifiers with zero size and position no longer cause
1484 an error during type derivation.
1485 ** bignum multiplication on the Alpha platform now returns the
1487 * porting: The system now builds on SuSE AMD64, although it still
1488 generates a 32-bit binary.
1489 * .fasl file incompatibility: The fasl file version number has
1490 been incremented (because of the changes to internal compiler
1491 data structures referred to above).
1493 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1494 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1495 more accurately, on the Darwin kernel running on PowerPC hardware.
1496 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1497 * Compiler code deletion notes now signal a condition of type
1498 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1499 an associated MUFFLE-WARNING restart.
1500 * The compiler now performs limited argument count validation of
1501 constant format strings in FORMAT, and where appropriate in ERROR,
1502 CERROR and WARN. (thanks to Gerd Moellmann)
1503 * New ASDF-INSTALL contrib can be used for automatic download and
1504 installation of third-party Lisp code from CCLAN or other sites
1506 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1507 which forces another thread to execute a function supplied by the
1509 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1510 now accept and act upon their :ELEMENT-TYPE keyword argument.
1511 (reported by Martin Atzmueller, Edi Weitz)
1512 * bug fix: FILE-POSITION now accepts position designators up to
1513 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1514 is the greater. (thanks to Patrik Nordebo)
1515 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1516 Kalle Olavi Niemitalo)
1517 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1518 printable. (reported by Eric Marsden)
1519 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1520 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1521 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1522 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1523 (thanks to Stig Sandoe)
1524 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1525 UNREAD-CHAR now correctly includes the unread character in the
1526 target sequence. (thanks to Gerd Moellmann)
1527 * bug fix in threaded builds: the system can now be suspended and
1528 resumed by shell job control with minimal disruption.
1529 * bug fixes in times and timezones >2038AD
1530 * better handling of "where is GNU make?" problem in build scripts
1531 (thanks to Nikodemus Siivola)
1532 * new optimization: inside a named function any reference to a
1533 function with the same name is considered to be a self-reference;
1534 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1535 optimization quality.
1536 * new optimization on x86: logical functions and + now have
1537 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1538 used when the result is truncated to 32 bits.
1539 * VALUES declaration is partially enabled.
1540 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1541 * fixed some bugs revealed by Paul Dietz' test suite:
1542 ** The system now obeys the constraint imposed by
1543 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1544 types form a lattice under type intersection.
1545 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1546 ** ASSOC now ignores NIL elements in an alist.
1547 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1548 and (1+ MOST-POSITIVE-FIXNUM) answers.
1549 ** The addition of a method with invalid qualifiers to a generic
1550 function does not cause an error to be signalled immediately;
1551 a warning is signalled, and the error is generated only on
1552 calling the generic function.
1553 * changed .fasl file version number, in part to add type codes for
1554 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1555 obscure ANSI requirements
1557 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1558 * fixed bug 148: failure to inline-expand a local function left
1559 garbage, confusing the compiler.
1560 * fixed bugs 3cd: structure slot readers perform type check if the
1561 slot can have an invalid value (i.e. it is either not initialized
1562 or can be written with a less specific slot writer).
1563 * bug fix: the compiler now traps array references to elements off
1564 the end of an array; previously, the bounds checking in some
1565 circumstances could go off-by-one.
1566 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1568 * type declarations for array element types now obey the description
1569 on the CLHS page "Declaration TYPE", as per discussions on
1570 sbcl-help around 2003-05-08. This means that a declaration
1571 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1572 declaration, all references to BAR will be asserted or assumed
1573 (with THE, so dependent on compiler policy) to involve objects of
1574 type FOO. Note that no such declaration is implied in
1575 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1576 * declared types of functions from the "Conditions"
1577 chapter. (reported by Paul Dietz)
1578 * bug fix: CERROR accepts a function as its first argument.
1579 * bug fix: NTH an NTHCDR accept a bignum as index
1580 arguments. (reported by Adam Warner)
1581 * optimization: character compare routines now optimize comparing
1582 against a constant character. (reported by Gilbert Baumann)
1583 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1584 argument now works properly on the MIPS platform.
1585 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1586 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1587 declared to have a functional type, does not cause an error
1589 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1590 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1591 * fixed bug: if last continuation of a deleted block has a
1592 destination, this destination should be deleted too. (reported by
1594 * fixed a bug in the bootstrap process: the host compiler's values
1595 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1596 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1597 test case from Patrik Nordebo)
1598 * improved the ability of the disassembler on the PPC platform to
1599 provide helpful disassembly notes.
1600 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1601 the class in more cases than previously.
1602 * bug fix: FILE-POSITION now understands :START and :END for
1603 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1604 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1605 8) data. (thanks to David Lichteblau)
1606 * bug fix: it is possible to add a method to a generic function
1607 without lambda list.
1608 * bug fix: reader failed to signal END-OF-FILE inside an
1609 object representation. (reported by Nikodemus Siivola)
1610 * fixed some bugs revealed by Paul Dietz' test suite:
1611 ** LAST and [N]BUTLAST should accept a bignum.
1612 ** condition slot accessors are methods.
1613 ** (VECTOR NIL) is a subtype of STRING.
1615 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1616 * minor incompatible change: some nonsensical specialized lambda
1617 lists (used in DEFMETHOD) which were previously ignored now signal
1619 * minor incompatible change: the system is now aware of the types of
1620 variables in the COMMON-LISP package, and will signal errors for
1621 most violations of these type constraints (where previously they
1622 were silently accepted).
1623 * minor incompatible change: COMPILE-FILE now uses the freedom
1624 afforded (ANSI 3.2.2.3) to use derived function types for
1625 functions defined in the same file. This also permits the system
1626 to warn on static type mismatches and function
1627 redefinition. (Currently it does not work with high DEBUG level.)
1628 * minor incompatible change: VALUES declaration is disabled.
1629 * When issuing notes, the compiler now signals a condition of type
1630 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1631 restart for use in user handlers. It is expected that the
1632 COMPILER-NOTE condition will eventually become a condition
1633 supertype to a hierarchy of note types, which will then be
1634 handleable in a similar fashion. However, at the moment, no such
1635 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1636 but it's an implementation detail, not a classification for the
1638 * Changes in type checking closed the following bugs:
1639 ** type checking of unused values (192b, 194d, 203);
1640 ** template selection based on unsafe type assertions (192c, 236);
1641 ** type checking in branches (194bc).
1642 * A short form of VALUES type specifier has ANSI meaning (it has
1643 increased the number of situations when SBCL cannot perform type
1645 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1646 keywords or constants is permissible.
1647 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1648 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1649 classes. (thanks to Antonio Martinez)
1650 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1651 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1652 argument to a pretty-print stream. Instead, it leaves any such
1653 operations to DESCRIBE-OBJECT methods.
1654 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1656 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1657 lambda lists. (thanks to David Lichteblau)
1658 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1659 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1660 (thanks to Gerd Moellmann)
1661 * fixed bug 246: increased compilation speed of long
1662 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1664 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1666 * DEFCONSTANT now throws a condition of type
1667 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1668 constant to a non-EQL value; CONTINUE and ABORT restarts
1669 respectively change and preserve the value.
1670 * fixed bug 63: The code walker, part of the implementation of CLOS,
1671 is now better at handling symbol macros.
1672 * bug fix: There is no longer an internal implementation type named
1673 CL:LENGTH. (reported by Raymond Toy)
1674 * bug fix: In macro-like defining macros/special operators the
1675 implicit block does not enclose the lambda list.
1676 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1677 enclosed in parentheses) are not suitable as type specifiers, and
1678 their use properly signals an error now.
1679 * bug fix: An argument count mismatch for a type specifier in code
1680 being compiled no longer causes an unhandled error at compile
1681 time, but signals a compile-time warning.
1682 * fixed simple vector readable printing
1683 * bug fix: DESCRIBE takes more care over whether the class
1684 precedence list slot of a class is bound before accessing it.
1685 (reported by Markus Krummenacker)
1686 * bug fix: FORMATTER can successfully compile pretty-printer format
1687 strings which use variants of the ~* directive inside.
1688 * bug fix: SEARCH now applies its TEST predicate to the elements of
1689 the arguments in the correct order. (thanks to Wolfhard Buss)
1690 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1691 in type inference. (thanks to Robert E. Brown)
1692 * bug fix: Reading in symbols with an explicit package name of ""
1693 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1694 * fixed some bugs revealed by Paul Dietz' test suite:
1695 ** NIL is now allowed as a structure slot name.
1696 ** Arbitrary numbers, not just REALs, are allowed in certain
1697 circumstances in LOOP for-as-arithmetic clauses.
1698 ** Multiple class redefinitions before slot access no longer
1699 causes a type error.
1700 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1701 association between the name and a class.
1702 ** Generic functions with non-standard method-combination and over
1703 five methods all of which return constants no longer return NIL
1704 after the first few invocations. (thanks to Gerd Moellmann)
1705 ** CALL-NEXT-METHOD with no arguments now passes the original
1706 values of the arguments, even in the presence of assignment.
1707 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1709 ** DISASSEMBLE works with closures and funcallable instances.
1710 ** ADD-METHOD now returns the generic function, not the new method.
1711 ** FIND-METHOD signals an error if the lengths of the specializers
1712 is incompatible with the generic function, even if the ERRORP
1714 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1715 which its argument is a member.
1716 ** DEFCLASS only redefines the class named by its class-name
1717 argument if that name is the proper name of the class;
1718 otherwise, it creates a new class.
1719 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1720 of the UNBOUND-SLOT condition to the name of the slot.
1721 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1722 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1723 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1724 SLOT-MAKUNBOUND in the specified fashion.
1726 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1727 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1728 cross-compilation host. As a consequence, we can now bootstrap our
1729 way up to SBCL starting with a bare gcc toolchain and human-readable
1730 source code (first the source to CLISP, then the source to SBCL).
1731 * A contributed module containing a partial implementation of the
1732 simple-streams interface has been included. (thanks to Rudi
1734 * A contributed module implementing the RFC1321 Message Digest
1735 Algorithm, known as MD5, has been included.
1736 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1737 no longer has any effect, as the code controlled by this feature
1738 has been deleted. (As far as we know, no-one has ever built using
1739 this feature, and its semantics were confused in any case).
