1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
3 * new feature: Unicode character names are now known to the system
4 (through CHAR-NAME and NAME-CHAR).
5 * bug fix: as implied by AMOP, standardized classes no longer have
6 slots named by external symbols of public packages. (reported by
8 * optimization: calling functions via constant symbols -- as in
9 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
10 via the function object as in (FUNCALL #'FOO).
12 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
13 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
14 be used for bundling the runtime and the core file into one
15 executable binary. This feature is not currently supported on all SBCL
16 platforms. (thanks to James Bielman and NIIMI Satoshi)
17 * new feature: a generational or ephemeral garbage collector is now
18 the default on the PowerPC platform (both Linux and Darwin). The
19 old Cheney (stop and copy) collector is a build-time option.
20 (thanks to Cyrus Harmon, after Raymond Toy)
21 * minor incompatible change: the method by which SBCL finds its
22 contributed modules has changed; it no longer relies on symbolic
23 links from an $SBCL_HOME/systems directory, but searches directly
24 in the subdirectories of $SBCL_HOME.
25 * enhancement: the dynamic heap size on the Linux/PPC platform is
26 markedly larger, even using the older Cheney garbage collector.
27 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
28 certain complicated string types. (reported by Gary King)
29 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
30 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
31 * fixed bug: DOCUMENTATION on structures no longer signals an
32 error if no structure type of the right name exists. (reported by
34 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
35 applicable non-standard primary or :AROUND method on
36 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
37 no longer cause unbound variable errors. (reported by Kevin Reid)
38 * fixed bug: metacircle resolution in cases where methods have slots
39 added before the slots from STANDARD-METHOD. (reported by Jean
41 * fixed bug: the Power PC assembly code for calling into lisp
42 sometimes computed its return address wrongly (depending on the
43 15th bit of the address, fixed at link time). This used to
44 manifest itself in a segmentation violation while building PCL.
45 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
46 many others over the years)
47 * fixed bug: ROOM no longer reports silly numbers for stack usage if
48 the stack pointer (treated as a signed integer) is negative.
49 (thanks to Peter van Eynde)
51 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
52 * new platform: experimental support for the Windows operating
53 system has been added. (thanks to Alastair Bridgewater)
54 * fixed several bugs in and robustified the PPC FFI (including
55 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
56 * bug fix: allow non-simple string symbol names (reported by Paul
58 * bug fix: interrupt handling on NetBSD (thanks to Richard M
60 * bug fix: saving a core corrupted callbacks on x86/x86-64
61 * bug fix: closed a loophole in metacircularity detection and
62 grounding in the PCL implementation of CLOS.
63 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
64 (reported by tomppa on #lisp)
65 * bug fix: clear the direction flag on Lisp -> C transitions, as
66 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
67 (reported by Andrej Grozin and Hendrik Maryns)
68 * optimization: major improvements to GC efficiency on GENCGC platforms
69 * optimization: faster implementation of EQUAL
70 * optimization: emit more efficient opcodes for some common
71 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
72 * optimization: several other minor code-generation improvements on x86-64
73 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
75 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
76 * minor incompatible change: (SETF CLASS-NAME) and (SETF
77 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
78 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
79 by ANSI as a generic function, and for consistency (SETF
80 GENERIC-FUNCTION-NAME) is treated likewise.
81 * fixed bug #233.b: make constraint propagation notice when a variable
82 value is changed after it is referenced but before it is used
83 * fixed bug #296: no more arbitrary behaviour from filenames with
84 odd characters as --load arguments. (reported by Adam Warner)
85 * fixed bug #390: :CHARACTER-SET pathname components now work as
86 expected. (reported by Tim Daly Jr)
87 * fixed bug #391: complicated :TYPE intersections in slot
88 definitions no longer cause an error in PCL internals.
89 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
90 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
91 returning the number of octets which would be written to the
92 file-stream. (thanks to Robert J. Macomber)
93 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
94 arguments to the internal stream functions. (thanks to David
96 * bug fix: stack exhaustion is now handled on the x86/SunOS
97 platform. (thanks to Daisuke Homma)
98 * bug fix: more accurate ROOM results on GENCGC platforms
99 * optimization: improved type inference for arithmetic-for
100 index variables in LOOP
101 * optimization: faster floating-point SQRT on x86-64
102 * fixed some bugs revealed by Paul Dietz' test suite:
103 ** DOCUMENTATION returns NIL instead of "" for method combinations
104 that don't have a docstring
106 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
107 * minor incompatible change: (SETF CLASS-NAME) and (SETF
108 GENERIC-FUNCTION-NAME) are no longer generic functions, and
109 therefore are not customizeable by user code (as seems to be at
110 least permitted and maybe required by AMOP). As a consolation,
111 however, the SBCL implementation of these functions now calls
112 REINITIALIZE-INSTANCE as specified by AMOP.
113 * bug fix: slot-definition documentation is propagated as per ANSI
114 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
115 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
116 argument, as do eslotd initialization methods. (from Pascal
117 Costanza's "Closer" project)
118 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
119 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
121 * bug fix: it is now possible to have more than one subclass of
122 STANDARD-GENERIC-FUNCTION without causing stack overflow.
123 (reported by Bruno Haible, Pascal Costanza and others)
124 * bug fix: the dependent update protocol now works for generic
125 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
127 * bug fix: condition-class instances corresponding to
128 DEFINE-CONDITION forms are now created eagerly. (reported by
129 Kalle Olavi Niemitalo on comp.lang.lisp)
130 * bug fix: floating point printing is more accurate in some
131 circumstances. (thanks to Simon Alexander)
132 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
133 merged with *DEFAULT-PATHNAME-DEFAULTS*.
134 * bug fix: callbacks on OS X now preserve stack-alignment.
135 (thanks to Cyrus Harmon)
136 * enhancement: the x86-64 disassembler is much better at
137 disassembling SSE instructions. (thanks to Lutz Euler)
138 * enhancement: CHECK-TYPE on a variable now allows type-inference
139 (thanks to Vincent Arkesteijn)
140 * optimization: improved performance of EUC-JP external format.
141 (thanks to NIIMI Satoshi)
142 * optimization: performance improvements to IO on file streams of
143 :ELEMENT-TYPE CHARACTER
144 * optimization: much faster memory allocation on x86-64
145 * optimization: faster garbage collections (latency and throughput)
147 * optimization: faster bignum subtraction and division on x86 and x86-64
148 (thanks to Lutz Euler)
149 * optimization: more accurate type-inference for loops that use a
150 floating point index variable or a negative step.
152 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
153 * bug fix: add a workaround to SBCL looping infinitely at startup on
154 Linux kernels with apparently buggy implementations of personality().
155 (thanks to Svein Ove Aas)
156 * bug fix: Unicode symbols are correctly printed in LDB backtraces
157 (thanks to David Lichteblau)
158 * bug fix: local bindings shadow global symbol macros for
159 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
160 * optimization: non-open coded uses of numeric comparison operators
161 (e.g. >) no longer cons when called with more than one parameter
162 on platforms supporting dynamic-extent allocation.
163 * enhancement: saving cores with foreign code loaded is now supported
164 on MIPS/Linux in addition to the previously supported platforms.
165 * bug fix: FIND-RESTART now tests for activity, not applicability when given
166 a restart object as identifier. (reported by Helmut Eller for CMUCL)
167 * bug fix: division by zero in sb-sprof when no samples were collected
168 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
169 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
170 (thanks to Alexander Kjeldaas)
171 * new feature: ignoring the return values of destructive functions
172 like NREVERSE signals a compile-time style-warning.
173 (thanks to Kevin Reid)
174 * new feature: the alignment of alien structure fields can be
175 explicitly specified. (thanks to Cyrus Harmon)
176 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
177 and dump core on SIGQUIT
179 ** incompatible change: threads do not inherit values of specials
180 from their parents (see manual)
181 ** bug fix: threads stacks belonging to dead threads are freed by the
182 next exiting thread, no need to gc to collect thread stacks anymore
183 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
184 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
185 ** bug fix: dynamic variable and thread start related gc lossage
186 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
188 * fixed some bugs revealed by Paul Dietz' test suite:
189 ** SUBTYPEP is slightly more accurate on heinously complicated
190 CONS types where some of the members have uncertain (in the
191 NIL, NIL sense) type relationships to each other.
193 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
194 no longer purified unless :PURIFY T is explicitly specified.
195 ** Non-purified cores are significantly smaller than before
197 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
198 * new feature: timers based on Zach Beane's excellent timer package
199 * added support for the following external formats: koi8-u,
200 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
201 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
202 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
203 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
204 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
205 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
206 * incompatible change: a threaded SBCL will no longer revert to
207 non-threaded mode on non-NPTL systems, but refuse to start entirely.
208 * bug fix: interrupts are disabled until startup is complete; no
209 more sigsegvs when receiving a signal to soon
210 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
212 * bug fix: add a workaround for the memory randomization features in
213 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
214 workaround will only be in effect on systems with the proc filesystem
216 * bug fix: printing objects of type HASH-TABLE signals a
217 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
219 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
220 have been running for over 50 days. (reported by Gilbert Baumann)
221 * bug fix: the logic for getting names of functions gets less
222 confused when confronded with alternate-metaclass
223 funcallable-instances. (reported by Cyrus Harmon)
224 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
225 consistent, even on internal alternate-metaclass objects.
226 * bug fix: flush closure information collected by physical
227 environment analysis prepass before the main pass. (bug reported
229 * bug fix: compiler pack phase does not modify a hashtable, which is
230 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
231 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
232 may be expanded. (reported by Alexander <malishev> on c.l.l)
233 * bug fix: no more style-warnings when compiling code using LAST
234 under high-SPEED low-SPACE settings. (reported by David Wallin
236 * fixed some incompatibilities between SBCL's MOP and the MOP
238 ** the METAOBJECT class is now implemented;
239 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
240 STANDARD-OBJECT, as required;
241 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
242 now compatible (as required by VALIDATE-SUPERCLASS); there
243 remains a consistency requirement over the presence or absence
244 of the FUNCTION class in the superclass list of finalized
245 classes; see the manual for more details;
246 ** the :AROUND method for COMPUTE-SLOTS on
247 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
248 requested slot ordering.
250 ** bug fix: parent thread now can be gc'ed even with a live
252 ** bug fix: binding a special with PROGV to no value is not
253 the same as that symbol not having been bound (thanks to
255 ** bug fix: binding specials is thread safe (thanks to
257 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
258 and SB-SPROF do not die with 'no handler for signal XX in
259 interrupt_handle_now(..)' anymore
260 ** bug fix: WITH-TIMEOUT works with multiple running threads
261 * fixed some bugs revealed by Paul Dietz' test suite:
262 ** ENSURE-GENERIC-FUNCTION should take a method class object for
263 the :method-class keyword argument.
265 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
266 * new port: the Solaris operating system on x86 processors is now
267 mostly supported, though some rough edges in the environment
268 remain. (thanks to Daisuke Homma)
269 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
270 space for its heap. (thanks to Thiemo Seufer)
271 * optimization: SBCL on MIPS platforms now supports dynamic-extent
272 closures. (thanks to Thiemo Seufer)
273 * minor incompatible change: eof selects abort in the debugger.
274 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
275 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
277 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
278 deprecated, and will go away in a future revision of SBCL.
279 * minor incompatible change: GC-ON and GC-OFF are no longer
280 implemented with a counter, it does not matter how many times gc
281 is switched on or off
282 * bug fix: discriminating functions for generic function classes
283 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
284 make invalid assumptions about method precedence order. (reported
286 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
287 platforms. (thanks to Thiemo Seufer)
288 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
289 * bug fix: correct pathname printing with printer escaping is on.
290 (thanks to Kevin Reid)
291 * bug fix: complex VOP definitions in "user-space" no longer trigger
292 package locks. (reported by Zach Beane)
293 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
294 for funcallable-instance functions, and is no different from
296 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
297 characters in the current readtable. (reported by Nicholas Neuss)
298 * bug fix: SUBTYPEP on various CONS types returns more a more
299 accurate acknowledgment of its certainty.
300 * optimizations: REMOVE-DUPLICATES now runs in linear time on
301 lists in some cases. This partially fixes bug 384.
302 * flush all standard streams before prompting in the REPL and the
304 * bug fix: signal handling and triggering gc do not conflict
305 directly anymore, in particular a high frequency sb-sprof does
306 not prevent gc from running
307 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
308 approximation for timezone and DST information between the
309 universal time epoch and the smallest negative 32-bit time_t.
310 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
311 year 1899 so long as the encoded time is non-negative.
312 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
313 the correct answer for non-constant rotate amounts when performing
314 an inline 32-bit rotation.
316 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
317 there is only one thread in the session
318 ** bug fix: memory leak for streams created in one thread and
319 written to in another
320 ** bug fix: lockup when compiled with gcc4
321 ** bug fix: race that allows the gc to be triggered when gc is
323 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
324 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
326 ** bug fix: finished threads are gc'ed properly
327 * fixed some bugs revealed by Paul Dietz' test suite:
328 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
329 has arguments with a different set of applicable methods from
330 the orignal arguments.
331 ** The type error thrown by MAP now has a correct expected-type
333 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
334 name a compiled function.
335 ** Three MISC tests where a large bignum was improperly coerced to
336 a float that couldn't represent that bignum during type
337 derivation were fixed.
338 ** SUBTYPEP can now handle types involving the negation of a
339 list-form FUNCTION type.
340 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
341 as (COMPLEX (AND RATIO (NOT FIXNUM))).
342 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
344 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
345 * New feature: Experimental support for bivalent streams: streams
346 opened with :element-type :default now allow character and binary
347 (unsigned-byte 8) I/O
348 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
349 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
351 * bug fix: no more highly sporadic "couldn't check whether ~S is
352 readable" when reading a stream and an interrupt hits in the middle
353 of a select system call
354 * compiler better recognizes complex arrays (reported by Eduardo
356 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
357 for variables on SBCL built with linkage-tables. (reported by Luis
359 * various error reporting improvements.
360 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
361 (Thanks to Thiemo Seufer)
362 * bug fix: nested reader invokations work correctly
363 * bug fix: it is possible to have simultaneous references to foreign
364 code and foreign data with the same name.
366 ** added x86-64 support
367 ** incompatible change: the threading api now works with thread
368 objects instead of thread ids
369 ** bug fix: threads are protected from signals and interruption when
370 starting up or going down
371 ** bug fix: a race where an exiting thread could lose its stack to gc
372 ** bug fix: don't halt on infinite error in threads if possible
373 ** fixed numerous gc deadlocks introduced in the pthread merge
374 ** bug fix: fixed thread safety issues in read and print
375 ** bug fix: debugger doesn't hang on session lock if interrupted at
376 an inappropriate moment
377 ** bug fix: run-program is now thread safe(r)
378 ** bug fix: inner with-recursive-lock no longer releases the mutex
379 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
380 * fixed some bugs revealed by Paul Dietz' test suite:
381 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
382 EXPECTED-TYPE slots filled.
383 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
385 ** MAKE-CONDITION accepts classes as type-designators.
386 ** COMPILE may never return NIL.
387 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
388 range before calling Unix time functions
390 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
391 * numerous signal handling fixes to increase stability
392 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
393 * minor incompatible change: we now correctly canonize default
394 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
395 as per the MOP, rather than the historical (INITARG INITFUNCTION
396 INITFORM). (reported by Bruno Haible)
397 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
398 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
399 for more information.
400 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
401 pathname is a directory pathname.
402 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
403 * dynamic space size on PPC has been increased to 768Mb. (thanks to
405 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
406 :METACLASS argument in addition to a class name. (reported by
407 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
408 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
409 (reported by Bruno Haible; patch from Gerd Moellmann)
410 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
412 * bug fix: more cleanups to the floating point exception handling on
413 x86-64 (thanks to James Knight)
414 * bug fix: the compiler does not try to constant fold calls of
415 COERCE and BIT-* functions when they return freshly constructed
416 objects. (reported by Nikodemus Siivola and Paul Dietz)
417 * optimization: improved the MIPS versions of generic (in the
418 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
419 * optimization: direct conversion of (unsigned-byte 32) to floats on
420 the PowerPC platform.
421 * optimization: structure instances with raw slots now use less
422 memory, and probably show better memory locality. (thanks to
424 * optimization: DYNAMIC-EXTENT declarations for lists and closures
425 are treated as requests for stack allocation on the x86-64,
426 Alpha-32, PPC and SPARC platforms.
427 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
428 the wrong file descriptor; implementation of SOCKET-OPEN-P.
429 (thanks to Tony Martinez)
431 ** gcing a dead thread can no longer lead to lockups
432 ** threads block signals until they are set up properly
433 ** errno is no longer shared by threads
434 ** interrupt-thread restores the eflags register on x86
435 ** fixed some lockups due to gc/thread interaction
436 * fixed some bugs revealed by Paul Dietz' test suite:
437 ** invalid dotted lists no longer raise a read error when
439 ** signal an error if a symbol that names a declaration is used
440 as the name of a type, or vice versa
441 ** allow using the (declare (typespec var*)) abbreviation for
442 (declare (type typespec var*)) with all type specifiers
443 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
444 ** always return NIL from PPRINT-POP when OBJECT is NIL
445 ** don't signal errors when pretty-printing malformed LABELS,
446 FLET or MACROLET forms
447 ** declarations in a DOLIST body are also in scope for the
449 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
451 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
454 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
455 * fixed cross-compiler leakages that prevented building a 32-bit
456 target with a 64-bit host compiler.
457 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
458 opened with :IF-EXISTS :OVERWRITE.
459 * fixed bug 281: error for an invalid qualifier in a short-form method
460 combination method is not signalled until the faulty method is called.
461 * bug fix: iteration variable type inferrer failed to deal with open
462 intervals. (reported by Alan Shields)
463 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
464 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
465 or more runtime options were provided to the sbcl binary.
466 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
467 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
468 * optimization: there's now a fast-path for fixnum arguments in the
469 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
470 * optimization: the code generated on x86-64 is more compact thanks
471 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
472 * bug fix: floating-point exception handling now partly works on
473 x86-64. (Thanks to James Knight)
474 * improvement to the MIPS backend from Thiemo Seufer: C-style
475 64-bit long long arguments and return values to alien functions
477 * fixed some bugs revealed by Paul Dietz' test suite:
478 ** the type-error signalled from WARN has a filled-in DATUM slot.
479 ** the type-error required when a stream is not associated with
480 a file has the stream as its datum.
481 ** type-errors on single-floats on x86-64 no longer have
482 :INVALID-OBJECT as the datum
483 ** the type-errors signalled for invalid function names now have
484 a correct expected type
485 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
486 for structures defined with a :TYPE. Documentation strings for
487 typed structures are no longer immediately discarded
488 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
489 broadcast streams and synonym streams. FILE-LENGTH now also works
490 on broadcast streams.
492 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
493 * incompatible change: the --noprogrammer option, deprecated since
494 version 0.7.5, has been removed. Please use the equivalent
495 --disable-debugger option instead.
496 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
498 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
499 inoperational for a while now) has been completely removed.
500 * null lexical environments are now printed as #<NULL-LEXENV>,
501 significantly reducing the amount of clutter in typical backtraces.
502 * documentation on weak pointers, finalization, and after GC hooks
503 has been added to the manual.
504 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
505 as other specialized array types, is much faster. SUBSEQ and
506 COPY-SEQ on such arrays have also been sped up.
507 * optimization: EQL is now more efficient when at least other argument
508 is known to be of type (OR FIXNUM (NOT NUMBER)).
509 * fixed bug: the runtime is now less vulnerable to changes in the
510 size of the SBCL object on OS X, and virtual memory is reserved for
511 all spaces that need to be at a fixed address.
512 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
513 * fixed bug: finalizers and after GC hooks that cause consing are now
515 * fixed bug: compiler error messages and summaries are now printed to
516 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
517 * fixed inference of the upper bound of an iteration variable.
518 (reported by Rajat Datta).
519 * fixed bug 211e: calling local functions with duplicated constant
520 keyword argument no longer causes a bogus style warning about an
522 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
523 ftype declaration to be disregarded. (reported by Dave Roberts)
524 * fixed bug 373: caused by erronous compilation of references to alien
525 variables in the runtime on ppc/darwin.
526 * fixed bug 376: CONJUGATE type deriver.
527 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
528 (reported by Baughn on #lisp)
529 * a cross-compiler bug on non-x86 platforms has been identified and
530 fixed. (thanks to Bruno Haible)
531 * improvements to the MIPS runtime code for increased stability.
532 (thanks to Thiemo Seufer)
533 * increased the maximimum compact environment size to allow
534 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
535 * partial workaround for bug 135: don't name the function we're
536 calling for hairy cases of EVAL, so as not to accumulate one environment
537 entry per call to EVAL. (reported by Kevin Reid)
538 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
539 * optimization: type testing for non-vector arrays should be faster.
540 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
542 * bug fix: setting 31st element of a bit vector to zero did not work
544 * bug fix: redefining a class definition which failed due to a
545 previous accessor / function clash now works (but see BUGS entry
546 #380 for more problems in this area). (thanks to Zach Beane)
547 * the long form of DEFINE-METHOD-COMBINATION disables method group
548 checking when given a single method group with pattern *. (thanks
550 * on x86 compiler supports stack allocation of results of simple
551 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
552 * support for building and running on Mac OS X 10.4 "Tiger" has been added
553 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
554 * fixed some bugs related to Unicode integration:
555 ** the restarts for recovering from input and output encoding
556 errors only appear when there is in fact such an error to
558 * fixed some bugs revealed by Paul Dietz' test suite:
559 ** MISC.549 and similar: late transformation of unsafe type
560 assertions into derived types caused unexpected code
562 ** SCALE-FLOAT type deriver is less wrong.
563 ** type derivers for EXP, LOG and similar functions compute result
564 types for complex arguments better.
565 ** (MISC.563) CONJUGATE type deriver works for very restricted
567 ** out-of-line type testers for character strings are available.
568 ** EQUAL compiler transform understands specialness of objects
570 ** accessing double-floats stored on the stack now works on x86-64.
571 ** debugger internals could sometimes create invalid lispobjs,
572 resulting in GC crashes.
573 ** MISC.548: type check weakening can convert required type into
575 ** initialization forms for bindings are not in scope of free special
578 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
579 * incompatible change: thread support for non-NPTL systems has
580 been removed - locking is buggy and unreliable. A threaded
581 SBCL build will now warn at startup and refuse to create
582 new threads, unless futex support is detected in the kernel
583 * incompatible change: the top level REPL now has only an ABORT
584 restart associated with it, not TOPLEVEL and ABORT as it used to.
585 TOP and TOPLEVEL are now available as debugger commands for
586 returning to the top level.
587 * incompatible change: forms evaluated in the REPL now use the
588 global optimization policy.
589 * incompatible change: user- and system-initialization files are
590 no longer processed with LOAD, but by READ and EVAL; hence the
591 global optimization policy, startup package, readtable, etc,
593 * The .fasl file format number has been incremented because of
594 various incompatible changes.
595 * internal entry point details and argument counts no longer appear
596 in backtraces unless explicitly requested by setting
597 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
598 * built-in and standard functions no longer have names like "top
599 level local call to FOO".
600 * fixed bug 32: functions defined in non-null lexical environments
601 now have more legible printed representation
602 * fixed bug 33: functions defined in non-null lexical environemnts
603 are now more amenable to inspection by INSPECT.
604 * workaround for bug 354: XEPs no longer appear in backtraces unless
605 explicitly requested.
606 * fixed bug: receiving the signal which results from attempting to
607 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
608 SIGBUS on other *BSDs) on architectures where the C stack is also
609 the Lisp stack (x86 and x86-64) from foreign code no longer leads
610 to debugger confusion or wild instability. (reported by Cheuksan
612 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
613 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
614 (reported by Lutz Euler)
615 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
616 compiled code. (reported by Paul Dietz)
617 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
618 specialized parameter to a method to something that is not TYPEP
619 the specializer is now possible.
620 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
621 face of package deletion.
622 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
623 pathname. (reported by David Tolpin on c.l.l)
624 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
625 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
626 STRING-STREAM and FILE-STREAM, respectively.
627 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
628 than previously; in addition, its test suite is now run on build.
629 (thanks to Luke Gorrie)
630 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
631 * optimization: added a immediate representation for single-floats
633 * fixed some bugs revealed by Paul Dietz' test suite:
634 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
635 ** attempting to create a package with a colliding nickname causes
636 correctable errors to be signalled.
637 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
638 ** several x86-64 backend bugs related to sign-extension of immediate
641 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
642 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
643 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
645 * fixed loading of multiply forward-referenced layouts.
646 (thanks to Cheuksan Wang)
647 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
648 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
649 (see COMPILE-FILE documentation for details of :PRINT option)
650 * fixed bugs 19 and 317: fixed-format floating point printing is
651 more accurate. This also fixes a bug reported by Adam Warner
652 related to the ~@F format directive.
653 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
655 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
656 dispatch table if the from-char is a dispatch macro character.
657 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
658 by the wordsize no longer ignore the last word. (reported by Lutz
660 * fixed bug in type checking of dynamic-extent variables. (reported
662 * optimization: sequence traversal functions use their freedom to
663 coerce function designators to functions.
664 * optimization: code with many calls to CLOS methods specialized on
665 CLOS classes has had redundant type checks removed.
666 * optimization: added declarations to speed up operations that access
667 the internal character database (for example STRING-UPCASE)
668 * optimization: comparison operations between floats and sufficiently small
669 fixnums no longer create extra rationals
670 * fixed some bugs related to Unicode integration:
671 ** portions of multibyte characters at the end of buffers for
672 character-based file input are correctly transferred to the
673 start of the buffer at the next read.
674 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
675 passing it through to OPEN.
676 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
677 argument given to internal calls to OPEN.
678 ** the built-in comment readers (introduced by character sequences
679 ";" and "#|") are more forgiving to encoding errors; they will
680 STYLE-WARN and then attempt to resync the stream at a character
681 boundary. (thanks to Teemu Kalvas)
682 * fixed some bugs in the x86-64 port:
683 ** Negative short int return values from c-calls are sign-extended
685 ** The stack is aligned to 16-bytes for c-calls, as required by
686 the ABI. (thanks to Cheuksan Wang)
687 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
688 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
689 for enums. (thanks to Vincent Arkesteijn)
690 ** Multiple small optimizations and bugfixes for floating point
692 * fixed some bugs revealed by Paul Dietz' test suite:
693 ** Space, Tab, Linefeed, Return and Page have the invalid
694 secondary constituent character trait.
695 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
697 ** WITH-INPUT-FROM-STRING should only update the index place on
699 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
701 ** Bit-array operations (BIT-AND and similar) worked incorrectly
702 with one-dimensional arrays with fill pointers.
703 ** TYPE-OF failed on a complex with an integer realpart and a
705 ** compiler failure during type inference for the code of form
706 (IF (EQL X (THE ...)) ...) (MISC.535).
708 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
709 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
710 platform. The port supports 61-bit fixnums, large memory spaces
711 and reloading shared object files.
712 * enhancement: saving cores with foreign code loaded is now
713 supported on ppc/Darwin in addition to the previously supported
715 * enhancement: the statistical profiler now walks deeper into the
716 call stack for more meaningful call-graphs and accrued time
717 reports (x86/x86-64 only). It also now reports time spent in
719 * enhancement: it is now possible to trace most individual methods
720 of a generic function in addition to tracing the generic function
722 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
724 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
725 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
726 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
727 produces an error. (thanks to Vincent Arkesteijn)
728 * bug fix: NAMESTRING on pathnames with :WILD components in their
729 directories works correctly. (thanks to Artem V. Andreev)
730 * fixed bug 125: compiler preserves identity of closures. (reported
732 * bug fixed: functions with &REST arguments sometimes failed with
733 "Undefined function" when compiled with (DEBUG 3). (reported by
735 * bug fix: overflow during compiling of setting element of a bit
736 vector with constant index and value. (reported by Timmy Douglas)
737 * build fix: fixed the dependence on *LOAD-PATHNAME* and
738 *COMPILE-FILE-PATHNAME* being absolute pathnames.
739 * on x86 compiler partially supports stack allocation of dynamic-extent
741 * GO and RETURN-FROM do not check the extent of their exit points
742 when compiled with SAFETY 0.
743 * fixed some bugs related to Unicode integration:
744 ** encoding and decoding errors are now much more robustly
745 handled; it should now be possible to recover even from invalid
746 input or output to the terminal. (thanks to Teemu Kalvas)
747 ** provided a first cut at implementing STRING-TO-OCTETS and
748 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
749 ** altered the SB-MD5 contributed module slightly, changing the
750 interface just enough for it to be supportable for builds where
751 lisp characters are not eight bits.
752 * fixed some bugs revealed by Paul Dietz' test suite:
753 ** the FORMATTER-generated functions for ~V[ conditionals require
754 the correct number of arguments.
755 ** READ-FROM-STRING returns the mandated second value when applied
756 to displaced strings.
757 ** the #\Rubout and #\Backspace characters are treated as invalid
758 constituent characters by the tokenizer.
760 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
761 * new feature: reloading changed shared object files with
762 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
763 * new feature: references to foreign variables and functions
764 can now be compiled and loaded before the corresponding shared
765 object file is loaded, as long as the foreign definitions are
766 available at runtime.
767 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
768 supported. (thanks to Dan Debertin)
769 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
770 just keywords. (thanks to Vincent Arkesteijn)
771 * fixed bug #331: structure-class instances corresponding to
772 DEFSTRUCT forms are now created eagerly.
773 * fixed bug #345: backtraces from calls to undefined functions work
774 on x86 as well. Related bug #61 is now also partially fixed on x86
775 (backtraces from throws to unknown catch tags.)
776 * bug fix: lambda-list parsing is now stricter vrt. order and number
777 of lambda-list keywords.
778 * bug fix: as specified by AMOP, an error is signalled if a
779 class-option appears multiple times in a DEFCLASS form. (reported
781 * bug fix: RANDOM can be compiled when the compiler derives the type
782 of its numeric argument as a disjoint set of small integers.
783 (reported by Paul Dietz)
784 * bug fix: starting a core saved with shared objects loaded when
785 those objects are not available no longer causes threaded SBCL to
786 hang. (reported by Sean Ross)
787 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
789 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
790 argument, which must always be NIL. (reported by Kalle Niemitalo)
791 * bug fix: printing 1.0d+23 no longer results in an error.
792 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
793 * bug fix: structure slot setters preserve evaluation order. (thanks
795 * fixed some bugs related to Unicode integration:
796 ** RUN-PROGRAM can allow its child to take input from a Lisp
797 stream. (reported by Stefan Scholl)
798 ** ASDF-INSTALL successfully downloads .tar.gz files in all
799 locales. (reported by Ken Causey)
800 * fixed some bugs revealed by Paul Dietz' test suite:
801 ** INCF, DECF and REMF evaluate their place form as specified in
803 ** FORMATTER expands ~{ ~} iteration directives with V or #
804 parameters correctly.
805 ** FORMATTER deals with the ~@[ ~] conditional directive where the
806 consequent uses no arguments correctly.
807 ** the system has a partial understanding of the (COMPLEX RATIO)
809 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
812 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
813 * new feature: a build-time option (controlled by the :SB-UNICODE
814 keyword feature, enabled by default) for building the system with
815 support for the entire 21-bit character space defined by the
817 * new feature: the system now has rudimentary external-format
818 support; the primary user-visible change at this time is that
819 characters with the high bit set (such as lower-case-e-acute) will
820 print correctly to a terminal in a UTF-8 environment.
821 * minor incompatible change: BASE-CHAR no longer names a class;
822 however, CHARACTER continues to do so, as required by ANSI.
823 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
824 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
825 has been moved to the SB-EXT package (temporarily re-exported from
827 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
828 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
830 * fixed bug #351: better error-handlind and reporting for malformed
832 * fixed bug #350: bignum-printing is now more memory-efficient,
833 allowing printing of very large bignums, eg. (expt 2 10000000).
834 (reported by Bruno Haible)
835 * fixed bug #302: better primitive-type selection for intersection
837 * fixed bug #308: non-graphic characters now all have names, as
838 required. (reported by Bruno Haible)
839 * bug fix: redefining a class with different superclasses now correctly
840 removes it from the direct-subclasses of its previous superclasses.
841 (reported by David Morse)
842 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
843 new value now works. (reported by Bruno Haible)
844 * bug fix: correct canonicalization of multiple non-standard slot
845 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
846 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
847 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
848 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
849 now exists, an signals an error.
850 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
851 during macroexpansion for non-iteration variables as well. (reported
852 by Bruno Haible for CMUCL)
853 * bug fix: Cyclic structures and unprintable objects in compiler
854 messages no longer cause errors. (reported by Bruno Haible)
855 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
856 and READ-FROM-STRING all now return a primary value of NIL if
857 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
858 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
859 (reported by Bruno Haible for CMUCL)
860 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
861 do not designate a condition. (reported by Bruno Haible for
863 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
864 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
865 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
866 permitting resolution in favour of any of the conflicting symbols.
867 (reported by Bruno Haible for CMUCL)
868 * bug fix: EQUAL compiler optimizations is less aggressive on
869 strings which can potentially compare true despite having distinct
870 specialized array element types.
871 * bug fix: unit enumerations can be defined without dividing by
872 zero. (thanks to Vincent Arkesteijn)
873 * FORMAT compile-time argument count checking has been enhanced.
874 (report from Bruno Haible for CMUCL)
875 * a partial workaround for the bug 262: the compiler does not try to
876 inline-expand a local function doing RETURN-FROM from a deleted
877 BLOCK. (thanks to Peter Denno for the bug report and to David
878 Wragg for the simple test case)
879 * fixed some bugs revealed by Paul Dietz' test suite:
880 ** RENAME-PACKAGE allows all package designators as new package
882 ** constraint propagation and lambda variable substitution are
883 more cautious in dealing with partially deleted code.
884 ** compiler sometimes lost reoptimization passes.
885 ** CERROR, given a condition as condition designator, passes its
886 remaining arguments to the continue format control without
888 ** Case-altering FORMAT directives work correctly on non-ASCII
890 ** The REAL type specifier handles bounds outside the floating
891 point ranges without signalling FLOATING-POINT-OVERFLOW.
892 ** Functions with IR1-transformations can create intercomponent
893 references to global functions.
894 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
896 ** FORMAT ~R treats a nil value for its first parameter correctly.
898 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
899 * enhancement: saving cores with foreign code loaded is now
900 supported on x86/NetBSD and sparc/Linux in addition to the previously
902 * bug fix: on some platforms repeated installations caused multiple
903 copies of HTML documentation to be installed -- should not happen
904 any more. (reported by Stefan Scholl)
905 * bug fix: parsing self-recursive alien record types multiple times
906 no longer causes infinite recursion. (reported by Thomas F. Burdick,
907 original patch by Helmut Eller for CMUCL)
908 * bug fix: stack-exhaustion detection works now on NetBSD as well.