1740 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1741 work on conditions (as required by the ANSI specification),
1742 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1743 expected behaviour on conditions. Users should note, however,
1744 that such behaviour is not required by the ANSI specification,
1745 and so use of this behaviour may render their code unportable.
1746 * fixed some bugs revealed by Paul Dietz' test suite:
1747 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1749 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1750 ** SXHASH on condition objects no longer returns NIL.
1751 ** :ALLOCATION :CLASS slots are better treated; their values are
1752 updated on class redefinition, and initforms inherited from
1753 superclasses are applied.
1754 ** REMOVE-METHOD returns its generic function argument even when
1755 no method was removed.
1756 ** SHARED-INITIALIZE now initializes the values of the requested
1757 slots, including those with :ALLOCATION :CLASS.
1758 ** ALLOCATE-INSTANCE now works on structure classes defined via
1759 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1761 ** SLOT-EXISTS-P now works on conditions, as well as structures
1763 ** MAKE-LOAD-FORM now has the required methods on
1764 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1765 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1766 keyword, and now implements the SLOT-NAMES argument.
1767 ** methods with &OPTIONAL arguments no longer allow too many
1768 arguments to be passed in the call without error.
1769 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1770 option is consistent with the required arguments of the generic
1771 function lambda list.
1772 * bug fix: REQUIRE accepts a string designator. (Thanks to
1774 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1775 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1776 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1777 * bug fix: The compiler now checks for duplicated variables in macro
1779 * bug fix: SETQ on globals returns the correct value.
1780 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1781 not a COMPILER-ERROR (followed by some other strange error on
1782 choosing the CONTINUE restart).
1783 * bug fix: make.sh and friends are now more consistent in the way that
1784 they look for GNU "make".
1786 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1787 * experimental native threads support (on x86 Linux >=2.4 only).
1788 This is not compiled in by default: you need to add :SB-THREAD to
1789 the target features. See the "Beyond ANSI" chapter of the manual
1791 * fix for longstanding nonANSIism: The old distinction between
1792 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1793 The return value from CL:FIND-CLASS is now a CLOS class, and
1794 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1795 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1796 * An interface to the MetaObject Protocol, as described in Kiczales,
1797 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1798 MIT Press, 1991, is available from the SB-MOP package.
1799 * incompatible change: the SB-PCL package should now be considered
1800 a private implementation detail, and no longer a semi-private MOP
1802 * minor incompatible change: due to rearrangement for threads, the
1803 control stack and binding stack are now allocated at arbitrary
1804 addresses instead of being hardcoded per-port. Users affected by
1805 this probably have to be doing advanced things with shared
1806 libraries, and will know who they are.
1807 * minor incompatible change: Previously, all --eval forms used were
1808 processed with READ before any of them were processed with EVAL.
1809 Now each --eval form is processed with both READ and EVAL before
1810 the next --eval form is processed. (Thus package operations like
1811 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1812 work as the user might reasonably expect.)
1813 * minor incompatible change: *STANDARD-INPUT* is now only an
1814 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1816 * minor incompatible change: Y-OR-N-P is now character-oriented,
1817 not line oriented. Also, YES-OR-NO-P now works without errors.
1818 (thanks to Antonio Martinez)
1819 * sb-aclrepl module improvements: an integrated inspector, added
1820 repl features, and a bug fix to :trace command.
1821 * Known functions, which cannot be open coded by the backend, are
1822 considered to be able to check types of their arguments. (fixing
1823 a bug report by Nathan J. Froyd)
1824 * fixed a bug in computing method discriminating functions: It is
1825 now possible to define methods specialized on classes which have
1826 forward-referenced superclasses. (thanks to Gerd Moellmann)
1827 * fixed evaluation order in optional entries (reported by Gilbert
1829 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1830 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1831 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1832 required-by-AMOP NAME argument, as well as CLASS and
1833 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1834 * fixed bug 20: DEFMETHOD can define methods using names that are
1835 not the proper names of classes to designate class specializers.
1836 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1837 Unix file descriptors, instead of blowing up. (thanks to Antonio
1839 * Garbage collection refactoring: user-visible change is that a
1840 call to the GC function during WITHOUT-GCING will not do garbage
1841 collection until the end of the WITHOUT-GCING. If you were doing
1842 this you were probably losing anyway.
1843 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1844 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1845 TYPEP the latter but not the former.
1846 * The compiler issues a full WARNING for calls to undefined functions
1847 with names from the CL package.
1848 * MAP-INTO for a vector destination is open coded. (reported by
1849 Brian Downing on c.l.l)
1850 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1851 documentation string.
1852 * fixed some bugs revealed by Paul Dietz' test suite:
1853 ** COPY-ALIST now signals an error if its argument is a dotted
1855 ** Condition slots are now accessed more correctly in the presence
1856 of multiple initargs for a given slot.
1857 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1858 exclude restarts of the same name associated with a different
1860 ** DEFCLASS of forward-referenced classes with another
1861 forward-referenced class in the superclasses list no longer
1863 ** Condition slots are now initialized once each, not multiple
1864 times. (thanks to Gerd Moellmann)
1865 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1867 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1868 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1869 arguments contain duplicated elements.
1870 ** RESTART-CASE understands local macros.
1871 ** RESTART-CASE associates exactly its own restarts with a condition.
1872 ** ENDP in safe mode checks its argument to be of type LIST.
1873 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1874 in question is unbound.
1875 ** Optimized MAKE-INSTANCE functions no longer cause internal
1876 assertion failures in the presence of duplicate initargs.
1877 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1878 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1879 propagated even in the ctor optimized implementation of
1881 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1883 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1884 * a better implementation of SXHASH on (simple) bit vectors,
1885 measured both in execution speed and in distribution of results
1886 over the positive fixnums, has been installed. Likewise, a better
1887 implementation of EQUAL for simple bit vectors is now available.
1888 * fixed CEILING optimization for a divisor of form 2^k.
1889 * fixed bug 240 (emitting extra style warnings "using the lexical
1890 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1891 by Antonio Martinez)
1892 * fixed SXHASH, giving different results for NIL depending on type
1893 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1894 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1895 variable. (found by Rolf Wester)
1896 * SBCL does not ignore type declarations for special
1897 variables. (reported by rif on c.l.l 2003-03-05)
1898 * some bug fixes in contrib/sb-aclrepl/
1899 * fixed some bugs revealed by Paul Dietz' test suite:
1900 ** a bug in the CONS type specifier, whereby the CAR and CDR
1901 types got intertwined, has been fixed;
1902 ** the type system is now able to reason about the interaction
1903 between INTEGER and RATIO types more completely;
1904 ** APPEND, [N]REVERSE and NRECONC check that those their
1905 arguments, which must be proper lists, are really so;
1906 ** An array specialized to be unable to hold elements has been
1907 implemented, as required -- yes, really -- by ANSI;
1908 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1909 on malformed property lists;
1911 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1912 * incompatible packaging change: in line with Unix convention,
1913 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1914 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1915 * REQUIRE and PROVIDE are now optionally capable of doing something
1916 useful. See the documentation string for REQUIRE.
1917 * infrastructure for a managed SBCL contrib system: contributed
1918 modules in this release include:
1919 ** the ASDF system definition facility;
1920 ** an interface to the BSD Sockets API;
1921 ** an ACL-like convenience interface to the repl;
1922 (thanks to Kevin Rosenberg)
1923 ** an implementation of ROTATE-BYTE, with efficient implementation
1925 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1926 gives a closer approximation to the right answer than previously.
1927 (thanks to Raymond Toy)
1928 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1929 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1930 optional environment arguments, as required by ANSI.
1931 * fixed bugs in other functions taking environment objects, allowing
1932 calls with an explicit NIL environment argument to be compiled
1934 * fixed bug 228: primary return values from
1935 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1936 COMPILE or FUNCTION.
1937 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1938 :TYPE will no longer signal errors on innocuous objects.
1939 * fixed bug 231b: SETQ is better at respecting type declarations in
1940 the lexical environment.
1941 * fixed a bug in DEFCLASS: classes named by symbols with no or
1942 unprintable packages can now be defined.
1943 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1944 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1945 * fixed compiler failure related to checking types of functions.
1946 (reported by Robert E. Brown)
1947 * the compiler is now much more consistent in its error-checking
1948 treatment of bounding index arguments to sequence functions: in
1949 (SAFETY 3) code, errors will be signalled in almost all cases if
1950 invalid sequence bounding indices are passed to functions defined
1951 by ANSI to operate on sequences.
1952 * fixed a bug in the build procedure: documentation of SBCL-specific
1953 packages is now preserved and available in the final Lisp image.
1954 * lifted FDEFINITION lookup out of loops in the implementation of
1955 many list operations. (thanks to Robert E. Brown)
1956 * fixed a bug in the reader: the #n# reader macro now works for
1957 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1958 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1959 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1960 E. Brown and Tony Martinez respectively)
1961 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1962 SLOT-BOUNDP: in particular, it is now able to optimize them much
1963 better, and is now not vulnerable to having packages renamed.
1964 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1965 always called when a slot is not present in an instance. (thanks
1967 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1968 longer signal an error on unreading a character following EOF on
1969 the previous constituent stream. (thanks to Tony Martinez)
1970 * fixed some bugs revealed by Paul Dietz' test suite:
1971 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1972 not just nonnegative fixnums;
1973 ** the logical bit-array operators such as BIT-AND now accept an
1974 explicit NIL for their "opt-arg" argument (to indicate a
1975 freshly-consed result bit-array);
1976 ** ELT now signals an error on an invalid sequence index in safe
1978 ** the type system is now cleverer about negations of numeric
1979 types, and consequently understands the BIGNUM and RATIO types
1981 ** the type system is now cleverer about the interaction between
1982 INTEGER and RATIO types: while bugs still remain, many more
1983 cases are accurately computed;
1984 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1985 if it is in the last clause;
1986 ** CONSTANTLY now correctly returns a side-effect-free function in
1988 ** DECLARE is no longer treated as a special-operator; in
1989 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1990 * incremented fasl file version number due to the change in the
1991 DEFSTRUCT-SLOT-DESCRIPTION structure.
1993 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1994 * minor incompatible change: code processed by the "interpreter" or
1995 EVAL now has a compilation optimization policy of (DEBUG 2)
1996 (changed from (DEBUG 1)) to improve debuggability of interactive
1997 development, and to allow the use of the debug RETURN command in
1999 * an experimental implementation of the RETURN command for the
2000 debugger has been included. (thanks to Frederik Kuivinen)
2001 * fixed bug 62: constraints were not propagated into a loop.