909 (thanks to Richard Kreuter)
910 * bug fix: defining classes whose accessors are methods on existing
911 generic functions in other (locked) packages no longer signals
912 bogus package lock violations. (reported by François-René Rideau)
913 * bug fix: special variables as DEFMETHOD parameters no longer have
914 associated bogus type declarations. (reported by David Wragg and
916 * bug fix: read-write consistency on streams of element-type
917 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
918 * bug fix: redefiniton of the only method of a generic function with
919 no DEFGENERIC no longer emits a full WARNING. In addition,
920 redefinition of generic functions with no DEFGENERIC to an
921 incompatible lambda list now signals an error. (thanks to Zach
923 * bug fix: DEFGENERIC now works even when there's a function of the
924 same name in an enclosing lexical environment. (thanks to Zach
926 * fixed compiler failure, caused by instrumenting code during
927 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
928 * optimization: added loop analysis and improved register allocation
929 * fixed some bugs revealed by Paul Dietz' test suite:
930 ** POSITION on displaced vectors with non-zero displacement
931 returns the right answer.
932 ** (SIMPLE-STRING) is a valid type specifier for sequence
934 ** *PRINT-LEVEL* handling for slotless structures is pedantically
936 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
937 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
939 ** The justification version of the FORMAT ~< directive treats
940 non-zero minpad parameter correctly.
942 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
943 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
944 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
945 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
946 the supported interface.
947 * new feature: Single-stepping of code compiled with DEBUG 2 or
948 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
949 * new feature: saving cores with foreign code loaded is now
950 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
951 Timothy Moore's work for CMUCL)
952 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
953 parameters to * instead of NIL if no initform is supplied.
954 (reported by Johan Bockgaard)
955 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
956 correctly when FUNCALL appears as the car of the form. Note:
957 despite this FUNCALL forms are not currently subject to
958 compiler-macro expansion. (port of Raymond Toy's fix for the
959 same from CMUCL, reported by Johan Bockgaard)
960 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
961 (thanks for Teemu Kalvas)
962 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
963 case instead of terminating the iteration (thanks for Julian
964 Squires, Sean Champ and Raymond Toy)
965 * bug fix: incorrect expansion of defgeneric that caused a style
966 warning. (thanks for Zach Beane)
967 * bug fix: slot accessor effective method computation works properly
968 for classes with multiple non-standard applicable methods on
969 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
970 * on x86 compiler supports stack allocation of results of LIST and
971 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
972 CMUCL implementation by Gerd Moellmann)
973 * fixed some bugs revealed by Paul Dietz' test suite:
974 ** FORMAT strings with both the ~<~:;~> form of the justification
975 directive and pretty-printing directives cause an error.
977 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
978 * incompatible change: the internal functions
979 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
980 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
981 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
982 instead of the old functions.
983 * new feature: on platforms where "dladdr" is available foreign
984 function names now appear in backtraces. (based on Helmut Eller's
986 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
987 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
989 * documentation: ASDF manual is now distributed with SBCL.
990 * bug fix: SBCL can now load its contributed modules with REQUIRE
991 even if the system-provided entries have been removed from
992 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
994 * bug fix: backtraces involving undefined functions or assembly
995 routines are more informative. (thanks to Brian Downing)
996 * bug fix: mutually referent alien structures now work correctly.
997 (reported by Rick Taube)
998 * bug fix: structures defined by WITH-ALIEN can be referred to
999 within other definitions in the same WITH-ALIEN.
1000 * bug fix: division operators (MOD, TRUNCATE and the like) with
1001 constant zero divisors and integer dividends no longer generate
1003 * bug fix: provide default methods for INPUT-STREAM-P and
1004 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1005 * bug fix: improve the detection and resolution of MOP metacycles
1006 (where effective method computation depends on the generic function
1007 having its effective method computed). (reported by Bruno Haible)
1008 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1009 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1010 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1011 * bug fix: pathnames with relative directory components can be
1012 represented relative to default pathnames.
1013 * optimization: in taking the GCD of bignums, reduce the two bignums
1014 to approximately the same size (using Euclid's algorithm) before
1015 applying the more sophisticated binary GCD. (thanks to Juho
1017 * optimization: COUNT on bitvectors now operates word-at-a-time.
1018 * optimization: ASH with a positive, but not necessarily constant,
1019 (leftwards) shift, when in a modular context, is compiled to a
1021 * fixed some bugs revealed by Paul Dietz' test suite:
1022 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1023 if the corresponding argument is NIL.
1024 ** FORMAT directives accepting a minpad parameter treat negative
1026 ** literal commas as character parameters to FORMAT directives are
1028 ** literal spaces directly after ~<Newline> directives within a
1029 format-logical-block (~:< ~@:>) do not induce :FILL-style
1030 conditional newlines.
1031 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1032 as specified: it no longer includes conditional newlines).
1033 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1035 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1036 :PER-LINE-PREFIX argument does not evaluate to a string.
1038 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1039 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1040 the manual for details; for now, package locks can be disabled by
1041 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1042 if no major problems are found then it is likely that they will be
1043 compiled in unconditionally.
1044 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1045 unsupported operators on all platforms. To load a shared library
1046 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1047 object file, link it into a shared library outside of SBCL and
1048 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1050 * bug workaround: ROOM T and the GC were not working together
1051 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1052 reporting facility aren't preserved. That reporting has been
1053 disabled, so now until and unless someone figures out how to make it
1054 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1055 * minor incompatible change: as threatened around sbcl-0.8.0, the
1056 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1057 The symbols are also exported from SB-PCL for backwards
1058 compatibility, but more so than before SB-PCL should be treated as
1059 an implementation-internal package.
1060 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1062 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1063 It is known as of this release not to work on the Alpha, however.
1064 * fixed bug #167: errors signalled due to illegal syntax in method
1065 bodies are now more legible.
1066 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1067 designators and can hence be used with TYPEP.
1068 * fixed bug #333: CHECK-TYPE now ensures that the type error
1069 signalled, if any, has the right object to be accessed by
1070 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1071 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1072 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1074 * fixed bug #334: programmatic addition of slots using specialized
1075 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1076 and :ALLOCATION :CLASS slots.
1077 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1078 float exponents. (rereported by Peter Seibel)
1079 * fixed a bug: #\Space (and other whitespace characters) are no
1080 longer considered to be macro characters in standard syntax by
1081 GET-MACRO-CHARACTER.
1082 * fixed bug: initialization of condition class metaobjects no longer
1083 causes an instance of the condition to be created. (reported by
1085 * fixed bug: it is now possible to have slots such that
1086 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1087 description is neither :INSTANCE nor :CLASS.
1088 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1089 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1090 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1091 non-standard SLOT-DEFINITION-ALLOCATION values.
1092 * fixed bugs: various race conditions handling exiting threads.
1093 CL-PPCRE's thread torture test now passes.
1094 * fixed arguably-a-bug: GC time proportional to number of threads in
1095 system even when most of them are idle
1096 * optimization: improved performance of BIT and SBIT on bit-vectors.
1097 * .fasl file incompatibility: The fasl file version number has
1098 been incremented because of changes associated with package locks.
1100 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1101 * minor incompatible change: the system no longer provides
1102 optimization hints (conditions which are TYPEP
1103 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1104 mode; these hints are still emitted when higher SPEED optimization
1106 * new contrib module: a sampling profiler (profiling by statistical
1107 sampling, rather than by instrumenting functions) is available as
1108 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1109 * the behaviour of the standard function ED is now customizeable by
1110 third parties through a hook variable: see ED's documentation
1111 string for information on the protocol.
1112 * the compiler no longer emits efficiency notes for (FUNCALL X)
1113 when the type of X is uncertain under default optimization
1115 * fixed bug 276: mutating a binding of a specialized parameter to a
1116 method to something that is not TYPEP the specializer is now
1118 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1119 exhibit the required behaviour on the x86 platform. (thanks to
1120 Peter van Eynde, Eric Marsden and Bruno Haible)
1121 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1122 even for difficult arguments. (reported by Peter Graves)
1123 * fixed bug 141a: the backquote printer now descends quoted
1125 * fixed another bug in backquote printing: no more destructive
1126 modification of the form's list structure. (reported by Brian
1128 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1129 move between its address being taken and the call to
1130 interrupt_thread, fixing a crashing race condition.
1131 * the SB-POSIX contrib implementation has been adjusted so that it
1132 no longer exhibits ridiculously poor performance when constructing
1133 instances corresponding to C structs.
1135 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1136 * minor incompatible change: the sb-grovel contrib now treats C
1137 structures as alien (in the sense of SB-ALIEN) objects rather than
1138 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1139 has implications for memory management of client code
1140 (sb-grovel-returned objects must now be manually managed) and for
1141 type safety (alien objects now have full types).
1142 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1143 used to control emission of compiler diagnostics, rather than the
1144 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1145 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1146 quality should be considered deprecated.
1147 * install.sh now installs the user manual as well
1148 * (not quite a new documentable feature, but worth considering in
1149 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1150 beginnings of a semantically meaningful condition hierarchy is
1151 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1153 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1154 designator as the defaults argument.
1155 * fixed bug: Displaced arrays whose displaced-to array has become
1156 too small now cause ARRAY-DIMENSION to signal an error, providing
1157 for safer bounds-checking. (reported by Bruno Haible)
1158 * fixed bug: DEFCLASS slot definitions with identical :READER and
1159 :WRITER names now signal a reasonable error. (reported by Thomas
1161 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1163 * fixed bug: Invalid slot specification errors now print correctly.
1164 (thanks to Zach Beane)
1165 * fixed bug 320: Shared to local slot value transfers in class
1166 redefinitions now happen corrently from superclasses as
1167 well. (reported by Bruno Haible)
1168 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1170 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1171 improper lists correctly. (reported by Bruno Haible)
1172 * fixed bug 313: source-transform for <fun-name> was erroneously
1173 applied to a call of a value of a variable with name <fun-name>.
1174 (reported by Antonio Menezes Leitao)
1175 * fixed bug 307: The obsolete instance protocol ensures that
1176 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1177 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1178 remove dead unknown-values globs from the middle of the stack.
1179 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1180 easier to resume long computations after using *BREAK-ON-SIGNALS*
1181 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1182 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1183 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1184 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1185 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1186 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1188 * fixed a (fairly theoretical) bug in string printing: if
1189 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1190 does not have array-element-type equal to the most general string
1192 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1193 function name. (thanks to Nikodemus Siivola)
1194 * fixed bug: compile-time format string checker failed on
1195 non-closed ~{. (reported by Thomas F Burdick)
1196 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1197 don't warn on legal loop code involving more than one aggregate
1198 boolean. (thanks to Nikodemus Siivola)
1199 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1200 in keyword position to MAKE-PACKAGE was not being reported as
1202 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1203 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1205 * fixed bugs 280 and 312: the checking for multiple definitions in a
1206 file is less likely to become confused by uses of inline
1208 * fixed bug: the #S reader macro performs the keyword coercion
1209 specified for slot names. (reported by Kalle Niemitalo)
1210 * fixed bug: lambda lists may contain symbols whose names start with
1211 & but are not lambda-list-keywords; their occurrence triggers a
1213 * fixed bug 321: define-method-combination argument lambda lists do
1214 not cause invalid code to be generated when &optional or &aux
1215 variables have default values. (reported by Bruno Haible)
1216 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1217 have CLOS classes; this appears also to have fixed reported
1218 instability in stack exhaustion detection.
1219 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1220 MUFFLE-WARNING when no associated restart is present is now
1222 * optimization: rearranged the expansion of various defining macros
1223 so that each expands into only one top-level form in a
1224 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1226 * optimization: used a previously little-used slot in symbols to
1227 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1229 * fixed some bugs revealed by Paul Dietz' test suite:
1230 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1232 ** Fixed an optimization bug related to inheritance of initforms
1233 from local to shared slots.
1234 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1235 ** CAST optimizer forgot to flush argument derived type.
1236 ** print/read consistency on floats is now orders of magnitude
1237 more likely. (thanks also to Bruno Haible for a similar report
1239 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1240 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1241 for objects of type REAL. Make it so.
1242 ** SXHASH is specified (infelicitously) to respect similarity,
1243 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1244 it so. (thanks to Markus Ziegler)
1245 ** on the Alpha, the compiler succeeds in compiling functions
1246 returning a known number of arguments greater than 63.
1247 ** fixed handling of invalid NIL arguments in keyword position
1249 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1251 ** provided a readably-printable representation for RANDOM-STATE
1253 ** ensured that pathnames, where they have a namestring, always
1254 print using #P"..." syntax.
1256 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1257 * Support for the forthcoming 2.0 version of the NetBSD kernel
1258 running on x86 hardware has been added. (thanks to Perry
1259 E. Metzger most immediately, and others for their past work)
1260 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1261 current release version; SBCL's OpenBSD support had been broken
1262 since about the time of OpenBSD's switch to ELF binary format.)
1263 * [placeholder for DX summary]
1264 ** user code with &REST lists declared dynamic-extent, under high
1265 speed or space and low safety and debug optimization policy.
1266 * The manual has been converted to Texinfo format and the debugger
1267 chapter from the cmucl manual has been added.
1268 * A facility has been added to extract documentation strings from
1269 sbcl and store them as Texinfo-formatted snippets for inclusion in
1270 the manual (via Texinfo's @include directive)
1271 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1272 the test case to Dave Roberts)
1273 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1274 pointers. (reported by Sean Ross)
1275 * bug fix: PROFILE output is printed nicely even for large numerical
1276 values. (thanks to Zach Beane)
1277 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1278 greater than 32 handle EOF correctly.
1279 * bug fix: on X86 an immediate argument of the IMUL instruction is
1280 correctly printed in disassembly. (thanks to Lutz Euler)
1281 * bug fix: class slots in redefined classes preserve their old
1282 values. (thanks to Bruno Haible and Nikodemus Siivola)
1283 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1284 succeeds. (reported by Marco Baringer)
1285 * fixed some bugs revealed by Paul Dietz' test suite:
1286 ** READ-SEQUENCE now works on ECHO-STREAMs.
1287 ** RATIONALIZE works more according to its specification. (thanks
1290 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1291 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1292 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1293 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1294 should matter to you only if you rebind the printer control
1295 variables and then find you want different bindings in the
1296 debugger than in the ordinary execution of your program.)
1297 * The runtime build system has been tweaked to support building
1298 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1299 assembler and linker. (thanks to Nikodemus Siivola)
1300 * Unbound, undefined, undeclared variables now trigger full
1301 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1302 more often programmer error than deliberate exploitation of undefined
1304 * optimization: the hash algorithm for strings has changed to one
1305 that is less vulnerable to spurious collisions. (thanks to Juho
1307 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1308 needless bounds checking. (thanks to Juho Snellman)
1309 * optimization: implemented multiplication as a modular
1310 (UNSIGNED-BYTE 32) operation on the PPC backend.
1311 * fixed some bugs revealed by Paul Dietz' test suite:
1312 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1314 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1315 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1316 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1318 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1320 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1322 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1323 their output stream on EOF from read.
1324 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1325 have been read to end-of-file.
1326 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1328 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1329 description of determination of which consecutive characters
1331 ** Printing the "Space" character with escaping on now yields "#\\ ",
1332 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1333 ** Reading floating-point numbers with *READ-BASE* set to a number
1334 less than 10 works correctly.
1335 ** Reading floating-point numbers with *READ-BASE* set to a number
1336 more than 10 works correctly.
1337 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1338 the readtable currently in effect.
1340 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1341 * minor incompatible change: parsing of namestrings on a physical
1342 (Unix) host has changed; numbers after the final #\. in a
1343 namestring are no longer interpreted as a version field. This is
1344 intented to be largely invisible to the user, except that the
1345 meaning of the namestring "*.*.*" has changed: it now refers to a
1346 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1347 should usually be replaced by
1348 (make-pathname :name :wild :type :wild :version :wild)
1349 with the added benefit that this is more likely to be portable.
1350 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1351 option to OPEN now signals an error if the file being opened
1352 exists; this may have an impact on existing code.
1353 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1354 signals received, on the PowerPC platforms (both Linux and
1355 Darwin). (thanks to Pierre Mai for pointing out the location of
1357 * several fixes on OS X: The system now builds and runs cleanly on
1358 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1359 (thanks to Brian Mastenbrook)
1360 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1361 second-resolution: integer multiples of 1/3600 between -24 and 24.
1362 (thanks to Vincent Arkesteijn)
1363 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1364 type when called with 1 argument; PEEK-CHAR checked type of
1365 PEEK-TYPE only after having read first character from a
1366 stream. (reported by Peter Graves)
1367 * bug fix: the garbage collector now has much better locality
1368 behaviour, and in particular no longer treats all memory as being
1369 exhausted when many small objects point to each other in a deeply
1371 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1372 recognized as being TYPEP their class.
1373 * bug fix: the PUSHNEW documentation string has been corrected.
1374 (thanks to Vincent Arkesteijn)
1375 * bug fix: defaulting of the value for the last of an atypically
1376 large number of multiple values being bound was not being
1377 performed correctly on the Alpha or PPC platforms
1378 * optimization: implemented multiplication as a modular
1379 (UNSIGNED-BYTE 32) operation on the x86 backend.
1380 * optimization: SEARCH on simple-base-strings can now be open-coded.
1381 (see also contrib/compiler-extras.lisp for inspiration for
1382 teaching the compiler about the Boyer-Moore algorithm).
1383 * value, returned by MAX (and MIN) called with several EQUALP, but
1384 not EQL, arguments now does not depend on compiler settings.
1385 * fixed some bugs revealed by Paul Dietz' test suite:
1386 ** in stack analysis liveness information is propagated from
1387 non-local entry points.
1388 ** pathwise CAST removing failed when the CAST node did not start
1390 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1391 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1393 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1394 host is already defined.
1395 ** RENAME-FILE works on streams instead of signalling an internal
1397 ** PEEK-CHAR uses the current readtable when determining whether
1398 or not a character is whitespace.
1399 ** MERGE-PATHNAMES handles the case when the pathname does not
1400 specify a name while the default-pathname specifies a version.
1401 ** Pathnames now stand a better chance of respecting print/read
1403 ** Attempting to use standardized file system operators with a
1404 pathname with invalid :DIRECTORY components signals a
1406 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1408 ** DIRECTORY on logical pathnames is more correct.
1409 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1410 signal a TYPE-ERROR if their argument is not a stream.
1411 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1412 designator argument does not designate a stream.
1413 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1414 examining the synonym.
1415 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1417 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1418 element-type larger than ([UN]SIGNED-BYTE 32).
1420 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1421 * When built with the :SB-FUTEX feature, threaded builds now take
1422 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1423 for faster/more reliable mutex and condition variable support.
1424 * Incompatible change (but one you probably shouldn't have been using
1425 anyway): the interface and code for arbitrating between multiple
1426 threads in the same user session has been redesigned.
1427 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1428 error when called without an explicit environment argument.
1429 (thanks to Nikodemus Siivola)
1430 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1431 confused about their position. (thanks to Adam Warner and Gerd
1433 * bug fix: Pretty printing backquoted forms with unquotations in the
1434 argument list position of various code constructs such as LAMBDA
1435 now works correctly. (reported by Paul Dietz)
1436 * bug fix: Pretty printing unquotations no longer loses all
1437 stream position information.
1438 * optimization: performance of string output streams is now less
1439 poor for multiple small sequence writes.
1440 * optimization: performance of CSUBTYPEP in the presence of complex
1441 expressions involving CONS and NOT many times has been improved.
1442 (reported by Paul Dietz)
1443 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1445 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1446 streams for the REPL, for the debugger, and so on. (thanks to
1448 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1449 * fixed some bugs revealed by Paul Dietz' test suite:
1450 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1451 is ignored if it is too small, rather than propagating through
1452 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1453 ** extremely complex negations of CONS types were not being
1454 sufficiently canonicalized, leading to inconsistencies in
1456 ** VALUES tranformer lost derived type.
1458 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1459 * fixed a bootstrapping bug: the build process no longer assumes
1460 that the various BOOLE-related constants have the same value in
1461 host and target lisps. (noted by Paul Dietz' test suite on an
1462 SBCL binary built from CLISP)
1463 * The system can now be dynamically linked on the MIPS platform,
1464 which enables dynamic loading of foreign code from Lisp. (thanks
1465 to Ralf Baechle for discussions on the MIPS ABI)
1466 * The system now records debugging information for its own source
1467 files in a filesystem-position-independent manner, relative to
1468 the "SYS" logical host.
1469 * fixed a compiler bug: MV-LET convertion did not check references
1470 to the "max args" entry point. (reported by Brian Downing)
1471 * tweaked disassembly notes to be less confident about proclaiming
1472 some instruction as an LRA. (thanks to Brian Downing)
1473 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1474 now each have their own history, command character, and other
1475 characteristics. (thanks to David Lichteblau)
1476 * fixed some bugs revealed by Paul Dietz' test suite:
1477 ** compiler failure in compiling LOGAND expressions including a
1479 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1480 shift greater than 32.
1481 ** FLUSH-DEST did not mark blocks for type check regeneration.
1482 ** HANDLER-CASE failed to accept declarations in handler clauses
1483 in some circumstances.
1485 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1486 * New code in contrib/sb-introspect (still probably not entirely
1487 stable yet) provides some support for smart Lisp development
1488 environments like SLIME.
1489 * The conditions signalled for errors occurring when loading .fasl
1490 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1491 in a way which should help ASDF recover gracefully.
1492 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1493 stuff has been cleaned up. If you code contrib/ stuff, this might
1494 affect you, and you can look at contrib/README, contrib/STANDARDS,
1495 and/or the 0.8.4.27 diff to check.
1496 * In full calls the compiler now does not generate checks for declared
1497 argument types for all arguments.
1498 * various threading fixes
1499 ** and some experimental patches which didn't make it into
1500 the main tree for this release, but which are shipped in
1501 contrib/experimental-thread.patch as a possible fix for some
1502 failures (deadlock, spinning...) in GC-intensive multithreaded
1504 * fixed PPC build problem (source code incompatibility of different
1505 library versions): added offsetof() hackery which attempts to divine
1506 where glibc maintainers put uc_mcontext today
1507 * fixed bug 282: compiler does not trust type assertions while passing
1508 arguments to a full call.
1509 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1510 &OPTIONAL VALUES type parameter.
1511 * fix bug 214: algorithm for noting rejected templates is now more
1512 similar to that of template seletion. (also reported by rydis on
1514 * fixed bug 141b: printing backquoted information readably and prettily
1515 inserts a space where necessary.
1516 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1517 (CHARACTER 10) are now reported as errors, rather than propagated
1518 as unknown types. (reported by piso on #lisp)
1519 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1520 with values NIL and :ERROR. (thanks to Milan Zamazal)
1521 * fixed bug 191c: CLOS now does proper keyword argument checking as
1522 described in CLHS 7.6.5 and 7.6.5.1.
1523 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1524 longer raise an error; further, using a list as a for-as-arithmetic
1525 counter now raises a meaningful error.
1526 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1527 now understood by sequence creation functions such as MAKE-SEQUENCE
1529 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1530 a STRING-INPUT-STREAM.
1531 * compiler enhancement: SIGNUM is now better able to derive the type
1533 * type declarations inside WITH-SLOTS are checked. (reported by
1535 * fixed some bugs revealed by Paul Dietz' test suite:
1536 ** incorrect optimization of TRUNCATE for a positive first
1537 argument and negative second.
1538 ** compiler failure in let-convertion during flushing dead code.
1539 ** compiler failure while deriving type of TRUNCATE on an
1540 interval, containing 0.
1541 ** ASH of a negative bignum by a negative bignum count now returns
1543 ** intersection of CONS types now canonicalizes properly, fixing
1544 inconsistencies in SUBTYPEP.
1546 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1547 * incompatible change: The --disable-debugger command line
1548 option now clobbers the debugger at a more fundamental
1549 level, by redefining #'INVOKE-DEBUGGER instead of by
1550 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1551 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1552 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1553 enter the debugger REPL and then suffer recursive errors
1554 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1555 while under the new system, BREAK in --disable-debugger mode
1556 terminates the system just as an unhandled error would.
1557 * fixed compiler performance when processing loops with a step >1;
1558 * bug fix: DOCUMENTATION now retrieves generic function
1559 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1560 support has been systematized, and now supports the methods
1561 specified by ANSI, along with a default method and a method for
1562 slot documentation. (reported by Nathan Froyd)
1563 * bug fix: effective methods associated with a generic function are
1564 no longer cached over a change of that generic function's method
1565 combination. (reported by Andreas Fuchs)
1566 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1567 names a non-existent directory. (thanks to Andreas Fuchs)
1568 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1569 the PPC platform, lead to stack corruption; this has been fixed.
1570 (reported by Rainer Joswig)
1571 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1572 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1573 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1574 arithmetic in general, is now much more reliable on the PPC
1576 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1577 platform now returns the right answer.
1578 * optimization: restored some effective method precomputation in
1579 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1580 precomputation is now tunable.
1581 * optimization: compiler-internal data structure use has been
1582 reviewed, and changes have been made that should improve the
1583 performance of the compiler by about 20%.
1584 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1585 simple-base-strings and simple-bit-vectors is improved.
1586 * optimization: the optimization of 32-bit logical and arithmetic
1587 functions introduced in version 0.8.3 on the x86 has been
1588 implemented on the mips, ppc and sparc platforms; an
1589 implementation of the same facility, but for 64-bit arithmetic,
1590 has been added for the alpha.
1591 * microoptimization: the compiler is better able to make use of the
1592 x86 LEA instruction for multiplication by constants.
1593 * bug fix: in some situations compiler did not report usage of
1594 generic arithmetic in (SPEED 3) policy.
1595 * bug 145b fix: compiler used wrong type specifier while converting
1596 MEMBER-types to numeric.
1597 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1599 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1600 index argument now works properly on the Alpha platform.
1601 * bug fix: floating point exception treatment on the Alpha platform
1603 * bug fix: FILE-POSITION works much better on string input and
1604 output streams. (thanks to Nikodemus Siivola)
1605 * bug fix: many threading/garbage collection symptoms sorted.
1606 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1607 might be pseudo-atomic.
1608 * internal change: Stopping for GC is now done with signals not
1609 ptrace. GC is now done in whichever thread wanted it, instead of
1611 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1613 * bug fix: result form in DO is not contained in the implicit
1615 * incompatible change: ICR structure is changed; the value part of
1616 CONTINUATION is now called LVAR; corresponding functions are
1617 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1618 * added type deriver for ISQRT (thanks to Robert E. Brown).
1619 * bug fix: better support for loading from the command line when an
1620 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1622 * fixed some bugs revealed by Paul Dietz' test suite:
1623 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1624 ** ROUND and FROUND now give the right answer when given very
1625 small float arguments.
1626 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1628 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1629 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1630 ** LCM should return a non-negative integer.
1631 ** PARSE-INTEGER returned the index of a terminator instead of the
1632 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1633 ** PARSE-INTEGER returned an incorrect index being applied to a
1635 ** LCM with two arguments of 0 returns 0 rather than signalling
1637 ** unsigned addition of a 32-bit constant with the high bit set no
1638 longer causes an internal compiler error on the x86.
1639 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1640 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1641 with negative last argument.
1642 ** byte specifiers with zero size and position no longer cause
1643 an error during type derivation.
1644 ** bignum multiplication on the Alpha platform now returns the
1646 * porting: The system now builds on SuSE AMD64, although it still
1647 generates a 32-bit binary.
1648 * .fasl file incompatibility: The fasl file version number has
1649 been incremented (because of the changes to internal compiler
1650 data structures referred to above).
1652 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1653 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1654 more accurately, on the Darwin kernel running on PowerPC hardware.
1655 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1656 * Compiler code deletion notes now signal a condition of type
1657 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1658 an associated MUFFLE-WARNING restart.
1659 * The compiler now performs limited argument count validation of
1660 constant format strings in FORMAT, and where appropriate in ERROR,
1661 CERROR and WARN. (thanks to Gerd Moellmann)
1662 * New ASDF-INSTALL contrib can be used for automatic download and
1663 installation of third-party Lisp code from CCLAN or other sites
1665 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1666 which forces another thread to execute a function supplied by the
1668 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1669 now accept and act upon their :ELEMENT-TYPE keyword argument.
1670 (reported by Martin Atzmueller, Edi Weitz)
1671 * bug fix: FILE-POSITION now accepts position designators up to
1672 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1673 is the greater. (thanks to Patrik Nordebo)
1674 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1675 Kalle Olavi Niemitalo)
1676 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1677 printable. (reported by Eric Marsden)
1678 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1679 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1680 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1681 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1682 (thanks to Stig Sandoe)
1683 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1684 UNREAD-CHAR now correctly includes the unread character in the
1685 target sequence. (thanks to Gerd Moellmann)
1686 * bug fix in threaded builds: the system can now be suspended and
1687 resumed by shell job control with minimal disruption.
1688 * bug fixes in times and timezones >2038AD
1689 * better handling of "where is GNU make?" problem in build scripts
1690 (thanks to Nikodemus Siivola)
1691 * new optimization: inside a named function any reference to a
1692 function with the same name is considered to be a self-reference;
1693 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1694 optimization quality.
1695 * new optimization on x86: logical functions and + now have
1696 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1697 used when the result is truncated to 32 bits.
1698 * VALUES declaration is partially enabled.
1699 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1700 * fixed some bugs revealed by Paul Dietz' test suite:
1701 ** The system now obeys the constraint imposed by
1702 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1703 types form a lattice under type intersection.
1704 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1705 ** ASSOC now ignores NIL elements in an alist.
1706 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1707 and (1+ MOST-POSITIVE-FIXNUM) answers.
1708 ** The addition of a method with invalid qualifiers to a generic
1709 function does not cause an error to be signalled immediately;
1710 a warning is signalled, and the error is generated only on
1711 calling the generic function.
1712 * changed .fasl file version number, in part to add type codes for
1713 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1714 obscure ANSI requirements
1716 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1717 * fixed bug 148: failure to inline-expand a local function left
1718 garbage, confusing the compiler.
1719 * fixed bugs 3cd: structure slot readers perform type check if the
1720 slot can have an invalid value (i.e. it is either not initialized
1721 or can be written with a less specific slot writer).
1722 * bug fix: the compiler now traps array references to elements off
1723 the end of an array; previously, the bounds checking in some
1724 circumstances could go off-by-one.
1725 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1727 * type declarations for array element types now obey the description
1728 on the CLHS page "Declaration TYPE", as per discussions on
1729 sbcl-help around 2003-05-08. This means that a declaration
1730 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1731 declaration, all references to BAR will be asserted or assumed
1732 (with THE, so dependent on compiler policy) to involve objects of
1733 type FOO. Note that no such declaration is implied in
1734 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1735 * declared types of functions from the "Conditions"
1736 chapter. (reported by Paul Dietz)
1737 * bug fix: CERROR accepts a function as its first argument.
1738 * bug fix: NTH an NTHCDR accept a bignum as index
1739 arguments. (reported by Adam Warner)
1740 * optimization: character compare routines now optimize comparing
1741 against a constant character. (reported by Gilbert Baumann)
1742 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1743 argument now works properly on the MIPS platform.
1744 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1745 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1746 declared to have a functional type, does not cause an error
1748 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1749 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1750 * fixed bug: if last continuation of a deleted block has a
1751 destination, this destination should be deleted too. (reported by
1753 * fixed a bug in the bootstrap process: the host compiler's values
1754 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1755 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1756 test case from Patrik Nordebo)
1757 * improved the ability of the disassembler on the PPC platform to
1758 provide helpful disassembly notes.
1759 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1760 the class in more cases than previously.
1761 * bug fix: FILE-POSITION now understands :START and :END for
1762 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1763 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1764 8) data. (thanks to David Lichteblau)
1765 * bug fix: it is possible to add a method to a generic function
1766 without lambda list.
1767 * bug fix: reader failed to signal END-OF-FILE inside an
1768 object representation. (reported by Nikodemus Siivola)
1769 * fixed some bugs revealed by Paul Dietz' test suite:
1770 ** LAST and [N]BUTLAST should accept a bignum.
1771 ** condition slot accessors are methods.
1772 ** (VECTOR NIL) is a subtype of STRING.
1774 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1775 * minor incompatible change: some nonsensical specialized lambda
1776 lists (used in DEFMETHOD) which were previously ignored now signal
1778 * minor incompatible change: the system is now aware of the types of
1779 variables in the COMMON-LISP package, and will signal errors for
1780 most violations of these type constraints (where previously they
1781 were silently accepted).
1782 * minor incompatible change: COMPILE-FILE now uses the freedom
1783 afforded (ANSI 3.2.2.3) to use derived function types for
1784 functions defined in the same file. This also permits the system
1785 to warn on static type mismatches and function
1786 redefinition. (Currently it does not work with high DEBUG level.)
1787 * minor incompatible change: VALUES declaration is disabled.
1788 * When issuing notes, the compiler now signals a condition of type
1789 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1790 restart for use in user handlers. It is expected that the
1791 COMPILER-NOTE condition will eventually become a condition
1792 supertype to a hierarchy of note types, which will then be
1793 handleable in a similar fashion. However, at the moment, no such
1794 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1795 but it's an implementation detail, not a classification for the
1797 * Changes in type checking closed the following bugs:
1798 ** type checking of unused values (192b, 194d, 203);
1799 ** template selection based on unsafe type assertions (192c, 236);
1800 ** type checking in branches (194bc).
1801 * A short form of VALUES type specifier has ANSI meaning (it has
1802 increased the number of situations when SBCL cannot perform type
1804 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1805 keywords or constants is permissible.
1806 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1807 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1808 classes. (thanks to Antonio Martinez)
1809 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1810 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1811 argument to a pretty-print stream. Instead, it leaves any such
1812 operations to DESCRIBE-OBJECT methods.
1813 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1815 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1816 lambda lists. (thanks to David Lichteblau)
1817 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1818 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1819 (thanks to Gerd Moellmann)
1820 * fixed bug 246: increased compilation speed of long
1821 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1823 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1825 * DEFCONSTANT now throws a condition of type
1826 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1827 constant to a non-EQL value; CONTINUE and ABORT restarts
1828 respectively change and preserve the value.
1829 * fixed bug 63: The code walker, part of the implementation of CLOS,
1830 is now better at handling symbol macros.
1831 * bug fix: There is no longer an internal implementation type named
1832 CL:LENGTH. (reported by Raymond Toy)
1833 * bug fix: In macro-like defining macros/special operators the
1834 implicit block does not enclose the lambda list.
1835 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1836 enclosed in parentheses) are not suitable as type specifiers, and
1837 their use properly signals an error now.
1838 * bug fix: An argument count mismatch for a type specifier in code
1839 being compiled no longer causes an unhandled error at compile
1840 time, but signals a compile-time warning.
1841 * fixed simple vector readable printing
1842 * bug fix: DESCRIBE takes more care over whether the class
1843 precedence list slot of a class is bound before accessing it.
1844 (reported by Markus Krummenacker)
1845 * bug fix: FORMATTER can successfully compile pretty-printer format
1846 strings which use variants of the ~* directive inside.
1847 * bug fix: SEARCH now applies its TEST predicate to the elements of
1848 the arguments in the correct order. (thanks to Wolfhard Buss)
1849 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1850 in type inference. (thanks to Robert E. Brown)
1851 * bug fix: Reading in symbols with an explicit package name of ""
1852 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1853 * fixed some bugs revealed by Paul Dietz' test suite:
1854 ** NIL is now allowed as a structure slot name.