2002 * fixed bug in embedded calls of SORT (reported and investigated by
2004 * fixed some bugs revealed by Paul Dietz' test suite:
2005 ** printing and reading of arrays with some dimensions having
2006 length 0 (thanks to Gerd Moellmann);
2007 ** BOA constructor with &AUX argument without a default value does
2008 not cause a type error;
2009 ** CONSTANTP now returns true for all self-evaluating objects.
2011 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2012 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2013 accessors that are related by inheritance, as specified in the
2014 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2015 Valtteri Vuorikoski)
2016 * The compiler is now able to inline functions that were defined in
2017 a complex lexical environment (e.g. inside a MACROLET).
2018 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2020 * Support for the upcoming FreeBSD-5.0 release has been included.
2021 (thanks to Dag-Erling Smorgrav)
2022 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2023 effect when it is not in a toplevel context.
2024 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2025 stand a better chance of being correct. (thanks to Gerd
2027 * fixed bug in COERCE, which now signals an error on coercing a
2028 rational to a bounded real type which excludes the expected
2030 * The compiler is now able to derive types more accurately from the
2031 COERCE and COMPILE functions.
2032 * fixed bug 223: functional binding is considered to be constant
2033 only for symbols in the CL package.
2034 * fixed bug 231: SETQ did not check the type of a variable being set
2035 (reported by Robert E. Brown)
2036 * A new optimization for MAKE-INSTANCE has been included, fixing
2037 various bugs (including relating to :ALLOCATION :CLASS slots and
2038 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2040 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2041 ** As required by ANSI, LOOP now disallows anonymous collection
2042 clauses such as COLLECT I in conjunction with aggregate boolean
2043 clauses such as THEREIS (= I 1);
2044 ** LOOP now signals an error when any variable is reused in the
2045 same loop (including the potentially useful construct analogous
2046 to WITH A = 1 WITH A = (1+ A);
2047 ** IT is only a special loop symbol within the first clause of a
2048 conditional loop clause;
2049 ** LOOP with a typed iteration variable over a hashtable now
2050 signals a type error iff it should.
2051 * fixed some other bugs revealed by Paul Dietz' test suite:
2052 ** FILE-STREAM now names the class previously known as FD-STREAM;
2053 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2054 argument) no longer signals an error;
2055 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2056 of the slot symbol, rather than using the current package
2057 ((:CONC-NAME "") continues to intern the slot's name in the
2059 * incremented fasl file version number, because of the incompatible
2060 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2061 of the new implementation of DEFINE-COMPILER-MACRO.
2063 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2064 * Support for building SBCL for MIPS platforms running in
2065 little-endian mode has now been checked in, and basic
2066 functionality on said platforms verified.
2067 * minor incompatible change: PCL now records the pathname of a file
2068 in which methods and the like are defined, rather than its
2070 * minor incompatible change: TRUENAME now considers the truename of
2071 a file naming a directory to be the pathname with :DIRECTORY
2072 component indicating that directory.
2073 * minor incompatible change: a NAMED clause in the extended form of
2074 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2075 reason for the previous behaviour is unclear.
2076 * more systematization and improvement of CLOS and MOP conformance
2077 in PCL (thanks to Gerd Moellman and Pierre Mai):
2078 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2080 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2081 primary methods with no specializers;
2082 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2084 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2085 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2086 FORWARD-REFERENCED-CLASSes; error reporting on
2087 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2089 ** SXHASH on CLOS instances now uses a slot internal to the
2090 instance to return different numbers on distinct instances,
2091 while preserving the same return value through invocations of
2093 ** DEFMETHOD signals errors when methods with longer incongruent
2094 lambda lists are added to generic functions;
2095 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2096 CLASS, as specified in AMOP;
2097 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2098 on the order returned by the primary method for classes of
2099 class STANDARD-CLASS;
2100 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2101 * fixed some bugs shown by Paul Dietz' test suite:
2102 ** DOLIST puts its body in TAGBODY;
2103 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2105 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2106 value producing form;
2107 ** if more variables are given to PROGV than values, extra
2108 variables are bound and made to have no value;
2109 ** NSUBSTITUTE on list arguments gets the right answer with
2111 ** ELT signals an error of type TYPE-ERROR when the index argument
2112 is not a valid sequence index;
2113 ** LOOP signals (at macroexpansion time) an error of type
2114 PROGRAM-ERROR when duplicate variable names are found;
2115 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2116 ** FUNCALL of special-operators now cause an error of type
2118 ** PSETQ now works as required in the presence of side-effecting
2119 symbol-macro places;
2120 ** NCONC accepts any object as its last argument;
2121 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2123 ** loop-for-as-package does not require a package to be explicitely
2125 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2127 * fixed bug 166: compiler preserves "there is a way to go"
2128 invariant when deleting code.
2129 * fixed bug 172: macro lambda lists with required arguments after
2130 &REST arguments now cause an error to be signalled. (thanks to
2132 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2133 bug. (thanks to Matthew Danish)
2134 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2136 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2137 arguments are lexically rebound. (thanks to Gerd Moellmann and
2139 * fixed bug 194: error messages are now more informative when there
2140 is no primary method applicable in a call to a generic
2141 function. (thanks to Gerd Moellmann)
2142 * fixed bug in command line argument checking (thanks to Julian
2144 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2145 (thanks to Matthew Danish)
2146 * incremented fasl file version number, because of the
2147 SXHASH-related changes in the layout of CLOS data structures
2149 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2150 * minor incompatible change: The runtime (the Unix executable named
2151 "sbcl") is now much pickier about the .core files it will load.
2152 Essentially it now requires .core files to descend from the same
2153 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2154 as the runtime does. (The intent is to prevent the crashes which
2155 can occur, and which can even be reported as mysterious failures,
2156 when people patch the sources or change the build parameters
2157 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2158 sbcl and .core files.)
2159 * fixed bug: VALUES-LIST is no longer optimized away.
2160 * fixed bug 142: The FFI conversion of C string values to Lisp
2161 string values no longer conses excessively. (thanks to Nathan
2162 Froyd porting Raymond Toy's fix to CMU CL)
2163 * began to systematize and improve MOP conformance in PCL (thanks to
2164 Nathan Froyd, Gerd Moellman and Pierre Mai):
2165 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2167 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2168 ** FINALIZE-INHERITANCE is now called on class finalization;
2169 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2170 argument precedence order.
2171 * fixed bug 202: The compiler no longer fails on functions whose
2172 derived types contradict their declared type.
2173 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2174 so it can be non-toplevel.
2175 * The fasl file version number has changed (because of the new
2176 implementation of DEFMACRO).
2177 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2178 safe code, that any length requirement by their type-specifier
2179 argument is valid. The exceptions to this are described in bug
2181 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2182 functions now check the types of their inputs as required by ANSI.
2183 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2184 introduced symbol is DECLAREd to be SPECIAL.
2185 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2186 * fixed bug: :COUNT argument to sequence functions may be negative
2187 * fixed bug: body of DO-SYMBOLS may contain declarations
2188 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2189 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2190 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2191 symbol macro only once
2192 * fixed printing of call frame when argument list is unavailable
2193 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2194 * compiler no longer signals WARNING on unknown keyword
2197 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2198 * A beta-quality port to the mips architecture running Linux,
2199 based on the old CMUCL backend, has been made. It has been tested
2200 on a big-endian kernel, and works sufficiently well to be able to
2201 rebuild itself; it has not been tested in little-endian mode.
2202 * fixed an inconsistency between gencgc.c and purify.c which made
2203 dumping/loading .core files unreliable
2204 * fixed bug 120a: The compiler now deals correctly with IFs where
2205 the consequent is the same as the alternative, instead of
2206 misderiving the return type. (thanks to Alexey Dejneka)
2207 * fixed bug 113: Logical pathnames are now dumpable (the logical
2208 host is resolved at load-time, throwing an error if it is not
2210 * fixed bug 174: FORMAT's error message is slightly clearer when a
2211 non-printing character is used in a format directive.
2212 * fixed several bugs in compiler checking of type declarations, i.e.
2213 violations of the Python "declarations are assertions" principle
2214 (thanks to Alexey Dejneka)
2215 * fixed several bugs in PCL's error checking (thanks to Gerd
2217 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2219 * fixed bug in compilation of functions as first class values
2220 (thanks to Antonio Martinez-Shotton)
2221 * The compiler's handling TYPE-ERRORs which it can prove will
2222 inevitably happen at runtime has been cleaned up and corrected
2223 in several ways. (thanks to Alexey Dejneka)
2224 * improved argument type checking for various basic arithmetic
2225 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2226 lavished on them in the past that they can be compiled in many
2227 ways in different special cases
2228 * fixed bug 181: compiler checks validity of user supplied type
2230 * cleaned up code flushing in optimization: Function calls which
2231 should signal errors for safety purposes (e.g. which ANSI says
2232 should signal errors when their arguments are of incorrect type)
2233 are no longer optimized away.
2234 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2235 * incremented fasl file version number, because changes in the
2236 implementation of sequence functions like COERCE caused
2237 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2238 in old inline expansions) to become undefined. (Actually these
2239 changes were later undone, so we might very well be binary
2240 compatible with 0.7.7 after all, but leaving the version number
2241 incremented seemed like the simplest and most conservative
2244 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2245 * An alpha-quality port to the parisc architecture running Linux,
2246 based on the old CMUCL backend, has been made. This, even more so
2247 than the other backends, should be considered still a work in
2248 progress; known problems include that the Linux kernel in 64-bit
2249 mode does not propagate the correct sigcontext structure to
2250 userspace, and consequently SBCL on a parisc64 kernel will not
2252 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2253 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2254 suppressed.) Also now that the compiler is looking at declarations
2255 in the environment, it checks optimization declarations as well,
2256 and suppresses inlining when (> DEBUG SPEED).