1855 ** Arbitrary numbers, not just REALs, are allowed in certain
1856 circumstances in LOOP for-as-arithmetic clauses.
1857 ** Multiple class redefinitions before slot access no longer
1858 causes a type error.
1859 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1860 association between the name and a class.
1861 ** Generic functions with non-standard method-combination and over
1862 five methods all of which return constants no longer return NIL
1863 after the first few invocations. (thanks to Gerd Moellmann)
1864 ** CALL-NEXT-METHOD with no arguments now passes the original
1865 values of the arguments, even in the presence of assignment.
1866 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1868 ** DISASSEMBLE works with closures and funcallable instances.
1869 ** ADD-METHOD now returns the generic function, not the new method.
1870 ** FIND-METHOD signals an error if the lengths of the specializers
1871 is incompatible with the generic function, even if the ERRORP
1873 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1874 which its argument is a member.
1875 ** DEFCLASS only redefines the class named by its class-name
1876 argument if that name is the proper name of the class;
1877 otherwise, it creates a new class.
1878 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1879 of the UNBOUND-SLOT condition to the name of the slot.
1880 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1881 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1882 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1883 SLOT-MAKUNBOUND in the specified fashion.
1885 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1886 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1887 cross-compilation host. As a consequence, we can now bootstrap our
1888 way up to SBCL starting with a bare gcc toolchain and human-readable
1889 source code (first the source to CLISP, then the source to SBCL).
1890 * A contributed module containing a partial implementation of the
1891 simple-streams interface has been included. (thanks to Rudi
1893 * A contributed module implementing the RFC1321 Message Digest
1894 Algorithm, known as MD5, has been included.
1895 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1896 no longer has any effect, as the code controlled by this feature
1897 has been deleted. (As far as we know, no-one has ever built using
1898 this feature, and its semantics were confused in any case).
1899 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1900 work on conditions (as required by the ANSI specification),
1901 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1902 expected behaviour on conditions. Users should note, however,
1903 that such behaviour is not required by the ANSI specification,
1904 and so use of this behaviour may render their code unportable.
1905 * fixed some bugs revealed by Paul Dietz' test suite:
1906 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1908 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1909 ** SXHASH on condition objects no longer returns NIL.
1910 ** :ALLOCATION :CLASS slots are better treated; their values are
1911 updated on class redefinition, and initforms inherited from
1912 superclasses are applied.
1913 ** REMOVE-METHOD returns its generic function argument even when
1914 no method was removed.
1915 ** SHARED-INITIALIZE now initializes the values of the requested
1916 slots, including those with :ALLOCATION :CLASS.
1917 ** ALLOCATE-INSTANCE now works on structure classes defined via
1918 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1920 ** SLOT-EXISTS-P now works on conditions, as well as structures
1922 ** MAKE-LOAD-FORM now has the required methods on
1923 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1924 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1925 keyword, and now implements the SLOT-NAMES argument.
1926 ** methods with &OPTIONAL arguments no longer allow too many
1927 arguments to be passed in the call without error.
1928 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1929 option is consistent with the required arguments of the generic
1930 function lambda list.
1931 * bug fix: REQUIRE accepts a string designator. (Thanks to
1933 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1934 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1935 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1936 * bug fix: The compiler now checks for duplicated variables in macro
1938 * bug fix: SETQ on globals returns the correct value.
1939 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1940 not a COMPILER-ERROR (followed by some other strange error on
1941 choosing the CONTINUE restart).
1942 * bug fix: make.sh and friends are now more consistent in the way that
1943 they look for GNU "make".
1945 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1946 * experimental native threads support (on x86 Linux >=2.4 only).
1947 This is not compiled in by default: you need to add :SB-THREAD to
1948 the target features. See the "Beyond ANSI" chapter of the manual
1950 * fix for longstanding nonANSIism: The old distinction between
1951 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1952 The return value from CL:FIND-CLASS is now a CLOS class, and
1953 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1954 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1955 * An interface to the MetaObject Protocol, as described in Kiczales,
1956 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1957 MIT Press, 1991, is available from the SB-MOP package.
1958 * incompatible change: the SB-PCL package should now be considered
1959 a private implementation detail, and no longer a semi-private MOP
1961 * minor incompatible change: due to rearrangement for threads, the
1962 control stack and binding stack are now allocated at arbitrary
1963 addresses instead of being hardcoded per-port. Users affected by
1964 this probably have to be doing advanced things with shared
1965 libraries, and will know who they are.
1966 * minor incompatible change: Previously, all --eval forms used were
1967 processed with READ before any of them were processed with EVAL.
1968 Now each --eval form is processed with both READ and EVAL before
1969 the next --eval form is processed. (Thus package operations like
1970 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1971 work as the user might reasonably expect.)
1972 * minor incompatible change: *STANDARD-INPUT* is now only an
1973 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1975 * minor incompatible change: Y-OR-N-P is now character-oriented,
1976 not line oriented. Also, YES-OR-NO-P now works without errors.
1977 (thanks to Antonio Martinez)
1978 * sb-aclrepl module improvements: an integrated inspector, added
1979 repl features, and a bug fix to :trace command.
1980 * Known functions, which cannot be open coded by the backend, are
1981 considered to be able to check types of their arguments. (fixing
1982 a bug report by Nathan J. Froyd)
1983 * fixed a bug in computing method discriminating functions: It is
1984 now possible to define methods specialized on classes which have
1985 forward-referenced superclasses. (thanks to Gerd Moellmann)
1986 * fixed evaluation order in optional entries (reported by Gilbert
1988 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1989 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1990 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1991 required-by-AMOP NAME argument, as well as CLASS and
1992 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1993 * fixed bug 20: DEFMETHOD can define methods using names that are
1994 not the proper names of classes to designate class specializers.
1995 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1996 Unix file descriptors, instead of blowing up. (thanks to Antonio
1998 * Garbage collection refactoring: user-visible change is that a
1999 call to the GC function during WITHOUT-GCING will not do garbage
2000 collection until the end of the WITHOUT-GCING. If you were doing
2001 this you were probably losing anyway.
2002 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2003 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2004 TYPEP the latter but not the former.
2005 * The compiler issues a full WARNING for calls to undefined functions
2006 with names from the CL package.
2007 * MAP-INTO for a vector destination is open coded. (reported by
2008 Brian Downing on c.l.l)
2009 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2010 documentation string.
2011 * fixed some bugs revealed by Paul Dietz' test suite:
2012 ** COPY-ALIST now signals an error if its argument is a dotted
2014 ** Condition slots are now accessed more correctly in the presence
2015 of multiple initargs for a given slot.
2016 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2017 exclude restarts of the same name associated with a different
2019 ** DEFCLASS of forward-referenced classes with another
2020 forward-referenced class in the superclasses list no longer
2022 ** Condition slots are now initialized once each, not multiple
2023 times. (thanks to Gerd Moellmann)
2024 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2026 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2027 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2028 arguments contain duplicated elements.
2029 ** RESTART-CASE understands local macros.
2030 ** RESTART-CASE associates exactly its own restarts with a condition.
2031 ** ENDP in safe mode checks its argument to be of type LIST.
2032 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2033 in question is unbound.
2034 ** Optimized MAKE-INSTANCE functions no longer cause internal
2035 assertion failures in the presence of duplicate initargs.
2036 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2037 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2038 propagated even in the ctor optimized implementation of
2040 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2042 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2043 * a better implementation of SXHASH on (simple) bit vectors,
2044 measured both in execution speed and in distribution of results
2045 over the positive fixnums, has been installed. Likewise, a better
2046 implementation of EQUAL for simple bit vectors is now available.
2047 * fixed CEILING optimization for a divisor of form 2^k.
2048 * fixed bug 240 (emitting extra style warnings "using the lexical
2049 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2050 by Antonio Martinez)
2051 * fixed SXHASH, giving different results for NIL depending on type
2052 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2053 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2054 variable. (found by Rolf Wester)
2055 * SBCL does not ignore type declarations for special
2056 variables. (reported by rif on c.l.l 2003-03-05)
2057 * some bug fixes in contrib/sb-aclrepl/
2058 * fixed some bugs revealed by Paul Dietz' test suite:
2059 ** a bug in the CONS type specifier, whereby the CAR and CDR
2060 types got intertwined, has been fixed;
2061 ** the type system is now able to reason about the interaction
2062 between INTEGER and RATIO types more completely;
2063 ** APPEND, [N]REVERSE and NRECONC check that those their
2064 arguments, which must be proper lists, are really so;
2065 ** An array specialized to be unable to hold elements has been
2066 implemented, as required -- yes, really -- by ANSI;
2067 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2068 on malformed property lists;
2070 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2071 * incompatible packaging change: in line with Unix convention,
2072 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2073 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2074 * REQUIRE and PROVIDE are now optionally capable of doing something
2075 useful. See the documentation string for REQUIRE.
2076 * infrastructure for a managed SBCL contrib system: contributed
2077 modules in this release include:
2078 ** the ASDF system definition facility;
2079 ** an interface to the BSD Sockets API;
2080 ** an ACL-like convenience interface to the repl;
2081 (thanks to Kevin Rosenberg)
2082 ** an implementation of ROTATE-BYTE, with efficient implementation
2084 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2085 gives a closer approximation to the right answer than previously.
2086 (thanks to Raymond Toy)
2087 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2088 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2089 optional environment arguments, as required by ANSI.
2090 * fixed bugs in other functions taking environment objects, allowing
2091 calls with an explicit NIL environment argument to be compiled
2093 * fixed bug 228: primary return values from
2094 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2095 COMPILE or FUNCTION.
2096 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2097 :TYPE will no longer signal errors on innocuous objects.
2098 * fixed bug 231b: SETQ is better at respecting type declarations in
2099 the lexical environment.
2100 * fixed a bug in DEFCLASS: classes named by symbols with no or
2101 unprintable packages can now be defined.
2102 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2103 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2104 * fixed compiler failure related to checking types of functions.
2105 (reported by Robert E. Brown)
2106 * the compiler is now much more consistent in its error-checking
2107 treatment of bounding index arguments to sequence functions: in
2108 (SAFETY 3) code, errors will be signalled in almost all cases if
2109 invalid sequence bounding indices are passed to functions defined
2110 by ANSI to operate on sequences.
2111 * fixed a bug in the build procedure: documentation of SBCL-specific
2112 packages is now preserved and available in the final Lisp image.
2113 * lifted FDEFINITION lookup out of loops in the implementation of
2114 many list operations. (thanks to Robert E. Brown)
2115 * fixed a bug in the reader: the #n# reader macro now works for
2116 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2117 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2118 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2119 E. Brown and Tony Martinez respectively)
2120 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2121 SLOT-BOUNDP: in particular, it is now able to optimize them much
2122 better, and is now not vulnerable to having packages renamed.
2123 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2124 always called when a slot is not present in an instance. (thanks
2126 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2127 longer signal an error on unreading a character following EOF on
2128 the previous constituent stream. (thanks to Tony Martinez)
2129 * fixed some bugs revealed by Paul Dietz' test suite:
2130 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2131 not just nonnegative fixnums;
2132 ** the logical bit-array operators such as BIT-AND now accept an
2133 explicit NIL for their "opt-arg" argument (to indicate a
2134 freshly-consed result bit-array);
2135 ** ELT now signals an error on an invalid sequence index in safe
2137 ** the type system is now cleverer about negations of numeric
2138 types, and consequently understands the BIGNUM and RATIO types
2140 ** the type system is now cleverer about the interaction between
2141 INTEGER and RATIO types: while bugs still remain, many more
2142 cases are accurately computed;
2143 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2144 if it is in the last clause;
2145 ** CONSTANTLY now correctly returns a side-effect-free function in
2147 ** DECLARE is no longer treated as a special-operator; in
2148 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2149 * incremented fasl file version number due to the change in the
2150 DEFSTRUCT-SLOT-DESCRIPTION structure.
2152 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2153 * minor incompatible change: code processed by the "interpreter" or
2154 EVAL now has a compilation optimization policy of (DEBUG 2)
2155 (changed from (DEBUG 1)) to improve debuggability of interactive
2156 development, and to allow the use of the debug RETURN command in
2158 * an experimental implementation of the RETURN command for the
2159 debugger has been included. (thanks to Frederik Kuivinen)
2160 * fixed bug 62: constraints were not propagated into a loop.
2161 * fixed bug in embedded calls of SORT (reported and investigated by
2163 * fixed some bugs revealed by Paul Dietz' test suite:
2164 ** printing and reading of arrays with some dimensions having
2165 length 0 (thanks to Gerd Moellmann);
2166 ** BOA constructor with &AUX argument without a default value does
2167 not cause a type error;
2168 ** CONSTANTP now returns true for all self-evaluating objects.
2170 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2171 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2172 accessors that are related by inheritance, as specified in the
2173 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2174 Valtteri Vuorikoski)
2175 * The compiler is now able to inline functions that were defined in
2176 a complex lexical environment (e.g. inside a MACROLET).
2177 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2179 * Support for the upcoming FreeBSD-5.0 release has been included.
2180 (thanks to Dag-Erling Smorgrav)
2181 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2182 effect when it is not in a toplevel context.
2183 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2184 stand a better chance of being correct. (thanks to Gerd
2186 * fixed bug in COERCE, which now signals an error on coercing a
2187 rational to a bounded real type which excludes the expected
2189 * The compiler is now able to derive types more accurately from the
2190 COERCE and COMPILE functions.
2191 * fixed bug 223: functional binding is considered to be constant
2192 only for symbols in the CL package.
2193 * fixed bug 231: SETQ did not check the type of a variable being set
2194 (reported by Robert E. Brown)
2195 * A new optimization for MAKE-INSTANCE has been included, fixing
2196 various bugs (including relating to :ALLOCATION :CLASS slots and
2197 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2199 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2200 ** As required by ANSI, LOOP now disallows anonymous collection
2201 clauses such as COLLECT I in conjunction with aggregate boolean
2202 clauses such as THEREIS (= I 1);
2203 ** LOOP now signals an error when any variable is reused in the
2204 same loop (including the potentially useful construct analogous
2205 to WITH A = 1 WITH A = (1+ A);
2206 ** IT is only a special loop symbol within the first clause of a
2207 conditional loop clause;
2208 ** LOOP with a typed iteration variable over a hashtable now
2209 signals a type error iff it should.
2210 * fixed some other bugs revealed by Paul Dietz' test suite:
2211 ** FILE-STREAM now names the class previously known as FD-STREAM;
2212 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2213 argument) no longer signals an error;
2214 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2215 of the slot symbol, rather than using the current package
2216 ((:CONC-NAME "") continues to intern the slot's name in the
2218 * incremented fasl file version number, because of the incompatible
2219 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2220 of the new implementation of DEFINE-COMPILER-MACRO.
2222 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2223 * Support for building SBCL for MIPS platforms running in
2224 little-endian mode has now been checked in, and basic
2225 functionality on said platforms verified.
2226 * minor incompatible change: PCL now records the pathname of a file
2227 in which methods and the like are defined, rather than its
2229 * minor incompatible change: TRUENAME now considers the truename of
2230 a file naming a directory to be the pathname with :DIRECTORY
2231 component indicating that directory.
2232 * minor incompatible change: a NAMED clause in the extended form of
2233 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2234 reason for the previous behaviour is unclear.
2235 * more systematization and improvement of CLOS and MOP conformance
2236 in PCL (thanks to Gerd Moellman and Pierre Mai):
2237 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2239 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2240 primary methods with no specializers;
2241 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2243 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2244 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2245 FORWARD-REFERENCED-CLASSes; error reporting on
2246 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2248 ** SXHASH on CLOS instances now uses a slot internal to the
2249 instance to return different numbers on distinct instances,
2250 while preserving the same return value through invocations of
2252 ** DEFMETHOD signals errors when methods with longer incongruent
2253 lambda lists are added to generic functions;
2254 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2255 CLASS, as specified in AMOP;
2256 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2257 on the order returned by the primary method for classes of
2258 class STANDARD-CLASS;
2259 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2260 * fixed some bugs shown by Paul Dietz' test suite:
2261 ** DOLIST puts its body in TAGBODY;
2262 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2264 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2265 value producing form;
2266 ** if more variables are given to PROGV than values, extra
2267 variables are bound and made to have no value;
2268 ** NSUBSTITUTE on list arguments gets the right answer with
2270 ** ELT signals an error of type TYPE-ERROR when the index argument
2271 is not a valid sequence index;
2272 ** LOOP signals (at macroexpansion time) an error of type
2273 PROGRAM-ERROR when duplicate variable names are found;
2274 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2275 ** FUNCALL of special-operators now cause an error of type
2277 ** PSETQ now works as required in the presence of side-effecting
2278 symbol-macro places;
2279 ** NCONC accepts any object as its last argument;
2280 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2282 ** loop-for-as-package does not require a package to be explicitely
2284 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2286 * fixed bug 166: compiler preserves "there is a way to go"
2287 invariant when deleting code.
2288 * fixed bug 172: macro lambda lists with required arguments after
2289 &REST arguments now cause an error to be signalled. (thanks to
2291 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2292 bug. (thanks to Matthew Danish)
2293 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2295 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2296 arguments are lexically rebound. (thanks to Gerd Moellmann and
2298 * fixed bug 194: error messages are now more informative when there
2299 is no primary method applicable in a call to a generic
2300 function. (thanks to Gerd Moellmann)
2301 * fixed bug in command line argument checking (thanks to Julian
2303 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2304 (thanks to Matthew Danish)
2305 * incremented fasl file version number, because of the
2306 SXHASH-related changes in the layout of CLOS data structures
2308 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2309 * minor incompatible change: The runtime (the Unix executable named
2310 "sbcl") is now much pickier about the .core files it will load.