2257 * More fixes have been made to treatment of floating point exception
2258 treatment and other Unix signals. In particular, floating point
2259 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2260 * The detection and handling of control stack exhaustion (infinite
2261 or very deeply nested recursion) has changed. Stack exhaustion
2262 detection is now done by write-protecting pages at the OS level
2263 and applies at all optimization settings; when found, a
2264 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2265 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2266 be caught using IGNORE-ERRORS.
2267 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2268 correctly when transforming recursive LABELS functions to LETs.
2269 (thanks to Alexey Dejneka)
2270 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2271 that are names of constants or global variables.
2272 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2273 alien routines with docstrings.
2274 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2275 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2277 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2278 (thanks to Eric Marsden)
2279 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2280 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2281 * Bug fix: The SPARC backend can now compile functions involving
2282 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2283 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2284 to a functions expecting a PATHNAME-DESIGNATOR.
2285 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2286 lambda lists. (thanks to Alexey Dejneka)
2287 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2288 functions so that EXPORTing the name of the class doesn't cause
2289 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2290 to Antonio Martinez for reporting this)
2291 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2292 constant-folded without failing an assertion. (thanks to Einar
2293 Floystad Dorum for reporting this)
2294 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2295 types are now dealt with more correctly by the compiler.
2296 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2297 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2298 behaviour analogous to RENAME-FILE. This puts its behaviour more
2299 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2301 * The fasl file version number has changed again. (because of the
2302 bug fix involving the names of PCL MAKE-INSTANCE functions)
2304 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2305 * bug fix: Floating point exceptions are treated much more
2306 consistently on the x86/Linux and PPC/Linux platforms.
2307 * Array initialization with :INITIAL-ELEMENT is now much faster for
2308 cases when the compiler cannot open code the array creation, but
2309 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2310 array accesses have also seen a speed increase.
2311 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2312 specified. (This was at the root of some bad interactions between
2313 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2315 * bug fix: Internal error arguments for undefined functions are now
2316 computed correctly on the PPC/Linux platform.
2317 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2318 Raymond Toy's patch for CMU CL)
2319 * Support for the Solaris 9 operating environment has been included
2320 (thanks to Daniel Merritt)
2321 * A very ugly but hopefully complete draft of the missing FFI chapter
2322 of the manual has been created by reformatting the corresponding
2323 CMU CL manual chapter into (currently very ugly and incoherent)
2324 DocBook and bringing it up to date for SBCL behavior. Thus, the
2325 manual is now essentially complete, at least by my extreme
2326 once-and-only-once standards, whereby it's acceptable to refer to
2327 the doc strings of SB-EXT functions as the primary documentation.
2328 * The fasl file version number has changed again, due to cleanup of
2329 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2330 is no longer a static symbol.)
2332 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2333 * SBCL now builds with OpenMCL (version 0.12) as the
2334 cross-compilation host; also, more progress has been made toward
2335 bootstrapping under CLISP.
2336 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2338 * bug 158 fixed: The compiler can now deal with integer loop
2339 increments different from 1; fixing this turned out also to fix
2341 * bug 169 fixed: no more bogus warnings about using lexical bindings
2342 despite the presence of perfectly good SPECIAL declarations (thanks
2343 to David Lichteblau)
2344 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2345 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2346 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2348 * bug 180 fixed: Method combination specifications no longer ignore
2349 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2350 * bug fix: Structure type predicate functions now check their argument
2351 count as they should.
2352 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2353 correctly. (thanks to Pierre Mai)
2354 * minor incompatible change: The --noprogrammer option is deprecated
2355 in favor of the new --disable-debugger option, which is very similar.
2356 (The major difference is that it takes effect at a slightly different
2357 time at startup, causing handling of errors in --sysinit and
2358 --userinit files will be affected differently.) The
2359 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2360 been added to allow this functionality to be controlled from ordinary
2361 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2362 maintainers, who might want to run non-interactive scripts to
2363 build SBCL cores which will later be used interactively.)
2364 * minor incompatible change: The LOAD function no longer, when given
2365 a wild pathname to load, loads all files matching that pathname.
2366 Instead, an error of type FILE-ERROR is signalled.
2368 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2369 * bug 147 fixed: The compiler preserves its block link/count
2370 invariants more correctly now so that it doesn't crash. (thanks
2372 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2374 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2375 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2376 no-value-for-this-character as NIL (as specified by ANSI).
2377 * HOST-NAMESTRING on physical pathnames now returns a string that is
2378 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2379 (thanks to Christophe Rhodes)
2380 * The Alpha port handles icache flushing more correctly. (thanks to
2382 * More progress has been made toward bootstrapping under CLISP. (thanks
2383 to Christophe Rhodes)
2384 * The fasl file format has changed again, because dynamic loading
2385 on OpenBSD (which has non-ELF object files) motivated some cleanups
2386 in the way that foreign symbols are transformed and passed around.
2387 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2388 is no longer treated as whitespace by the reader, but instead as
2389 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2390 |A
\7fB|, instead of A as it used to.
2392 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2393 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2394 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2395 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2396 this as of 0.7.1.45, but was left out of the previous news section
2397 (thanks to Dan Barlow)
2398 * SBCL now runs on the Solaris operating system on SPARC architectures
2399 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2400 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2401 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2402 * SPARC backend cleanups, allowing builds of cores optimized for V8
2403 and V9 SPARCS, and also emission of code targeted to a particular
2404 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2406 * SBCL is closer to bootstrapping under CLISP, thanks to various
2407 fixes by Christophe Rhodes.
2408 * The fasl file format has changed again, to allow the compiler's
2409 INFO database to support symbol macros.
2410 * The user manual (in doc/) is formatted into HTML more nicely.
2411 (thanks to coreythomas)
2412 * The system is smarter about SUBTYPEP relationships, especially
2413 those involving NOT types (including types such as ATOM which are
2414 represented internally using NOT types). Thus SUBTYPEP is less
2415 likely to return (VALUES NIL NIL) in general, and in particular
2416 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2418 * The fasl file format has changed again, because the internal
2419 representation of types now includes a new slot to support the new
2420 SUBTYPEP-of-NOT-types logic.
2421 * (not a change in the main branch of SBCL, but a related prototype
2422 which can hopefully be merged into the main branch of SBCL in the
2423 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2424 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2425 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2426 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2427 dispatch macro character combinations. (thanks to Alexey Dejneka)
2428 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2429 that superficially look like logical namestrings correctly.
2430 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2432 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2433 default supplied-p) for &optional and &key arguments. (thanks to
2436 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2437 * incompatible change: The compiler is now less aggressive about
2438 tail call optimization, doing it only when (> SPACE DEBUG) or
2439 (> SPEED DEBUG). (This is an incompatible change because there are
2440 programs which relied on the old CMU-CL-style behavior to optimize
2441 away their unbounded recursion which will now die of stack overflow.)
2442 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2443 for non-GENCGC systems has been increased to 20M (since that
2444 seems much closer to the likely performance optimum for modern
2445 systems than the old 4M value was)
2446 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2447 and *DEBUG-PRINT-LEVEL*
2448 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2449 of CMU CL's support for SPARC, and various endianness and other
2450 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2451 * new syntactic sugar for the Unix command line: --load foo.bar is now
2452 an alternate notation for --eval '(load "foo.bar")'.
2454 ** The system now detects stack overflow and handles it gracefully,
2455 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2456 optimization settings. (This is a good thing in general, and
2457 its introduction in this version should be particularly timely
2458 for anyone whose code fails because of suppression of tail
2460 ** The system now hunts for the C variable "environ" in a more
2461 devious way, to avoid segfaults when the C library version differs
2462 between compile time and run time. (thanks to Christophe Rhodes)
2463 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2464 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2465 ** The compiler no longer issues bogus style warnings for undefined
2466 classes in the same source file as the DEFCLASSes which defined
2467 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2469 ** fixes in CONDITION class precedence list for undefined function
2470 errors (thanks to Alexei Dejneka)
2471 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2472 correctly. (thanks to Dan Barlow)
2473 ** portability fixes aiming at bootstrapping under CLISP (thanks
2474 to Dave McDonald and Christophe Rhodes)
2475 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2476 ** fixes in type translation and and type inference (thanks to
2478 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2479 ** various fixes in the new ports (thanks to Dan Barlow)
2480 * several changes related to debugging:
2481 ** suppression of tail recursion, as noted above
2482 ** stack overflow detection, as noted above
2483 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2484 is now the default. (For some time encapsulation has been more
2485 reliable than the breakpoint-based :ENCAPSULATE NIL
2486 implementation, at least on X86 systems; and I just noticed that
2487 encapsulation also seems closer to the spirit of the ANSI
2490 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2492 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2493 up properly again. (There was a packaging bug in 0.7.0 which
2494 left their definitions in SB-SYS::LOAD-FOREIGN and
2495 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2496 vital for most things which interface to C-level interfaces,
2497 like extensions working with sockets or databases or
2498 Perl-compatible regexes or whatever, and the need to fix
2499 this bug is the main reason that 0.7.1 was released so
2501 ** DEFGENERIC is now choosier about the methods it redefines, so that
2502 reLOADing a previously-LOADed file containing DEFGENERICs does
2503 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2504 cycle now works as it should. (thanks to Alexey Dejneka)
2505 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2506 fixed by Christophe Rhodes. (He actually submitted this patch
2507 months ago, and I delayed until after 0.7.0.)
2508 ** Bug 111 (internal compiler confusion about runtime checks on
2509 FUNCTION types) was fixed by Alexey Dejneka.
2510 * Some internal cleanups (getting rid of variables which aren't
2511 needed now that the byte interpreter is gone) caused the fasl
2512 file format number to change again.
2514 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2515 * major incompatible change: The default fasl file extension, i.e. the
2516 default extension for files produced by COMPILE-FILE, has changed
2517 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2519 ** There are many changes in the implementation of the compiler.