2311 Essentially it now requires .core files to descend from the same
2312 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2313 as the runtime does. (The intent is to prevent the crashes which
2314 can occur, and which can even be reported as mysterious failures,
2315 when people patch the sources or change the build parameters
2316 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2317 sbcl and .core files.)
2318 * fixed bug: VALUES-LIST is no longer optimized away.
2319 * fixed bug 142: The FFI conversion of C string values to Lisp
2320 string values no longer conses excessively. (thanks to Nathan
2321 Froyd porting Raymond Toy's fix to CMU CL)
2322 * began to systematize and improve MOP conformance in PCL (thanks to
2323 Nathan Froyd, Gerd Moellman and Pierre Mai):
2324 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2326 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2327 ** FINALIZE-INHERITANCE is now called on class finalization;
2328 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2329 argument precedence order.
2330 * fixed bug 202: The compiler no longer fails on functions whose
2331 derived types contradict their declared type.
2332 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2333 so it can be non-toplevel.
2334 * The fasl file version number has changed (because of the new
2335 implementation of DEFMACRO).
2336 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2337 safe code, that any length requirement by their type-specifier
2338 argument is valid. The exceptions to this are described in bug
2340 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2341 functions now check the types of their inputs as required by ANSI.
2342 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2343 introduced symbol is DECLAREd to be SPECIAL.
2344 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2345 * fixed bug: :COUNT argument to sequence functions may be negative
2346 * fixed bug: body of DO-SYMBOLS may contain declarations
2347 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2348 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2349 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2350 symbol macro only once
2351 * fixed printing of call frame when argument list is unavailable
2352 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2353 * compiler no longer signals WARNING on unknown keyword
2356 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2357 * A beta-quality port to the mips architecture running Linux,
2358 based on the old CMUCL backend, has been made. It has been tested
2359 on a big-endian kernel, and works sufficiently well to be able to
2360 rebuild itself; it has not been tested in little-endian mode.
2361 * fixed an inconsistency between gencgc.c and purify.c which made
2362 dumping/loading .core files unreliable
2363 * fixed bug 120a: The compiler now deals correctly with IFs where
2364 the consequent is the same as the alternative, instead of
2365 misderiving the return type. (thanks to Alexey Dejneka)
2366 * fixed bug 113: Logical pathnames are now dumpable (the logical
2367 host is resolved at load-time, throwing an error if it is not
2369 * fixed bug 174: FORMAT's error message is slightly clearer when a
2370 non-printing character is used in a format directive.
2371 * fixed several bugs in compiler checking of type declarations, i.e.
2372 violations of the Python "declarations are assertions" principle
2373 (thanks to Alexey Dejneka)
2374 * fixed several bugs in PCL's error checking (thanks to Gerd
2376 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2378 * fixed bug in compilation of functions as first class values
2379 (thanks to Antonio Martinez-Shotton)
2380 * The compiler's handling TYPE-ERRORs which it can prove will
2381 inevitably happen at runtime has been cleaned up and corrected
2382 in several ways. (thanks to Alexey Dejneka)
2383 * improved argument type checking for various basic arithmetic
2384 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2385 lavished on them in the past that they can be compiled in many
2386 ways in different special cases
2387 * fixed bug 181: compiler checks validity of user supplied type
2389 * cleaned up code flushing in optimization: Function calls which
2390 should signal errors for safety purposes (e.g. which ANSI says
2391 should signal errors when their arguments are of incorrect type)
2392 are no longer optimized away.
2393 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2394 * incremented fasl file version number, because changes in the
2395 implementation of sequence functions like COERCE caused
2396 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2397 in old inline expansions) to become undefined. (Actually these
2398 changes were later undone, so we might very well be binary
2399 compatible with 0.7.7 after all, but leaving the version number
2400 incremented seemed like the simplest and most conservative
2403 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2404 * An alpha-quality port to the parisc architecture running Linux,
2405 based on the old CMUCL backend, has been made. This, even more so
2406 than the other backends, should be considered still a work in
2407 progress; known problems include that the Linux kernel in 64-bit
2408 mode does not propagate the correct sigcontext structure to
2409 userspace, and consequently SBCL on a parisc64 kernel will not
2411 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2412 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2413 suppressed.) Also now that the compiler is looking at declarations
2414 in the environment, it checks optimization declarations as well,
2415 and suppresses inlining when (> DEBUG SPEED).
2416 * More fixes have been made to treatment of floating point exception
2417 treatment and other Unix signals. In particular, floating point
2418 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2419 * The detection and handling of control stack exhaustion (infinite
2420 or very deeply nested recursion) has changed. Stack exhaustion
2421 detection is now done by write-protecting pages at the OS level
2422 and applies at all optimization settings; when found, a
2423 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2424 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2425 be caught using IGNORE-ERRORS.
2426 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2427 correctly when transforming recursive LABELS functions to LETs.
2428 (thanks to Alexey Dejneka)
2429 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2430 that are names of constants or global variables.
2431 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2432 alien routines with docstrings.
2433 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2434 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2436 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2437 (thanks to Eric Marsden)
2438 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2439 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2440 * Bug fix: The SPARC backend can now compile functions involving
2441 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2442 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2443 to a functions expecting a PATHNAME-DESIGNATOR.
2444 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2445 lambda lists. (thanks to Alexey Dejneka)
2446 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2447 functions so that EXPORTing the name of the class doesn't cause
2448 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2449 to Antonio Martinez for reporting this)
2450 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2451 constant-folded without failing an assertion. (thanks to Einar
2452 Floystad Dorum for reporting this)
2453 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2454 types are now dealt with more correctly by the compiler.
2455 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2456 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2457 behaviour analogous to RENAME-FILE. This puts its behaviour more
2458 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2460 * The fasl file version number has changed again. (because of the
2461 bug fix involving the names of PCL MAKE-INSTANCE functions)
2463 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2464 * bug fix: Floating point exceptions are treated much more
2465 consistently on the x86/Linux and PPC/Linux platforms.
2466 * Array initialization with :INITIAL-ELEMENT is now much faster for
2467 cases when the compiler cannot open code the array creation, but
2468 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2469 array accesses have also seen a speed increase.
2470 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2471 specified. (This was at the root of some bad interactions between
2472 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2474 * bug fix: Internal error arguments for undefined functions are now
2475 computed correctly on the PPC/Linux platform.
2476 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2477 Raymond Toy's patch for CMU CL)
2478 * Support for the Solaris 9 operating environment has been included
2479 (thanks to Daniel Merritt)
2480 * A very ugly but hopefully complete draft of the missing FFI chapter
2481 of the manual has been created by reformatting the corresponding
2482 CMU CL manual chapter into (currently very ugly and incoherent)
2483 DocBook and bringing it up to date for SBCL behavior. Thus, the
2484 manual is now essentially complete, at least by my extreme
2485 once-and-only-once standards, whereby it's acceptable to refer to
2486 the doc strings of SB-EXT functions as the primary documentation.
2487 * The fasl file version number has changed again, due to cleanup of
2488 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2489 is no longer a static symbol.)
2491 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2492 * SBCL now builds with OpenMCL (version 0.12) as the
2493 cross-compilation host; also, more progress has been made toward
2494 bootstrapping under CLISP.
2495 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2497 * bug 158 fixed: The compiler can now deal with integer loop
2498 increments different from 1; fixing this turned out also to fix
2500 * bug 169 fixed: no more bogus warnings about using lexical bindings
2501 despite the presence of perfectly good SPECIAL declarations (thanks
2502 to David Lichteblau)
2503 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2504 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2505 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2507 * bug 180 fixed: Method combination specifications no longer ignore
2508 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2509 * bug fix: Structure type predicate functions now check their argument
2510 count as they should.
2511 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2512 correctly. (thanks to Pierre Mai)
2513 * minor incompatible change: The --noprogrammer option is deprecated
2514 in favor of the new --disable-debugger option, which is very similar.
2515 (The major difference is that it takes effect at a slightly different
2516 time at startup, causing handling of errors in --sysinit and
2517 --userinit files will be affected differently.) The
2518 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2519 been added to allow this functionality to be controlled from ordinary
2520 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2521 maintainers, who might want to run non-interactive scripts to
2522 build SBCL cores which will later be used interactively.)
2523 * minor incompatible change: The LOAD function no longer, when given
2524 a wild pathname to load, loads all files matching that pathname.
2525 Instead, an error of type FILE-ERROR is signalled.
2527 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2528 * bug 147 fixed: The compiler preserves its block link/count
2529 invariants more correctly now so that it doesn't crash. (thanks
2531 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2533 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2534 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2535 no-value-for-this-character as NIL (as specified by ANSI).
2536 * HOST-NAMESTRING on physical pathnames now returns a string that is
2537 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2538 (thanks to Christophe Rhodes)
2539 * The Alpha port handles icache flushing more correctly. (thanks to
2541 * More progress has been made toward bootstrapping under CLISP. (thanks
2542 to Christophe Rhodes)
2543 * The fasl file format has changed again, because dynamic loading
2544 on OpenBSD (which has non-ELF object files) motivated some cleanups
2545 in the way that foreign symbols are transformed and passed around.
2546 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2547 is no longer treated as whitespace by the reader, but instead as
2548 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2549 |A
\7fB|, instead of A as it used to.
2551 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2552 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2553 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2554 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2555 this as of 0.7.1.45, but was left out of the previous news section
2556 (thanks to Dan Barlow)
2557 * SBCL now runs on the Solaris operating system on SPARC architectures
2558 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2559 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2560 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2561 * SPARC backend cleanups, allowing builds of cores optimized for V8
2562 and V9 SPARCS, and also emission of code targeted to a particular
2563 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2565 * SBCL is closer to bootstrapping under CLISP, thanks to various
2566 fixes by Christophe Rhodes.
2567 * The fasl file format has changed again, to allow the compiler's
2568 INFO database to support symbol macros.
2569 * The user manual (in doc/) is formatted into HTML more nicely.
2570 (thanks to coreythomas)
2571 * The system is smarter about SUBTYPEP relationships, especially
2572 those involving NOT types (including types such as ATOM which are
2573 represented internally using NOT types). Thus SUBTYPEP is less
2574 likely to return (VALUES NIL NIL) in general, and in particular
2575 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2577 * The fasl file format has changed again, because the internal
2578 representation of types now includes a new slot to support the new
2579 SUBTYPEP-of-NOT-types logic.
2580 * (not a change in the main branch of SBCL, but a related prototype
2581 which can hopefully be merged into the main branch of SBCL in the
2582 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2583 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2584 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2585 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2586 dispatch macro character combinations. (thanks to Alexey Dejneka)
2587 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2588 that superficially look like logical namestrings correctly.
2589 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2591 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2592 default supplied-p) for &optional and &key arguments. (thanks to
2595 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2596 * incompatible change: The compiler is now less aggressive about
2597 tail call optimization, doing it only when (> SPACE DEBUG) or
2598 (> SPEED DEBUG). (This is an incompatible change because there are
2599 programs which relied on the old CMU-CL-style behavior to optimize
2600 away their unbounded recursion which will now die of stack overflow.)
2601 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2602 for non-GENCGC systems has been increased to 20M (since that
2603 seems much closer to the likely performance optimum for modern
2604 systems than the old 4M value was)
2605 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2606 and *DEBUG-PRINT-LEVEL*
2607 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2608 of CMU CL's support for SPARC, and various endianness and other
2609 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2610 * new syntactic sugar for the Unix command line: --load foo.bar is now
2611 an alternate notation for --eval '(load "foo.bar")'.
2613 ** The system now detects stack overflow and handles it gracefully,
2614 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2615 optimization settings. (This is a good thing in general, and
2616 its introduction in this version should be particularly timely
2617 for anyone whose code fails because of suppression of tail
2619 ** The system now hunts for the C variable "environ" in a more
2620 devious way, to avoid segfaults when the C library version differs
2621 between compile time and run time. (thanks to Christophe Rhodes)
2622 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2623 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2624 ** The compiler no longer issues bogus style warnings for undefined
2625 classes in the same source file as the DEFCLASSes which defined
2626 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2628 ** fixes in CONDITION class precedence list for undefined function
2629 errors (thanks to Alexei Dejneka)
2630 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2631 correctly. (thanks to Dan Barlow)
2632 ** portability fixes aiming at bootstrapping under CLISP (thanks
2633 to Dave McDonald and Christophe Rhodes)
2634 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2635 ** fixes in type translation and and type inference (thanks to
2637 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2638 ** various fixes in the new ports (thanks to Dan Barlow)
2639 * several changes related to debugging:
2640 ** suppression of tail recursion, as noted above
2641 ** stack overflow detection, as noted above
2642 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2643 is now the default. (For some time encapsulation has been more
2644 reliable than the breakpoint-based :ENCAPSULATE NIL
2645 implementation, at least on X86 systems; and I just noticed that
2646 encapsulation also seems closer to the spirit of the ANSI
2649 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2651 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2652 up properly again. (There was a packaging bug in 0.7.0 which
2653 left their definitions in SB-SYS::LOAD-FOREIGN and
2654 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2655 vital for most things which interface to C-level interfaces,
2656 like extensions working with sockets or databases or
2657 Perl-compatible regexes or whatever, and the need to fix
2658 this bug is the main reason that 0.7.1 was released so
2660 ** DEFGENERIC is now choosier about the methods it redefines, so that
2661 reLOADing a previously-LOADed file containing DEFGENERICs does
2662 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2663 cycle now works as it should. (thanks to Alexey Dejneka)
2664 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2665 fixed by Christophe Rhodes. (He actually submitted this patch
2666 months ago, and I delayed until after 0.7.0.)
2667 ** Bug 111 (internal compiler confusion about runtime checks on
2668 FUNCTION types) was fixed by Alexey Dejneka.
2669 * Some internal cleanups (getting rid of variables which aren't
2670 needed now that the byte interpreter is gone) caused the fasl
2671 file format number to change again.
2673 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2674 * major incompatible change: The default fasl file extension, i.e. the
2675 default extension for files produced by COMPILE-FILE, has changed
2676 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2678 ** There are many changes in the implementation of the compiler.
2679 SBCL is now essentially a compiler-only implementation of ANSI
2680 Common Lisp. EVAL still "interprets" a few special cases, but
2681 almost all the interesting cases are handled by creating
2682 a LAMBDA expression, calling COMPILE on it, then calling
2683 FUNCALL on the result.
2684 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2685 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2686 Since the code is newer, there might still be some new bugs
2687 (though not as many as before Martin Atzmueller's fixes:-). But
2688 the new code is substantially simpler and clearer, and hopefully
2689 any remaining bugs will be simpler, less fundamental, and more
2690 fixable then the bugs in the old code.
2691 ** The revised compiler is still a little unsteady on its feet.
2693 *** The debugging information it produces (particularly the names
2694 of FUNCTION objects) is sometimes much less useful than what
2695 the old compiler produced.
2696 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2697 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2698 within a MACROLET) has been temporarily weakened.
2699 ** There are new compiler optimizations for various functions:
2700 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2701 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2702 *** the math functions TRUNCATE, FLOOR, and CEILING
2703 *** the function-of-all-trades COERCE
2704 Mostly these should be transparent, but there's one
2705 potentially-annoying problem (bug 117): when the compiler
2706 inline-expands a function and does type analysis on the result,
2707 it can create control paths which have type mismatches, and
2708 when it can't prove that those control paths aren't taken,
2709 it will issue WARNINGs about the type mismatches. This is
2710 a particular problem in practice for the new sequence functions.
2711 It's not clear how this should be fixed, and for now, a
2712 workaround is given in the entry for 117 in the BUGS file.
2713 ** (Because of the interaction between the two previous items --
2714 occasional inlining problems and new inline expansions -- some
2715 of the new sequence function optimizations won't really kick in
2716 completely until debugging information, and then inlining, are
2717 straightened out in some future version.)
2718 * minor incompatible changes:
2719 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2720 DIRECTORY no longer implicitly promotes NIL slots of its
2721 pathname argument to :WILD. In particular, when you ask for the
2722 contents of a directory (which you used to be able to do without
2723 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2724 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2725 ** changes in behavior that ANSI explicitly defines to be
2726 implementation dependent:
2727 *** The new compiler-only implementation still conforms with ANSI,
2728 but acts a little different than before. Besides the obvious
2729 changes in performance tradeoffs (that the cost per form passed
2730 to EVAL has gone up, and the cost per form executed by EVAL
2731 has gone down), the behavior of the system changes a little
2732 because there are no longer any interpreted function objects.
2733 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2734 e.g. doing COMPILE on the output of interactive DEFUN is
2736 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2738 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2739 has changed from (:CL) to NIL.
2740 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2741 e.g. #\Newline and #\Tab, don't have names specified in the
2742 ANSI Common Lisp standard, is now based on their ASCII symbolic
2743 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2744 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2745 are no longer used for output.
2746 ** changes in internal implementation constants:
2747 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2748 4 million. (If your application spends a lot of time GCing and
2749 you have a lot of RAM, you might want to experiment with
2750 increasing it even more.)
2751 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2752 However, almost all old code should still continue to work without
2753 immediate update, as SB-C-CALL is now a (deprecated) nickname
2755 ** Old operator names in the style DEF-FOO are now deprecated in
2756 favor of new corresponding names DEFINE-FOO, for consistency with
2757 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2758 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2759 internal symbols, but a few supported extensions like
2760 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2761 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2762 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2763 etc. as you get deeper into recursive calls to the debugger
2764 command loop, instead of the old "5]", "5]]", "5]]]"
2765 sequence. (I was motivated to do this when squabbles between
2766 ILISP and SBCL left me very deeply nested in the debugger. In the
2767 short term, this change will probably provoke more ILISP/SBCL
2768 squabbles, but hopefully it will be an improvement in the long run.)