2520 SBCL is now essentially a compiler-only implementation of ANSI
2521 Common Lisp. EVAL still "interprets" a few special cases, but
2522 almost all the interesting cases are handled by creating
2523 a LAMBDA expression, calling COMPILE on it, then calling
2524 FUNCALL on the result.
2525 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2526 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2527 Since the code is newer, there might still be some new bugs
2528 (though not as many as before Martin Atzmueller's fixes:-). But
2529 the new code is substantially simpler and clearer, and hopefully
2530 any remaining bugs will be simpler, less fundamental, and more
2531 fixable then the bugs in the old code.
2532 ** The revised compiler is still a little unsteady on its feet.
2534 *** The debugging information it produces (particularly the names
2535 of FUNCTION objects) is sometimes much less useful than what
2536 the old compiler produced.
2537 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2538 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2539 within a MACROLET) has been temporarily weakened.
2540 ** There are new compiler optimizations for various functions:
2541 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2542 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2543 *** the math functions TRUNCATE, FLOOR, and CEILING
2544 *** the function-of-all-trades COERCE
2545 Mostly these should be transparent, but there's one
2546 potentially-annoying problem (bug 117): when the compiler
2547 inline-expands a function and does type analysis on the result,
2548 it can create control paths which have type mismatches, and
2549 when it can't prove that those control paths aren't taken,
2550 it will issue WARNINGs about the type mismatches. This is
2551 a particular problem in practice for the new sequence functions.
2552 It's not clear how this should be fixed, and for now, a
2553 workaround is given in the entry for 117 in the BUGS file.
2554 ** (Because of the interaction between the two previous items --
2555 occasional inlining problems and new inline expansions -- some
2556 of the new sequence function optimizations won't really kick in
2557 completely until debugging information, and then inlining, are
2558 straightened out in some future version.)
2559 * minor incompatible changes:
2560 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2561 DIRECTORY no longer implicitly promotes NIL slots of its
2562 pathname argument to :WILD. In particular, when you ask for the
2563 contents of a directory (which you used to be able to do without
2564 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2565 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2566 ** changes in behavior that ANSI explicitly defines to be
2567 implementation dependent:
2568 *** The new compiler-only implementation still conforms with ANSI,
2569 but acts a little different than before. Besides the obvious
2570 changes in performance tradeoffs (that the cost per form passed
2571 to EVAL has gone up, and the cost per form executed by EVAL
2572 has gone down), the behavior of the system changes a little
2573 because there are no longer any interpreted function objects.
2574 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2575 e.g. doing COMPILE on the output of interactive DEFUN is
2577 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2579 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2580 has changed from (:CL) to NIL.
2581 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2582 e.g. #\Newline and #\Tab, don't have names specified in the
2583 ANSI Common Lisp standard, is now based on their ASCII symbolic
2584 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2585 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2586 are no longer used for output.
2587 ** changes in internal implementation constants:
2588 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2589 4 million. (If your application spends a lot of time GCing and
2590 you have a lot of RAM, you might want to experiment with
2591 increasing it even more.)
2592 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2593 However, almost all old code should still continue to work without
2594 immediate update, as SB-C-CALL is now a (deprecated) nickname
2596 ** Old operator names in the style DEF-FOO are now deprecated in
2597 favor of new corresponding names DEFINE-FOO, for consistency with
2598 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2599 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2600 internal symbols, but a few supported extensions like
2601 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2602 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2603 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2604 etc. as you get deeper into recursive calls to the debugger
2605 command loop, instead of the old "5]", "5]]", "5]]]"
2606 sequence. (I was motivated to do this when squabbles between
2607 ILISP and SBCL left me very deeply nested in the debugger. In the
2608 short term, this change will probably provoke more ILISP/SBCL
2609 squabbles, but hopefully it will be an improvement in the long run.)
2610 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2611 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2612 function, since declaiming return types involving aliens is
2613 (1) annoyingly messy to do by hand and (2) vital to efficient
2614 compilation of code which calls such functions.
2615 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2616 longer reexported by the SB-EXT package. They're solely useful
2617 for alien code, so it seems more logical that you should get
2618 them from the SB-ALIEN package, not in SB-EXT.
2619 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2620 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2621 features. Instead, the code that they used to control is always
2622 built into the system.
2623 * many other bug fixes
2624 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2625 advantage of the new EVAL-WHEN stuff and to clean them up in
2626 general, and they are now more ANSI-compliant in a number of
2627 ways. Martin Atzmueller is responsible for a lot of this.
2628 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2630 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2631 *** correct ERROR type for various file operations
2632 *** some fixes for Lisp streams
2633 *** DEFMETHOD syntax checking
2634 *** changing old weird representation of debug information as
2635 strings (which, among their other deficiencies, don't transform
2636 correctly when you rename packages, and don't change their
2637 print representation when you change things like *PACKAGE*
2638 and *PRINT-LENGTH*) to symbols and lists of symbols
2639 He also made several improvements and fixed several bugs in DESCRIBE.
2640 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2642 *** misbehavior of WRITE-STRING/WRITE-LINE
2643 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2644 and several other LOOP problems as well
2645 *** DIRECTORY when similar filenames are present
2646 *** DEFGENERIC with :METHOD options
2647 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2648 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2649 *** argument ordering in FIND with :TEST option
2650 *** mishandled package designator argument in APROPOS-LIST
2651 *** various problems in the backquote readmacro
2652 *** a bug in APROPOS
2653 *** probably some others that I'm not describing very well here,
2654 since the CVS log documents them by reference to sbcl-devel
2655 messages, and the SourceForge archives aren't working well.:-(
2656 ** Dan Barlow improved the Alpha port (and is making progress on the
2657 PPC port, for those of you who think different).
2658 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2659 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2660 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2661 continues to work on the SPARC port (for those of you in a position
2662 to look down upon our little PC-compatible boxes from a great height).
2663 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2664 and uses that copy, rather than the current dynamic value, when
2665 it's trying to decide whether to truncate output. Thus e.g.
2666 (let ((*print-lines* 50))
2667 (pprint-logical-block (stream nil)
2669 (let ((*print-lines* 8))
2670 (print (aref possiblybigthings i) stream)))))
2671 should now truncate the logical block only at 50 lines, instead of
2672 often truncating it at 8 lines, as it did before.
2673 * The doc/cmucl/ directory, containing old CMU CL documentation
2674 from the time of the fork, is no longer part of the base system.
2675 SourceForge has shut down its anonymous FTP service, and with it
2676 my original plan for distributing the old CMU CL documentation
2677 there. For now, if you need these files you can download an old
2678 SBCL source release and extract them from it.
2679 * The fasl file version number changed again, for dozens of reasons,
2680 some of which are apparent above.
2682 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2683 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2684 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2685 patch, so that DISASSEMBLE output is much nicer.
2686 * The code in the SB-PROFILE package now seems reasonably stable.
2687 I still haven't decided what the final interface should look like
2688 (I'd like PROFILE to interact cleanly with TRACE, since both
2689 facilities use function encapsulation) but if you have a need
2690 for profiling now, you can probably use it successfully with
2691 the current CMU-CL-style interface.
2692 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2693 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2694 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2695 link now returns the dangling link itself, and for similar
2696 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2697 link itself. (In these cases the old code signalled an error and
2698 looped endlessly, respectively.) Thus, DIRECTORY now works even
2699 in the presence of dangling and cyclic symbolic links.
2700 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2701 is now a supported extension again, since the consensus on
2702 sbcl-devel was that it can be useful for ordinary development
2703 work, not just for debugging SBCL itself.
2704 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2705 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2706 that currently-defined functions might be redefined later with
2707 different return types.
2708 * Hash tables can be printed readably, as inspired by CMU CL code
2709 of Eric Marsden and SBCL code of Martin Atzmueller.
2710 * better error handling in CLOS method combination, thanks to
2711 Martin Atzmueller porting Pierre Mai's CMU CL patches
2712 * more overflow fixes for >16Mbyte I/O buffers
2713 * A bug in READ has been fixed, so that now a single Ctrl-D
2714 character suffices to cause end-of-file on character streams.
2715 In particular, now you only need one Ctrl-D at the command
2716 line (not two) to exit SBCL.
2717 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2719 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2720 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2721 does the right thing.
2722 * The implementation of some type tests, especially for CONDITION
2723 types, is now tidier and maybe faster, due to CMU CL code
2724 originally by Douglas Crosher, ported by Martin Atzmueller.
2725 * Some math functions have been fixed, and there are new
2726 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2727 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2728 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2729 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2730 likely to be built into sbcl-0.7.0.)
2731 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2732 CL, ported to SBCL by Martin Atzmueller)
2733 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2734 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2735 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2736 * There's a new slam.sh hack to shorten the edit/compile/debug
2737 cycle for low-level changes to SBCL itself, and a new
2738 :SB-AFTER-XC-CORE target feature to control the generation of
2739 the after-xc.core file needed by slam.sh.
2740 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2741 COMPILE-FILE is no longer supported, so that now every function
2742 gets an entry point, so that block compilation looks a little
2743 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2744 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2745 returns the number of bytes consed since the system started,
2746 rather than the number consed since the first time the function
2747 was called. (The new definition parallels ANSI functions like
2748 CL:GET-INTERNAL-RUN-TIME.)
2749 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2750 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2751 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2752 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2753 * Fasl file version numbers are now independent of the target CPU,
2754 since historically most system changes which required version
2755 number changes have affected all CPUs equally. Similarly,
2756 the byte fasl file version is now equal to the ordinary
2759 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2760 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2761 is no longer recognized. I apologize for this, because it was
2762 listed in SB-EXT as a supported extension, but I found that
2763 its existing behavior was poorly specified, as well as incorrectly
2764 specified, and it looked like too much of a mess to straighten it
2765 out. I have enough on my hands trying to get ANSI stuff to work..
2766 * many patches ported from CMU CL by Martin Atzmueller, with
2767 half a dozen bug fixes in pretty-printing and the debugger, and
2768 half a dozen others elsewhere
2769 * fixed bug 13: Floating point infinities are now supported again.
2770 They might still be a little bit flaky, but thanks to bug reports
2771 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2772 as flaky as they were.
2773 * The --noprogrammer command line option is now supported. (Its
2774 behavior is slightly different in detail from what the old man
2775 page claimed it would do, but it's still appropriate under the
2776 same circumstances that the man page talks about.)
2777 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2778 are now supported, and enabled by default. Thus, the compiler can
2779 handle many floating point and complex operations much less
2780 inefficiently. (Thus e.g. you can implement a complex FFT
2782 * The compiler now detects type mismatches between DECLAIM FTYPE
2783 and DEFUN better, and implements CHECK-TYPE more correctly, and
2784 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2785 thanks to patches from Martin Atzmueller.
2786 * various fixes to make the cross-compiler more portable to
2787 ANSI-conforming-but-different cross-compilation hosts (notably
2788 Lispworks for Windows, following bug reports from Arthur Lemmens)
2789 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2790 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2791 fixed thanks to Pierre Mai's CMU CL patches.
2792 * fixes to keep the system from overflowing internal counters when
2793 it tries to use i/o buffers larger than 16M bytes
2794 * fixed bug 45a: Various internal functions required to support
2795 complex special functions have been merged from CMU CL sources.
2796 (When I was first setting up SBCL, I misunderstood a compile-time
2797 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2798 * improved support for type intersection and union, fixing bug 12
2799 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2800 more obscure bugs as well
2801 * some steps toward byte-compiling non-performance-critical
2802 parts of the system, courtesy of patches from Martin Atzmueller
2803 * Christophe Rhodes has made some debian packages of sbcl at
2804 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2805 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2806 stable, but are nonetheless usable. When he's ready, I'd be happy
2807 to add them to the SourceForge "File Releases" section. (And if
2808 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2809 * new fasl file format version number (because of changes in
2810 internal representation of (OR ..) types to accommodate the new
2811 support for (AND ..) types, among other things)
2813 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2814 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2816 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2818 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2819 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2820 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2821 * DESCRIBE now works on structure objects again.
2822 * Most function call argument type mismatches are now handled as
2823 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2824 know whether the function will be redefined before the call is
2825 executed. (The compiler could flag local calls with full WARNINGs,
2826 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2827 it doesn't keep track of enough information to know whether calls
2828 are local in this sense.)
2829 * Compiler output is now more verbose, with messages truncated
2830 later than before. (There should be some supported way for users
2831 to override the default verbosity, but I haven't decided how to
2832 provide it yet, so this behavior is still controlled by the internal
2833 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2834 src/compiler/ir1util.lisp.)
2835 * Fasl file format version numbers have increased again, because
2836 support for the Gray streams extension changes the layout of the
2837 system's STREAM objects.
2838 * The Gray subclassable streams extension now works, thanks to a
2839 patch from Martin Atzmueller.
2840 * The full LOAD-FOREIGN extension (not just the primitive
2841 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2842 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2843 but like most other programs, it defaults to copying the Unix
2844 environment from the original process instead of starting the
2845 new process in an empty environment.
2846 * Extensions which manipulate the Unix environment now support
2847 an :ENVIRONMENT keyword option which doesn't smash case or
2848 do other bad things. The CMU-CL-style :ENV option is retained
2849 for porting convenience.
2850 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2851 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2853 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2855 * A patch from Martin Atzmueller seems to have solved the SIGINT
2856 problem, and as far as we know, signal-handling now works cleanly.
2857 (If you find any new bugs, please report them!)
2858 * The system no longer defaults Lisp source file names to types
2859 ".l", ".cl", or ".lsp", but only to ".lisp".
2860 * The compiler no longer uses special default file extensions for
2861 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2862 seems to expect a single default extension for all compiled code,
2863 and there's no compelling reason to try to stretch the standard
2864 to allow two different extensions.) Instead, byte-compiled files
2865 default to the same extension as native-compiled files.
2866 * Fasl file format version numbers have increased again, because
2867 a rearrangement of internal implementation packages made some
2868 dumped symbols in old fasl files unreadable in new cores.
2869 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2870 many fewer weird special cases.
2871 * Bug #17 (differing COMPILE-FILE behavior between logical and
2872 physical pathnames) has been fixed, and some related misbehavior too,
2873 thanks to a patch from Martin Atzmueller.
2874 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2875 by Tim Moore, ported to SBCL by Martin Atzmueller.
2876 * Martin Atzmueller fixed several filesystem-related problems,
2877 including bug #36, in part by porting CMU CL patches, which were
2878 written in part by Paul Werkowski.
2879 * More compiler warnings in src/runtime/ are gone, thanks to
2880 more patches from Martin Atzmueller.
2881 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2884 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2886 * DESCRIBE now works on CONDITION objects.
2887 * The debugger now handles errors which arise when trying to print
2888 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2890 * The build system now uses an additional file, customize-target-features.lisp,
2891 to allow local modifications to the target *FEATURES* list. (The point of
2892 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2893 debugging features enabled, without having to worry about propagating it
2894 into everyone's system when I do a "cvs update".) When no
2895 customize-target-features.lisp file exists, the target *FEATURES* list
2896 should be constructed the same way as before.
2897 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2898 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2899 warn about behavior which is undefined under the ANSI spec. Note: This
2900 is specified by ANSI, but it's not very popular with programmers.
2901 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2902 macro in the SBCL sources for an example of a workaround which you
2903 might use to make portable ANSI-standard code which does what you want.
2904 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2905 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2906 when it's used as a non-toplevel form. (This is required in order
2907 to implement the DEFCONSTANT-EQX macro.)
2908 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2909 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2910 funny things, due to bugs in the implementation of EVAL-WHEN
2911 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2912 non-toplevel DEFCONSTANTs are uncommon.)
2913 * The core file version number and fasl file version number have been
2914 incremented, because the old noncompliant DEFCONSTANT behavior involved
2915 calling functions which no longer exist, and because I also took the
2916 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2917 * fixed bug 1 (error handling before read-eval-print loop starts), and
2918 redid debugger restarts and related debugger commands somewhat while
2920 ** The QUIT debugger command is gone, since it did something
2921 rather different than the SB-EXT:QUIT command, and since it never
2922 worked properly outside the main toplevel read/eval/print loop.
2923 Invoking the new TOPLEVEL restart provides the same functionality.
2924 ** The GO debugger command is also gone, since you can just invoke
2925 the CONTINUE restart directly instead.
2926 ** The TOP debugger command is also gone, since it's redundant with the
2927 FRAME 0 command, and since it interfered with abbreviations for the
2929 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2931 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2932 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2933 some porting work by Martin Atzmueller)
2934 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2935 with a boatload of patches which he ported to SBCL. Now that those
2937 ** The system tries to make sure that its low-priority messages
2938 are prefixed by semicolons, to help people who like to use
2939 syntax highlighting in their ILISP buffer. (This patch
2940 was originally due to Raymond Toy.)
2941 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2942 patches originally written by Raymond Toy.
2943 ** The compiler understands coercion between single-value and
2944 multiple-VALUES type expressions better, getting rid of some very
2945 weird behavior, thanks to patches originally by Robert MacLachlan
2946 and Douglas Crosher.
2947 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2948 lambda lists, thanks to a patch originally by Pierre Mai.
2949 ** The system no longer bogusly warns about "abbreviated type
2951 ** The compiler gets less confused by inlining and RETURN-FROM,
2952 thanks to some patches originally by Tim Moore.
2953 ** The system no longer hangs when dumping circular lists to fasl
2954 files, thanks to a patch originally from Douglas Crosher.
2955 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2956 undefined function error.
2957 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2958 in structure slot initforms) for now, documented workaround instead:-|
2959 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2960 * fixed bug 5: added stubs for various Gray stream functions called
2961 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2962 installed, at least appropriate type errors are generated
2963 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2964 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2965 consistently in DEFMETHOD forms.
2966 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2967 it doesn't seem to affect SBCL after all
2968 * The C runtime system now builds with better optimization and many
2969 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2971 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2973 * The system is now under CVS at SourceForge (instead of the
2974 CVS repository on my home machine).
2975 * The new signal handling code has been tweaked to treat register
2976 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2977 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2978 downstream, e.g. in debug-int.lisp, has implicit dependencies
2979 on the unsignedness of integer representation of machine words,
2980 and that caused the system to bomb out with infinite regress
2981 when trying to recover from type errors involving signed values,
2982 e.g. (BUTLAST '(1 2 3) -1).)
2983 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2984 (This was one of the bugs Peter Van Eynde reported back in July.)
2985 * The system now uses code inspired by Colin Walters' O(N)
2986 implementation of MAP (from the cmucl-imp@cons.org mailing
2987 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2988 inline the MAP operation, and there is more than one
2989 sequence argument to the MAP call (so that it can't just
2990 do ETYPECASE once and for all based on the type of the
2991 single sequence argument). (The old non-inline implementation
2992 of the general M-argument sequence-of-length-N case required
2993 O(M*N*N) time when any of the sequence arguments were LISTs.)
2994 * The QUIT :UNIX-CODE keyword argument has been renamed to
2995 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2996 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2998 * Raymond Wiker's patches to port dynamic loading from Linux to
2999 FreeBSD have been added.
3000 * The BUGS file is now more nearly up to date, thanks in large part
3001 to Martin Atzmueller's review of it.
3002 * The debugger now flushes standard output streams before it begins
3003 its output ("debugger invoked" and so forth).
3004 * The core version number and fasl file version number have both
3005 been incremented, because of incompatible changes in the layout
3007 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3008 used to not be called for a saved Lisp image.)
3009 * Martin Atzmueller's version of a patch to fix a compiler crash,
3010 as posted on sbcl-devel 13 September 2000, has been installed.
3011 * Instead of installing Martin Atzmueller's patch for the
3012 compiler transform for SUBSEQ, I deleted the compiler transform,
3013 and transforms for some similar consing operations.
3014 * A bug in signal handling which kept TRACE from working on OpenBSD
3016 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3017 away by constant folding
3018 * The system now defines its address space constants in one place
3019 (in the Lisp sources), and propagates them automatically elsewhere
3020 (through GENESIS and the sbcl.h file). Therefore, patching the
3021 address map is less unnecessarily tedious and error-prone. The
3022 Lisp names of address space constants have also been systematized.
3023 * CVS tags like dollar-Header-dollar have been removed from
3024 the sources, because they have never saved me trouble and
3025 they've been source of trouble working with patches and other
3026 diff-related operations.