2769 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2770 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2771 function, since declaiming return types involving aliens is
2772 (1) annoyingly messy to do by hand and (2) vital to efficient
2773 compilation of code which calls such functions.
2774 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2775 longer reexported by the SB-EXT package. They're solely useful
2776 for alien code, so it seems more logical that you should get
2777 them from the SB-ALIEN package, not in SB-EXT.
2778 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2779 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2780 features. Instead, the code that they used to control is always
2781 built into the system.
2782 * many other bug fixes
2783 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2784 advantage of the new EVAL-WHEN stuff and to clean them up in
2785 general, and they are now more ANSI-compliant in a number of
2786 ways. Martin Atzmueller is responsible for a lot of this.
2787 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2789 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2790 *** correct ERROR type for various file operations
2791 *** some fixes for Lisp streams
2792 *** DEFMETHOD syntax checking
2793 *** changing old weird representation of debug information as
2794 strings (which, among their other deficiencies, don't transform
2795 correctly when you rename packages, and don't change their
2796 print representation when you change things like *PACKAGE*
2797 and *PRINT-LENGTH*) to symbols and lists of symbols
2798 He also made several improvements and fixed several bugs in DESCRIBE.
2799 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2801 *** misbehavior of WRITE-STRING/WRITE-LINE
2802 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2803 and several other LOOP problems as well
2804 *** DIRECTORY when similar filenames are present
2805 *** DEFGENERIC with :METHOD options
2806 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2807 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2808 *** argument ordering in FIND with :TEST option
2809 *** mishandled package designator argument in APROPOS-LIST
2810 *** various problems in the backquote readmacro
2811 *** a bug in APROPOS
2812 *** probably some others that I'm not describing very well here,
2813 since the CVS log documents them by reference to sbcl-devel
2814 messages, and the SourceForge archives aren't working well.:-(
2815 ** Dan Barlow improved the Alpha port (and is making progress on the
2816 PPC port, for those of you who think different).
2817 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2818 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2819 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2820 continues to work on the SPARC port (for those of you in a position
2821 to look down upon our little PC-compatible boxes from a great height).
2822 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2823 and uses that copy, rather than the current dynamic value, when
2824 it's trying to decide whether to truncate output. Thus e.g.
2825 (let ((*print-lines* 50))
2826 (pprint-logical-block (stream nil)
2828 (let ((*print-lines* 8))
2829 (print (aref possiblybigthings i) stream)))))
2830 should now truncate the logical block only at 50 lines, instead of
2831 often truncating it at 8 lines, as it did before.
2832 * The doc/cmucl/ directory, containing old CMU CL documentation
2833 from the time of the fork, is no longer part of the base system.
2834 SourceForge has shut down its anonymous FTP service, and with it
2835 my original plan for distributing the old CMU CL documentation
2836 there. For now, if you need these files you can download an old
2837 SBCL source release and extract them from it.
2838 * The fasl file version number changed again, for dozens of reasons,
2839 some of which are apparent above.
2841 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2842 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2843 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2844 patch, so that DISASSEMBLE output is much nicer.
2845 * The code in the SB-PROFILE package now seems reasonably stable.
2846 I still haven't decided what the final interface should look like
2847 (I'd like PROFILE to interact cleanly with TRACE, since both
2848 facilities use function encapsulation) but if you have a need
2849 for profiling now, you can probably use it successfully with
2850 the current CMU-CL-style interface.
2851 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2852 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2853 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2854 link now returns the dangling link itself, and for similar
2855 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2856 link itself. (In these cases the old code signalled an error and
2857 looped endlessly, respectively.) Thus, DIRECTORY now works even
2858 in the presence of dangling and cyclic symbolic links.
2859 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2860 is now a supported extension again, since the consensus on
2861 sbcl-devel was that it can be useful for ordinary development
2862 work, not just for debugging SBCL itself.
2863 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2864 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2865 that currently-defined functions might be redefined later with
2866 different return types.
2867 * Hash tables can be printed readably, as inspired by CMU CL code
2868 of Eric Marsden and SBCL code of Martin Atzmueller.
2869 * better error handling in CLOS method combination, thanks to
2870 Martin Atzmueller porting Pierre Mai's CMU CL patches
2871 * more overflow fixes for >16Mbyte I/O buffers
2872 * A bug in READ has been fixed, so that now a single Ctrl-D
2873 character suffices to cause end-of-file on character streams.
2874 In particular, now you only need one Ctrl-D at the command
2875 line (not two) to exit SBCL.
2876 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2878 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2879 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2880 does the right thing.
2881 * The implementation of some type tests, especially for CONDITION
2882 types, is now tidier and maybe faster, due to CMU CL code
2883 originally by Douglas Crosher, ported by Martin Atzmueller.
2884 * Some math functions have been fixed, and there are new
2885 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2886 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2887 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2888 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2889 likely to be built into sbcl-0.7.0.)
2890 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2891 CL, ported to SBCL by Martin Atzmueller)
2892 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2893 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2894 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2895 * There's a new slam.sh hack to shorten the edit/compile/debug
2896 cycle for low-level changes to SBCL itself, and a new
2897 :SB-AFTER-XC-CORE target feature to control the generation of
2898 the after-xc.core file needed by slam.sh.
2899 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2900 COMPILE-FILE is no longer supported, so that now every function
2901 gets an entry point, so that block compilation looks a little
2902 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2903 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2904 returns the number of bytes consed since the system started,
2905 rather than the number consed since the first time the function
2906 was called. (The new definition parallels ANSI functions like
2907 CL:GET-INTERNAL-RUN-TIME.)
2908 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2909 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2910 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2911 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2912 * Fasl file version numbers are now independent of the target CPU,
2913 since historically most system changes which required version
2914 number changes have affected all CPUs equally. Similarly,
2915 the byte fasl file version is now equal to the ordinary
2918 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2919 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2920 is no longer recognized. I apologize for this, because it was
2921 listed in SB-EXT as a supported extension, but I found that
2922 its existing behavior was poorly specified, as well as incorrectly
2923 specified, and it looked like too much of a mess to straighten it
2924 out. I have enough on my hands trying to get ANSI stuff to work..
2925 * many patches ported from CMU CL by Martin Atzmueller, with
2926 half a dozen bug fixes in pretty-printing and the debugger, and
2927 half a dozen others elsewhere
2928 * fixed bug 13: Floating point infinities are now supported again.
2929 They might still be a little bit flaky, but thanks to bug reports
2930 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2931 as flaky as they were.
2932 * The --noprogrammer command line option is now supported. (Its
2933 behavior is slightly different in detail from what the old man
2934 page claimed it would do, but it's still appropriate under the
2935 same circumstances that the man page talks about.)
2936 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2937 are now supported, and enabled by default. Thus, the compiler can
2938 handle many floating point and complex operations much less
2939 inefficiently. (Thus e.g. you can implement a complex FFT
2941 * The compiler now detects type mismatches between DECLAIM FTYPE
2942 and DEFUN better, and implements CHECK-TYPE more correctly, and
2943 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2944 thanks to patches from Martin Atzmueller.
2945 * various fixes to make the cross-compiler more portable to
2946 ANSI-conforming-but-different cross-compilation hosts (notably
2947 Lispworks for Windows, following bug reports from Arthur Lemmens)
2948 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2949 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2950 fixed thanks to Pierre Mai's CMU CL patches.
2951 * fixes to keep the system from overflowing internal counters when
2952 it tries to use i/o buffers larger than 16M bytes
2953 * fixed bug 45a: Various internal functions required to support
2954 complex special functions have been merged from CMU CL sources.
2955 (When I was first setting up SBCL, I misunderstood a compile-time
2956 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2957 * improved support for type intersection and union, fixing bug 12
2958 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2959 more obscure bugs as well
2960 * some steps toward byte-compiling non-performance-critical
2961 parts of the system, courtesy of patches from Martin Atzmueller
2962 * Christophe Rhodes has made some debian packages of sbcl at
2963 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2964 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2965 stable, but are nonetheless usable. When he's ready, I'd be happy
2966 to add them to the SourceForge "File Releases" section. (And if
2967 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2968 * new fasl file format version number (because of changes in
2969 internal representation of (OR ..) types to accommodate the new
2970 support for (AND ..) types, among other things)
2972 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2973 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2975 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2977 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2978 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2979 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2980 * DESCRIBE now works on structure objects again.
2981 * Most function call argument type mismatches are now handled as
2982 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2983 know whether the function will be redefined before the call is
2984 executed. (The compiler could flag local calls with full WARNINGs,
2985 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2986 it doesn't keep track of enough information to know whether calls
2987 are local in this sense.)
2988 * Compiler output is now more verbose, with messages truncated
2989 later than before. (There should be some supported way for users
2990 to override the default verbosity, but I haven't decided how to
2991 provide it yet, so this behavior is still controlled by the internal
2992 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2993 src/compiler/ir1util.lisp.)
2994 * Fasl file format version numbers have increased again, because
2995 support for the Gray streams extension changes the layout of the
2996 system's STREAM objects.
2997 * The Gray subclassable streams extension now works, thanks to a
2998 patch from Martin Atzmueller.
2999 * The full LOAD-FOREIGN extension (not just the primitive
3000 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3001 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3002 but like most other programs, it defaults to copying the Unix
3003 environment from the original process instead of starting the
3004 new process in an empty environment.
3005 * Extensions which manipulate the Unix environment now support
3006 an :ENVIRONMENT keyword option which doesn't smash case or
3007 do other bad things. The CMU-CL-style :ENV option is retained
3008 for porting convenience.
3009 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3010 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3012 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3014 * A patch from Martin Atzmueller seems to have solved the SIGINT
3015 problem, and as far as we know, signal-handling now works cleanly.
3016 (If you find any new bugs, please report them!)
3017 * The system no longer defaults Lisp source file names to types
3018 ".l", ".cl", or ".lsp", but only to ".lisp".
3019 * The compiler no longer uses special default file extensions for
3020 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3021 seems to expect a single default extension for all compiled code,
3022 and there's no compelling reason to try to stretch the standard
3023 to allow two different extensions.) Instead, byte-compiled files
3024 default to the same extension as native-compiled files.
3025 * Fasl file format version numbers have increased again, because
3026 a rearrangement of internal implementation packages made some
3027 dumped symbols in old fasl files unreadable in new cores.
3028 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3029 many fewer weird special cases.
3030 * Bug #17 (differing COMPILE-FILE behavior between logical and
3031 physical pathnames) has been fixed, and some related misbehavior too,
3032 thanks to a patch from Martin Atzmueller.
3033 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3034 by Tim Moore, ported to SBCL by Martin Atzmueller.
3035 * Martin Atzmueller fixed several filesystem-related problems,
3036 including bug #36, in part by porting CMU CL patches, which were
3037 written in part by Paul Werkowski.
3038 * More compiler warnings in src/runtime/ are gone, thanks to
3039 more patches from Martin Atzmueller.
3040 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3043 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3045 * DESCRIBE now works on CONDITION objects.
3046 * The debugger now handles errors which arise when trying to print
3047 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3049 * The build system now uses an additional file, customize-target-features.lisp,
3050 to allow local modifications to the target *FEATURES* list. (The point of
3051 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3052 debugging features enabled, without having to worry about propagating it
3053 into everyone's system when I do a "cvs update".) When no
3054 customize-target-features.lisp file exists, the target *FEATURES* list
3055 should be constructed the same way as before.
3056 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3057 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3058 warn about behavior which is undefined under the ANSI spec. Note: This
3059 is specified by ANSI, but it's not very popular with programmers.
3060 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3061 macro in the SBCL sources for an example of a workaround which you
3062 might use to make portable ANSI-standard code which does what you want.
3063 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3064 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3065 when it's used as a non-toplevel form. (This is required in order
3066 to implement the DEFCONSTANT-EQX macro.)
3067 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3068 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3069 funny things, due to bugs in the implementation of EVAL-WHEN
3070 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3071 non-toplevel DEFCONSTANTs are uncommon.)
3072 * The core file version number and fasl file version number have been
3073 incremented, because the old noncompliant DEFCONSTANT behavior involved
3074 calling functions which no longer exist, and because I also took the
3075 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3076 * fixed bug 1 (error handling before read-eval-print loop starts), and
3077 redid debugger restarts and related debugger commands somewhat while
3079 ** The QUIT debugger command is gone, since it did something
3080 rather different than the SB-EXT:QUIT command, and since it never
3081 worked properly outside the main toplevel read/eval/print loop.
3082 Invoking the new TOPLEVEL restart provides the same functionality.
3083 ** The GO debugger command is also gone, since you can just invoke
3084 the CONTINUE restart directly instead.
3085 ** The TOP debugger command is also gone, since it's redundant with the
3086 FRAME 0 command, and since it interfered with abbreviations for the
3088 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3090 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3091 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3092 some porting work by Martin Atzmueller)
3093 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3094 with a boatload of patches which he ported to SBCL. Now that those
3096 ** The system tries to make sure that its low-priority messages
3097 are prefixed by semicolons, to help people who like to use
3098 syntax highlighting in their ILISP buffer. (This patch
3099 was originally due to Raymond Toy.)
3100 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3101 patches originally written by Raymond Toy.
3102 ** The compiler understands coercion between single-value and
3103 multiple-VALUES type expressions better, getting rid of some very
3104 weird behavior, thanks to patches originally by Robert MacLachlan
3105 and Douglas Crosher.
3106 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3107 lambda lists, thanks to a patch originally by Pierre Mai.
3108 ** The system no longer bogusly warns about "abbreviated type
3110 ** The compiler gets less confused by inlining and RETURN-FROM,
3111 thanks to some patches originally by Tim Moore.
3112 ** The system no longer hangs when dumping circular lists to fasl
3113 files, thanks to a patch originally from Douglas Crosher.
3114 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3115 undefined function error.
3116 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3117 in structure slot initforms) for now, documented workaround instead:-|
3118 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3119 * fixed bug 5: added stubs for various Gray stream functions called
3120 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3121 installed, at least appropriate type errors are generated
3122 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3123 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3124 consistently in DEFMETHOD forms.
3125 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3126 it doesn't seem to affect SBCL after all
3127 * The C runtime system now builds with better optimization and many
3128 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3130 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3132 * The system is now under CVS at SourceForge (instead of the
3133 CVS repository on my home machine).
3134 * The new signal handling code has been tweaked to treat register
3135 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3136 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3137 downstream, e.g. in debug-int.lisp, has implicit dependencies
3138 on the unsignedness of integer representation of machine words,
3139 and that caused the system to bomb out with infinite regress
3140 when trying to recover from type errors involving signed values,
3141 e.g. (BUTLAST '(1 2 3) -1).)
3142 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3143 (This was one of the bugs Peter Van Eynde reported back in July.)
3144 * The system now uses code inspired by Colin Walters' O(N)
3145 implementation of MAP (from the cmucl-imp@cons.org mailing
3146 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3147 inline the MAP operation, and there is more than one
3148 sequence argument to the MAP call (so that it can't just
3149 do ETYPECASE once and for all based on the type of the
3150 single sequence argument). (The old non-inline implementation
3151 of the general M-argument sequence-of-length-N case required
3152 O(M*N*N) time when any of the sequence arguments were LISTs.)
3153 * The QUIT :UNIX-CODE keyword argument has been renamed to
3154 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3155 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3157 * Raymond Wiker's patches to port dynamic loading from Linux to
3158 FreeBSD have been added.
3159 * The BUGS file is now more nearly up to date, thanks in large part
3160 to Martin Atzmueller's review of it.
3161 * The debugger now flushes standard output streams before it begins
3162 its output ("debugger invoked" and so forth).
3163 * The core version number and fasl file version number have both
3164 been incremented, because of incompatible changes in the layout
3166 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3167 used to not be called for a saved Lisp image.)
3168 * Martin Atzmueller's version of a patch to fix a compiler crash,
3169 as posted on sbcl-devel 13 September 2000, has been installed.
3170 * Instead of installing Martin Atzmueller's patch for the
3171 compiler transform for SUBSEQ, I deleted the compiler transform,
3172 and transforms for some similar consing operations.
3173 * A bug in signal handling which kept TRACE from working on OpenBSD
3175 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3176 away by constant folding
3177 * The system now defines its address space constants in one place
3178 (in the Lisp sources), and propagates them automatically elsewhere
3179 (through GENESIS and the sbcl.h file). Therefore, patching the
3180 address map is less unnecessarily tedious and error-prone. The
3181 Lisp names of address space constants have also been systematized.
3182 * CVS tags like dollar-Header-dollar have been removed from
3183 the sources, because they have never saved me trouble and
3184 they've been source of trouble working with patches and other
3185 diff-related operations.
3186 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3187 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3189 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3191 * The system has been ported to OpenBSD.
3192 * The system now compiles with a simple "sh make.sh" on the systems
3193 that it's supported on. I.e., now you no longer need to tweak
3194 text in the target-features.lisp-expr and symlinks in src/runtime/
3195 by hand, the make.sh takes care of it for you.
3196 * The system is no longer so grossly inefficient when compiling code
3197 involving vectors implemented as general (not simple) vectors (VECTOR T),
3198 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3199 which dares to use the various sequence functions on non-simple
3200 vectors, takes less of a performance hit.
3201 * There is now a primitive type predicate VECTOR-T-P
3202 to test for the (VECTOR T) type, so that e.g.
3203 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3204 can now be compiled with some semblance of efficiency. (The old code
3205 turned the type declaration into a full call to %TYPEP at runtime!)
3206 * AREF on (VECTOR T) is still not fast, since it's still compiled
3207 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3208 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3209 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3210 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3211 without making full calls to SUBTYPEP at runtime.