3027 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3028 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3030 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3032 * The system has been ported to OpenBSD.
3033 * The system now compiles with a simple "sh make.sh" on the systems
3034 that it's supported on. I.e., now you no longer need to tweak
3035 text in the target-features.lisp-expr and symlinks in src/runtime/
3036 by hand, the make.sh takes care of it for you.
3037 * The system is no longer so grossly inefficient when compiling code
3038 involving vectors implemented as general (not simple) vectors (VECTOR T),
3039 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3040 which dares to use the various sequence functions on non-simple
3041 vectors, takes less of a performance hit.
3042 * There is now a primitive type predicate VECTOR-T-P
3043 to test for the (VECTOR T) type, so that e.g.
3044 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3045 can now be compiled with some semblance of efficiency. (The old code
3046 turned the type declaration into a full call to %TYPEP at runtime!)
3047 * AREF on (VECTOR T) is still not fast, since it's still compiled
3048 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3049 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3050 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3051 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3052 without making full calls to SUBTYPEP at runtime.
3053 (Some analogous efficiency issues for non-simple vectors specialized to
3054 element types other than T, or for non-simple multidimensional arrays,
3055 have not been addressed. They could almost certainly be handled the
3056 same way if anyone is motivated to do so.)
3057 * The changes in array handling break binary compatibility, so
3058 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3059 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3060 instead of (VALUES T T).
3061 * By following the instructions that Dan Barlow posted to sbcl-devel
3062 on 2 July 2000, I was able to enable primitive dynamic object
3063 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3064 functionality is not implemented (since it calls ld to resolve
3065 library references automatically, requiring RUN-PROGRAM for its
3066 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3067 try to resolve library references) is now supported.
3068 * The system now flushes the standard output streams when it terminates,
3069 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3070 them at several other probably-convenient times, e.g. in each pass of
3071 the toplevel read-eval-print loop, and after evaluating a form given
3072 as an "--eval" command-line option. (These changes were motivated by a
3073 discussion of stream flushing issues on cmucl-imp in August 2000.)
3074 * The source transform for TYPEP of array types no longer assumes
3075 that an array whose element type is a not-yet-defined type
3076 is implemented as an array of T, but instead punts, so that the
3077 type will be interpreted at runtime.
3078 * There is now some support for cross-compiling in make.sh: each of
3079 the phases of make.sh has its own script. (This should be transparent
3080 to people doing ordinary, non-cross-compile builds.)
3081 * Since my laptop doesn't have hundreds of megabytes of memory like
3082 my desktop machine, I became more motivated to do some items on
3083 my to-do list in order to reduce the size of the system a little:
3084 ** Arrange for various needed-only-at-cold-init things to be
3085 uninterned after cold init. To support this, those things have
3086 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3087 symbols with such names are now uninterned after cold init).
3088 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3089 fasl files for cold load.
3090 ** Remove the old compiler structure pooling code (which used to
3091 be conditional on the target feature :SB-ALLOC) completely.
3092 ** Redo the representation of some data in cold init to be more compact.
3093 (I also looked into supporting byte compiled code at bootstrap time,
3094 which would probably reduce the size of the system a lot, but that
3095 looked too complicated, so I punted for now.)
3096 * The maximum signal nesting depth in the src/runtime/ support code has
3097 been reduced from 4096 to 256. (I don't know any reason for the very
3098 large old value. If the new smaller value turns out to break something,
3099 I'll probably just bump it back up.)
3100 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3102 * Many, many bugs reported by Peter Van Eynde have been added to
3103 the BUGS list; some have even been fixed.
3104 * While enabling dynamic object file loading, I tried to make the
3105 code easier to understand, renaming various functions and variables
3106 with less ambiguous names, and changing some function calling
3107 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3108 * While trying to figure out how to do the OpenBSD port, I tried to
3109 clean up some of the code in src/runtime/. In particular, I dropped
3110 support for non-POSIX signal handling, added various comments,
3111 tweaked the code to reduce the number of compilation warnings, and
3112 renamed some files to increase consistency.
3113 * To support the new automatic configuration functionality in make.sh,
3114 the source file target-features.lisp-expr has been replaced with the
3115 source file base-target-features.lisp-expr and the machine-generated
3116 file local-target-features.lisp-expr.
3117 * fixed a stupid quoting error in make.sh so that using CMU CL
3118 "lisp -batch" as cross-compilation host works again
3120 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3122 * DESCRIBE no longer tries to call itself recursively to describe
3123 bound/fbound values, so that it no longer fails on symbols which are
3124 bound to themselves (like keywords, T, and NIL).
3125 * DESCRIBE now works on generic functions.
3126 * The printer now prints less-screwed-up representations of closures
3127 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3128 * A private symbol is used instead of the :EMPTY keyword previously
3129 used to mark empty slots in hash tables. Thus
3130 (DEFVAR *HT* (MAKE-HASH-TABLE))
3131 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3132 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3133 now does what ANSI says that it should. (You can still get
3134 similar noncompliant behavior if bang on the hash table
3135 implementation with all the symbols you get back from
3136 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3137 This breaks binary compatibility, since tests for equality to
3138 :EMPTY are wired into things like the macroexpansion of
3139 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3141 * There's now a minimal placeholder implementation for CL:STEP,
3142 as required by ANSI.
3143 * An obscure bug in the interaction of the normal compiler, the byte
3144 compiler, inlining, and structure predicates has been patched
3145 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3146 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3148 * Missing ordinary arguments in a macro call are now detected even
3149 when the macro lambda list contains &KEY or &REST.
3150 * The debugger no longer complains about encountering the top of the
3151 stack when you type "FRAME 0" to explicitly instruct it to go to
3152 the top of the stack. And it now prints the frame you request even
3153 if it's the current frame (instead of saying "You are here.").
3154 * As specified by ANSI, the system now always prints keywords
3155 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3157 * The default initial SIZE of HASH-TABLEs is now smaller.
3158 * Type information from CLOS class dispatch is now propagated
3159 into DEFMETHOD bodies, so that e.g.
3160 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3162 is now basically equivalent to
3163 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3164 (DECLARE (TYPE SINGLE-FLOAT X))
3166 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3167 operation, without having to do run-time type dispatch.
3168 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3169 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3170 * Since I don't seem to be making big file reorganizations very often
3171 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3172 is overflowing my ability to conveniently back them up), I've finally
3173 checked the system into CVS. (The CVS repository is on my home system,
3174 not at SourceForge -- putting it on SourceForge might come later.)
3175 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3176 high-level GC-NOTIFY-FOO functions send their output. (There's
3177 still very little control of where low-level verbose GC functions
3178 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3179 less than it used to -- the GC-NOTIFY-FOO functions are now under
3180 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3181 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3182 in only one place in the source code, and propagates it automatically
3183 everywhere that it's needed. Thus e.g. when I bump the version from
3184 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3185 * The C source files now include boilerplate legalese and documentation
3186 at the head of each file (just as the Lisp source files already did).
3187 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3188 to his page will be replaced with a link to his new CLiki service.
3190 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3192 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3193 * The build process now looks for GNU make under the default name "gmake",
3194 instead of "make" as it used to. If GNU make is not available as "gmake"
3195 on your system, you can change this default behavior by setting the
3196 GNUMAKE environment variable.
3197 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3198 can build without error under CMU CL.
3200 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3202 * There is now a partial SBCL user manual (with some new text and some
3203 text cribbed from the CMU CL manual).
3204 * The beginnings of a profiler have been added (starting with the
3205 CMU CL profiler and simplifying and cleaning up). Eventually the
3206 main interface should be through the TRACE macro, but for now,
3207 it's still accessed through vaguely CMU-CL-style functions and macros
3208 exported from the package SB-PROFILE.
3209 * Some problems left over from porting CMU CL to the new
3210 cross-compilation bootstrap process have been cleaned up:
3211 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3212 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3213 part when getting rid of the compiler *BACKEND* stuff.)
3214 ** The value of *NULL-TYPE* was screwed up, because it was
3215 being initialized before the type system knew the final
3216 definition of the 'NULL type. This screwed up several key
3217 optimizations in the compiler, causing inefficiency in all sorts
3218 of places. (I found it because I wanted to understand why
3219 GET-INTERNAL-RUN-TIME was consing.)
3220 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3221 PROCLAIM FTYPE information. Unfortunately this broke binary
3222 compatibility again, since now the forms output by DEFGENERIC
3223 to refer to functions which didn't exist in 0.6.3.
3224 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3225 can use the new (as of 0.6.3) transform for SOME into MAP into
3227 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3228 values to (INTEGER 0 1000000), so that the time code will no longer
3229 occasionally get blown up by Linux returning 1000000 microseconds
3230 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3231 its output conform to the ANSI spec. (Alas, this makes its output
3232 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3233 * A full call to MAP NIL with a single sequence argument no longer conses.
3234 * fixes to problems pointed out by Martin Atzmueller:
3235 * The manual page no longer talks about multiprocessing as though
3236 it were currently supported.
3237 * The ILISP support patches have been removed from the distribution,
3238 because as of version 5.10.1, ILISP now supports SBCL without us
3239 having to maintain patches.
3240 * added a modified version of Raymond Toy's recent CMU CL patch for
3241 EQUALP comparison of HASH-TABLE
3243 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3245 * The system still can't cross-compile itself with
3246 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3247 entails), but at least it can compile more of itself that way
3248 than it used to be able to, and various buglets which were uncovered
3249 by trying to cross-compile itself that way have now been fixed.
3250 * This release breaks binary compatibility again. This time
3251 at least I've incremented the FASL file format version to 2, so that the
3252 problem can be detected reliably instead of just causing weird errors.
3253 * various new style warnings:
3254 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3255 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3256 ** using the lexical binding of a variable named in the *FOO* style
3257 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3258 as specified by ANSI.
3259 * *RANDOM-STATE* is no longer automatically initialized from
3260 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3261 default behavior of the system is to repeat its behavior every time
3262 it's run. If you'd like to change this behavior, you can always
3263 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3264 old convention there was no comparably easy way to get the system to
3265 repeat its behavior every time it was run.
3266 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3267 FTYPE declarations has been removed, in favor of their ANSI
3268 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3269 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3270 the types of their sequence arguments can be determined at compile time.
3271 This is done through a new open code expansion for MAP which eliminates
3272 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3273 argument types can be determined at compile time.
3274 * The optimizer now transforms COERCE into an identity operation when it
3275 can prove that the coerced object is already of the correct type. (This
3276 can be a win for machine generated code, including the output of other
3277 optimization transforms, such as the MAP transform above.)
3278 * Credit information has been moved from source file headers into CREDITS.
3279 * Source file headers have been made more standard.
3280 * The CASE macro now compiles without complaining even when it has
3283 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3285 * (Note that the way that the PCL macroexpansions were rewritten
3286 to accommodate the change in DEFGENERIC below breaks binary
3287 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3288 not run under sbcl-0.6.2. Once we get out of alpha releases,
3289 i.e. hit release 1.0.0, we'll probably try to maintain binary
3290 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3291 and sbcl-1.4.4. Until then, however, it might be fairly common
3292 for maintenance releases to break binary compatibility.)
3293 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3295 * The reporting of unhandled signals has been changed to print some
3296 explanatory text as well as the report form. (Previously only
3297 the report form was printed.)
3298 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3299 it defines, so that the compiler no longer issues undefined function
3300 warnings for compiled-but-not-yet-loaded generic functions.
3301 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3302 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3303 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3304 section of the ANSI spec are supported.
3305 * The "" nickname for the "KEYWORD" package has been removed.
3306 The reader still handles symbol tokens which begin with a package marker
3307 as keywords, but it doesn't expose its mechanism for doing so in the
3308 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3309 * The system now issues STYLE-WARNINGs for contradictory TYPE
3310 proclamations. (Warnings for contradictory FTYPE proclamations would
3311 be nice too, but those can't be done usefully unless the type system
3312 is made smarter about FUNCTION types.)
3313 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3314 systematized, so that "*target-*.lisp is supposed to exist only on the
3315 target and imply that there's a related file which exists on the
3316 host, and *host-*.lisp is supposed to exist only on the host and imply
3317 that there's a related file which exists on the target. This involves a
3318 lot of renaming. Hopefully the acute confusion caused by the renaming
3319 will be justified by the reduction in chronic confusion..
3320 ** runtime-type.lisp -> early-target-type.lisp
3321 ** target-type.lisp -> late-target-type.lisp
3322 ** early-host-format.lisp -> early-format.lisp
3323 ** late-host-format.lisp -> late-format.lisp
3324 ** host-error.lisp -> misc-error.lisp
3325 ** early-error.lisp -> early-target-error.lisp
3326 ** late-error.lisp -> late-target-error.lisp
3327 ** host-defboot.lisp -> early-defboot.lisp
3328 ** code/misc.lisp -> code/target-misc.lisp
3329 ** code/host-misc.lisp -> code/misc.lisp
3330 ** code/numbers.lisp -> code/target-numbers.lisp
3331 ** code/early-numbers.lisp -> numbers.lisp
3332 ** early-host-type.lisp -> early-type.lisp
3333 ** late-host-type.lisp -> late-type.lisp
3334 ** host-typep.lisp -> typep.lisp
3335 ** load.lisp -> target-load.lisp
3336 ** host-load.lisp -> load.lisp
3337 ** host-disassem.lisp -> disassem.lisp
3338 ** host-insts.lisp -> insts.lisp
3339 ** byte-comp.lisp -> target-byte-comp.lisp
3340 ** host-byte-comp.lisp -> byte-comp.lisp
3341 ** host-signal.lisp -> signal.lisp
3342 ** host-defstruct.lisp -> defstruct.lisp
3343 ** late-target-type.lisp -> deftypes-for-target.lisp
3344 Furthermore, several other previously target-only files foo.lisp (e.g.
3345 hash-table.lisp and random.lisp) have been split into a target-and-host
3346 foo.lisp file and a target-only target-foo.lisp file, with their key type
3347 definitions in the target-and-host part, so that the cross-compiler will
3348 know more about target types.
3349 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3350 gone away. In their place are various *BACKEND-FOO* variables
3351 corresponding to the slots of the old structure.
3352 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3353 into the target SBCL has been fixed.
3354 * The chill.lisp system for loading cold code into a running SBCL
3356 * Support for the CMU CL "scavenger hook" extension has been removed.
3357 (It was undocumented and unused in the CMU CL sources that SBCL was
3358 derived from, and stale in sbcl-0.6.1.)
3359 * Various errors in the cross-compiler type system were detected
3360 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3361 (enabling various consistency checks). Many of them were fixed,
3362 but some hard problems remain, so the compiler is back to
3363 running without *TYPE-SYSTEM-INITIALIZED* for now.
3364 * As part of the cross-compiler type system cleanup, I implemented
3365 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3366 * I have started adding UNCROSS calls throughout the type system
3367 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3368 on cross-compiler input files will be able to go away, and various
3370 * CONSTANTP now returns true for quoted forms (as explicitly required
3373 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3375 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3376 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3377 (MAKE-LIST -1). In the longer term, it ought to become true
3378 that declarations are assertions even at SAFETY 1. For now, it's not
3379 quite true even at SAFETY 3, but it's at least more nearly true..
3380 (Note that this change seems to increases the size of the system by
3381 O(5%) and to decrease the speed of the compiler by 20% or more.)
3382 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3383 for the problem of printing dozens of lines of distracting information
3384 about low-level system machinery as part of the top stack frame
3385 on entry to the debugger when an undefined function was called.
3386 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3387 is not reset to COMMON-LISP-USER.
3388 * Compilation of stuff related to dyncount.lisp has been made conditional
3389 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3390 smaller. The various dyncount-related symbols have been moved into
3391 a new "SB-DYNCOUNT" package.
3392 * tty-inspect.lisp has been renamed to inspect.lisp.
3393 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3394 feature has gone away. (When we eventually port to other flavors of
3395 libc and/or Unix, we'll try to make the differences between flavors
3396 invisible at the user level.)
3397 * Various other *FEATURES* tags, and/or their associated conditionals,
3398 have been removed if obsolescent, or given better documentation, or
3399 sometimes given more-mnemonic names.
3401 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3403 * tidied up "make.sh" script
3404 * tidied up system directory structure
3405 * better "clean.sh" behavior
3406 * added doc/FOR-CMUCL-DEVELOPERS
3407 * many many small tweaks to output format, e.g. removing possibly-confusing
3408 trailing #\. character in DESCRIBE-INSTANCE
3409 * (EQUALP #\A 'A) no longer signals an error.
3410 * new hashing code, including EQUALP hashing
3411 * tidied up Lisp initialization and toplevel
3412 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3413 * command line argument processing
3414 * added POSIX-GETENV function to deal with Unix-ish environment variables
3415 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3416 terminating SBCL on EOF
3417 * non-verbose GC by default
3418 * There is no more "sbcl" shell script; the sbcl file is now the C
3419 runtime executable (just like CMU CL).
3420 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3422 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3423 symbol and package information as Lisp native symbol and package objects
3424 instead of strings naming symbols and strings naming packages. This way,
3425 whenever packages are renamed (as in warm init), debug information is
3426 transformed along with everything else.
3427 * tweaked the optimization policy declarations which control the building
3428 of SBCL itself. Now, among other things, the system no longer saves
3429 source location debugging information. (This helps two problems at once
3430 by reducing SBCL size and by keeping SBCL from trying to look for its
3431 sources -- which may not exist -- when reporting errors.)
3432 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3433 debugging and testing purposes
3434 * cleaned up printing, making the printer call PRINT-OBJECT for
3435 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3436 methods, giving nearly-ANSI behavior
3437 * converted almost all special variables to use *FOO* naming convention
3438 * deleted PARSE-TIME functionality, since it can be done portably
3439 * moved some files out of cold init into warm init
3440 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3442 * regularized formatting of source files
3443 * added an install.sh script
3444 * fixed ridiculous memory usage of cross-compiler by making
3445 compiler/alloc.lisp not try to do pooling unless it can hook
3446 itself into the GC of the cross-compilation host. Now the system
3447 builds nicely on my old laptop.
3448 * added :SB-ALLOC in target-features.lisp-expr
3449 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3450 was not implemented)
3451 * re-did condition handling and note reporting in the compiler. Notes
3452 are no longer handled by signalling conditions. Style warnings
3453 and warnings are handled more correctly and reported in such a way
3454 that it's easy to find one or the other in your output (so that you
3455 can e.g. figure out which of many problems caused COMPILE-FILE to
3457 * changed the severity of several compiler warnings from full WARNING
3458 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3459 compiler note reporting so that it doesn't use the condition system
3460 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3461 COMPILE-FILE command)
3462 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3463 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3464 no longer supported, since their implementation was deeply intertwingled
3465 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3466 * removed bogus "support" for compiler macros named (SETF FOO), and
3467 removed the compiler macro for SETF INFO (but only after making a fool
3468 of myself on the cmucl-imp mailing list by posting a bogus patch for
3469 DEFINE-COMPILER-MACRO..)
3470 * Compiled files containing forms which have side effects on the Lisp
3471 reader (such as DEFPACKAGE forms) are now handled more correctly.
3472 (Compiler queuing of top level lambdas has been suppressed by setting
3473 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3474 * deleted various currently-unused source files, e.g. gengc.lisp. They
3475 may be added back at some point e.g. when porting to other architectures,
3476 but until they are it's distracting to distribute them and to try to
3478 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3479 there were so many of them they're just distractions, and UNCROSS is
3480 known to be able to handle the current sources
3481 * moved PROFILE functionality into TRACE, so that it will be clear
3482 how the wrapping and unwrapping of functions when you profile them
3483 interacts with the wrapping and unwrapping of functions when you
3484 trace them. (Actually, the functionality isn't there yet, but at least
3485 the interface specification is there. Hopefully, the functionality will
3486 arrive with some future maintenance release.)
3487 * removed host-oops.lisp
3488 * changed signature of QUIT function to allow UNIX-CODE argument
3489 * fixed READ-SEQUENCE bug
3490 * tweaked verbose GC output so that it looks more like the progress
3491 output that ANSI specifies for functions like LOAD
3492 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3493 * added <http://sbcl.sourceforge.com> to the banner information printed by