3212 (Some analogous efficiency issues for non-simple vectors specialized to
3213 element types other than T, or for non-simple multidimensional arrays,
3214 have not been addressed. They could almost certainly be handled the
3215 same way if anyone is motivated to do so.)
3216 * The changes in array handling break binary compatibility, so
3217 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3218 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3219 instead of (VALUES T T).
3220 * By following the instructions that Dan Barlow posted to sbcl-devel
3221 on 2 July 2000, I was able to enable primitive dynamic object
3222 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3223 functionality is not implemented (since it calls ld to resolve
3224 library references automatically, requiring RUN-PROGRAM for its
3225 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3226 try to resolve library references) is now supported.
3227 * The system now flushes the standard output streams when it terminates,
3228 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3229 them at several other probably-convenient times, e.g. in each pass of
3230 the toplevel read-eval-print loop, and after evaluating a form given
3231 as an "--eval" command-line option. (These changes were motivated by a
3232 discussion of stream flushing issues on cmucl-imp in August 2000.)
3233 * The source transform for TYPEP of array types no longer assumes
3234 that an array whose element type is a not-yet-defined type
3235 is implemented as an array of T, but instead punts, so that the
3236 type will be interpreted at runtime.
3237 * There is now some support for cross-compiling in make.sh: each of
3238 the phases of make.sh has its own script. (This should be transparent
3239 to people doing ordinary, non-cross-compile builds.)
3240 * Since my laptop doesn't have hundreds of megabytes of memory like
3241 my desktop machine, I became more motivated to do some items on
3242 my to-do list in order to reduce the size of the system a little:
3243 ** Arrange for various needed-only-at-cold-init things to be
3244 uninterned after cold init. To support this, those things have
3245 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3246 symbols with such names are now uninterned after cold init).
3247 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3248 fasl files for cold load.
3249 ** Remove the old compiler structure pooling code (which used to
3250 be conditional on the target feature :SB-ALLOC) completely.
3251 ** Redo the representation of some data in cold init to be more compact.
3252 (I also looked into supporting byte compiled code at bootstrap time,
3253 which would probably reduce the size of the system a lot, but that
3254 looked too complicated, so I punted for now.)
3255 * The maximum signal nesting depth in the src/runtime/ support code has
3256 been reduced from 4096 to 256. (I don't know any reason for the very
3257 large old value. If the new smaller value turns out to break something,
3258 I'll probably just bump it back up.)
3259 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3261 * Many, many bugs reported by Peter Van Eynde have been added to
3262 the BUGS list; some have even been fixed.
3263 * While enabling dynamic object file loading, I tried to make the
3264 code easier to understand, renaming various functions and variables
3265 with less ambiguous names, and changing some function calling
3266 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3267 * While trying to figure out how to do the OpenBSD port, I tried to
3268 clean up some of the code in src/runtime/. In particular, I dropped
3269 support for non-POSIX signal handling, added various comments,
3270 tweaked the code to reduce the number of compilation warnings, and
3271 renamed some files to increase consistency.
3272 * To support the new automatic configuration functionality in make.sh,
3273 the source file target-features.lisp-expr has been replaced with the
3274 source file base-target-features.lisp-expr and the machine-generated
3275 file local-target-features.lisp-expr.
3276 * fixed a stupid quoting error in make.sh so that using CMU CL
3277 "lisp -batch" as cross-compilation host works again
3279 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3281 * DESCRIBE no longer tries to call itself recursively to describe
3282 bound/fbound values, so that it no longer fails on symbols which are
3283 bound to themselves (like keywords, T, and NIL).
3284 * DESCRIBE now works on generic functions.
3285 * The printer now prints less-screwed-up representations of closures
3286 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3287 * A private symbol is used instead of the :EMPTY keyword previously
3288 used to mark empty slots in hash tables. Thus
3289 (DEFVAR *HT* (MAKE-HASH-TABLE))
3290 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3291 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3292 now does what ANSI says that it should. (You can still get
3293 similar noncompliant behavior if bang on the hash table
3294 implementation with all the symbols you get back from
3295 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3296 This breaks binary compatibility, since tests for equality to
3297 :EMPTY are wired into things like the macroexpansion of
3298 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3300 * There's now a minimal placeholder implementation for CL:STEP,
3301 as required by ANSI.
3302 * An obscure bug in the interaction of the normal compiler, the byte
3303 compiler, inlining, and structure predicates has been patched
3304 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3305 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3307 * Missing ordinary arguments in a macro call are now detected even
3308 when the macro lambda list contains &KEY or &REST.
3309 * The debugger no longer complains about encountering the top of the
3310 stack when you type "FRAME 0" to explicitly instruct it to go to
3311 the top of the stack. And it now prints the frame you request even
3312 if it's the current frame (instead of saying "You are here.").
3313 * As specified by ANSI, the system now always prints keywords
3314 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3316 * The default initial SIZE of HASH-TABLEs is now smaller.
3317 * Type information from CLOS class dispatch is now propagated
3318 into DEFMETHOD bodies, so that e.g.
3319 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3321 is now basically equivalent to
3322 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3323 (DECLARE (TYPE SINGLE-FLOAT X))
3325 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3326 operation, without having to do run-time type dispatch.
3327 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3328 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3329 * Since I don't seem to be making big file reorganizations very often
3330 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3331 is overflowing my ability to conveniently back them up), I've finally
3332 checked the system into CVS. (The CVS repository is on my home system,
3333 not at SourceForge -- putting it on SourceForge might come later.)
3334 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3335 high-level GC-NOTIFY-FOO functions send their output. (There's
3336 still very little control of where low-level verbose GC functions
3337 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3338 less than it used to -- the GC-NOTIFY-FOO functions are now under
3339 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3340 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3341 in only one place in the source code, and propagates it automatically
3342 everywhere that it's needed. Thus e.g. when I bump the version from
3343 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3344 * The C source files now include boilerplate legalese and documentation
3345 at the head of each file (just as the Lisp source files already did).
3346 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3347 to his page will be replaced with a link to his new CLiki service.
3349 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3351 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3352 * The build process now looks for GNU make under the default name "gmake",
3353 instead of "make" as it used to. If GNU make is not available as "gmake"
3354 on your system, you can change this default behavior by setting the
3355 GNUMAKE environment variable.
3356 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3357 can build without error under CMU CL.
3359 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3361 * There is now a partial SBCL user manual (with some new text and some
3362 text cribbed from the CMU CL manual).
3363 * The beginnings of a profiler have been added (starting with the
3364 CMU CL profiler and simplifying and cleaning up). Eventually the
3365 main interface should be through the TRACE macro, but for now,
3366 it's still accessed through vaguely CMU-CL-style functions and macros
3367 exported from the package SB-PROFILE.
3368 * Some problems left over from porting CMU CL to the new
3369 cross-compilation bootstrap process have been cleaned up:
3370 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3371 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3372 part when getting rid of the compiler *BACKEND* stuff.)
3373 ** The value of *NULL-TYPE* was screwed up, because it was
3374 being initialized before the type system knew the final
3375 definition of the 'NULL type. This screwed up several key
3376 optimizations in the compiler, causing inefficiency in all sorts
3377 of places. (I found it because I wanted to understand why
3378 GET-INTERNAL-RUN-TIME was consing.)
3379 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3380 PROCLAIM FTYPE information. Unfortunately this broke binary
3381 compatibility again, since now the forms output by DEFGENERIC
3382 to refer to functions which didn't exist in 0.6.3.
3383 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3384 can use the new (as of 0.6.3) transform for SOME into MAP into
3386 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3387 values to (INTEGER 0 1000000), so that the time code will no longer
3388 occasionally get blown up by Linux returning 1000000 microseconds
3389 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3390 its output conform to the ANSI spec. (Alas, this makes its output
3391 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3392 * A full call to MAP NIL with a single sequence argument no longer conses.
3393 * fixes to problems pointed out by Martin Atzmueller:
3394 * The manual page no longer talks about multiprocessing as though
3395 it were currently supported.
3396 * The ILISP support patches have been removed from the distribution,
3397 because as of version 5.10.1, ILISP now supports SBCL without us
3398 having to maintain patches.
3399 * added a modified version of Raymond Toy's recent CMU CL patch for
3400 EQUALP comparison of HASH-TABLE
3402 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3404 * The system still can't cross-compile itself with
3405 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3406 entails), but at least it can compile more of itself that way
3407 than it used to be able to, and various buglets which were uncovered
3408 by trying to cross-compile itself that way have now been fixed.
3409 * This release breaks binary compatibility again. This time
3410 at least I've incremented the FASL file format version to 2, so that the
3411 problem can be detected reliably instead of just causing weird errors.
3412 * various new style warnings:
3413 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3414 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3415 ** using the lexical binding of a variable named in the *FOO* style
3416 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3417 as specified by ANSI.
3418 * *RANDOM-STATE* is no longer automatically initialized from
3419 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3420 default behavior of the system is to repeat its behavior every time
3421 it's run. If you'd like to change this behavior, you can always
3422 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3423 old convention there was no comparably easy way to get the system to
3424 repeat its behavior every time it was run.
3425 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3426 FTYPE declarations has been removed, in favor of their ANSI
3427 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3428 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3429 the types of their sequence arguments can be determined at compile time.
3430 This is done through a new open code expansion for MAP which eliminates
3431 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3432 argument types can be determined at compile time.
3433 * The optimizer now transforms COERCE into an identity operation when it
3434 can prove that the coerced object is already of the correct type. (This
3435 can be a win for machine generated code, including the output of other
3436 optimization transforms, such as the MAP transform above.)
3437 * Credit information has been moved from source file headers into CREDITS.
3438 * Source file headers have been made more standard.
3439 * The CASE macro now compiles without complaining even when it has
3442 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3444 * (Note that the way that the PCL macroexpansions were rewritten
3445 to accommodate the change in DEFGENERIC below breaks binary
3446 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3447 not run under sbcl-0.6.2. Once we get out of alpha releases,
3448 i.e. hit release 1.0.0, we'll probably try to maintain binary
3449 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3450 and sbcl-1.4.4. Until then, however, it might be fairly common
3451 for maintenance releases to break binary compatibility.)
3452 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3454 * The reporting of unhandled signals has been changed to print some
3455 explanatory text as well as the report form. (Previously only
3456 the report form was printed.)
3457 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3458 it defines, so that the compiler no longer issues undefined function
3459 warnings for compiled-but-not-yet-loaded generic functions.
3460 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3461 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3462 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3463 section of the ANSI spec are supported.
3464 * The "" nickname for the "KEYWORD" package has been removed.
3465 The reader still handles symbol tokens which begin with a package marker
3466 as keywords, but it doesn't expose its mechanism for doing so in the
3467 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3468 * The system now issues STYLE-WARNINGs for contradictory TYPE
3469 proclamations. (Warnings for contradictory FTYPE proclamations would
3470 be nice too, but those can't be done usefully unless the type system
3471 is made smarter about FUNCTION types.)
3472 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3473 systematized, so that "*target-*.lisp is supposed to exist only on the
3474 target and imply that there's a related file which exists on the
3475 host, and *host-*.lisp is supposed to exist only on the host and imply
3476 that there's a related file which exists on the target. This involves a
3477 lot of renaming. Hopefully the acute confusion caused by the renaming
3478 will be justified by the reduction in chronic confusion..
3479 ** runtime-type.lisp -> early-target-type.lisp
3480 ** target-type.lisp -> late-target-type.lisp
3481 ** early-host-format.lisp -> early-format.lisp
3482 ** late-host-format.lisp -> late-format.lisp
3483 ** host-error.lisp -> misc-error.lisp
3484 ** early-error.lisp -> early-target-error.lisp
3485 ** late-error.lisp -> late-target-error.lisp
3486 ** host-defboot.lisp -> early-defboot.lisp
3487 ** code/misc.lisp -> code/target-misc.lisp
3488 ** code/host-misc.lisp -> code/misc.lisp
3489 ** code/numbers.lisp -> code/target-numbers.lisp
3490 ** code/early-numbers.lisp -> numbers.lisp
3491 ** early-host-type.lisp -> early-type.lisp
3492 ** late-host-type.lisp -> late-type.lisp
3493 ** host-typep.lisp -> typep.lisp
3494 ** load.lisp -> target-load.lisp
3495 ** host-load.lisp -> load.lisp
3496 ** host-disassem.lisp -> disassem.lisp
3497 ** host-insts.lisp -> insts.lisp
3498 ** byte-comp.lisp -> target-byte-comp.lisp
3499 ** host-byte-comp.lisp -> byte-comp.lisp
3500 ** host-signal.lisp -> signal.lisp
3501 ** host-defstruct.lisp -> defstruct.lisp
3502 ** late-target-type.lisp -> deftypes-for-target.lisp
3503 Furthermore, several other previously target-only files foo.lisp (e.g.
3504 hash-table.lisp and random.lisp) have been split into a target-and-host
3505 foo.lisp file and a target-only target-foo.lisp file, with their key type
3506 definitions in the target-and-host part, so that the cross-compiler will
3507 know more about target types.
3508 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3509 gone away. In their place are various *BACKEND-FOO* variables
3510 corresponding to the slots of the old structure.
3511 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3512 into the target SBCL has been fixed.
3513 * The chill.lisp system for loading cold code into a running SBCL
3515 * Support for the CMU CL "scavenger hook" extension has been removed.
3516 (It was undocumented and unused in the CMU CL sources that SBCL was
3517 derived from, and stale in sbcl-0.6.1.)
3518 * Various errors in the cross-compiler type system were detected
3519 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3520 (enabling various consistency checks). Many of them were fixed,
3521 but some hard problems remain, so the compiler is back to
3522 running without *TYPE-SYSTEM-INITIALIZED* for now.
3523 * As part of the cross-compiler type system cleanup, I implemented
3524 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3525 * I have started adding UNCROSS calls throughout the type system
3526 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3527 on cross-compiler input files will be able to go away, and various
3529 * CONSTANTP now returns true for quoted forms (as explicitly required
3532 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3534 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3535 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3536 (MAKE-LIST -1). In the longer term, it ought to become true
3537 that declarations are assertions even at SAFETY 1. For now, it's not
3538 quite true even at SAFETY 3, but it's at least more nearly true..
3539 (Note that this change seems to increases the size of the system by
3540 O(5%) and to decrease the speed of the compiler by 20% or more.)
3541 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3542 for the problem of printing dozens of lines of distracting information
3543 about low-level system machinery as part of the top stack frame
3544 on entry to the debugger when an undefined function was called.
3545 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3546 is not reset to COMMON-LISP-USER.
3547 * Compilation of stuff related to dyncount.lisp has been made conditional
3548 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3549 smaller. The various dyncount-related symbols have been moved into
3550 a new "SB-DYNCOUNT" package.
3551 * tty-inspect.lisp has been renamed to inspect.lisp.
3552 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3553 feature has gone away. (When we eventually port to other flavors of
3554 libc and/or Unix, we'll try to make the differences between flavors
3555 invisible at the user level.)
3556 * Various other *FEATURES* tags, and/or their associated conditionals,
3557 have been removed if obsolescent, or given better documentation, or
3558 sometimes given more-mnemonic names.
3560 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3562 * tidied up "make.sh" script
3563 * tidied up system directory structure
3564 * better "clean.sh" behavior
3565 * added doc/FOR-CMUCL-DEVELOPERS
3566 * many many small tweaks to output format, e.g. removing possibly-confusing
3567 trailing #\. character in DESCRIBE-INSTANCE
3568 * (EQUALP #\A 'A) no longer signals an error.
3569 * new hashing code, including EQUALP hashing
3570 * tidied up Lisp initialization and toplevel
3571 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3572 * command line argument processing
3573 * added POSIX-GETENV function to deal with Unix-ish environment variables
3574 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3575 terminating SBCL on EOF
3576 * non-verbose GC by default
3577 * There is no more "sbcl" shell script; the sbcl file is now the C
3578 runtime executable (just like CMU CL).
3579 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3581 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3582 symbol and package information as Lisp native symbol and package objects
3583 instead of strings naming symbols and strings naming packages. This way,
3584 whenever packages are renamed (as in warm init), debug information is
3585 transformed along with everything else.
3586 * tweaked the optimization policy declarations which control the building
3587 of SBCL itself. Now, among other things, the system no longer saves
3588 source location debugging information. (This helps two problems at once
3589 by reducing SBCL size and by keeping SBCL from trying to look for its
3590 sources -- which may not exist -- when reporting errors.)
3591 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3592 debugging and testing purposes
3593 * cleaned up printing, making the printer call PRINT-OBJECT for
3594 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3595 methods, giving nearly-ANSI behavior
3596 * converted almost all special variables to use *FOO* naming convention
3597 * deleted PARSE-TIME functionality, since it can be done portably
3598 * moved some files out of cold init into warm init
3599 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3601 * regularized formatting of source files
3602 * added an install.sh script
3603 * fixed ridiculous memory usage of cross-compiler by making
3604 compiler/alloc.lisp not try to do pooling unless it can hook
3605 itself into the GC of the cross-compilation host. Now the system
3606 builds nicely on my old laptop.
3607 * added :SB-ALLOC in target-features.lisp-expr
3608 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3609 was not implemented)
3610 * re-did condition handling and note reporting in the compiler. Notes
3611 are no longer handled by signalling conditions. Style warnings
3612 and warnings are handled more correctly and reported in such a way
3613 that it's easy to find one or the other in your output (so that you
3614 can e.g. figure out which of many problems caused COMPILE-FILE to
3616 * changed the severity of several compiler warnings from full WARNING
3617 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3618 compiler note reporting so that it doesn't use the condition system
3619 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3620 COMPILE-FILE command)
3621 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3622 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3623 no longer supported, since their implementation was deeply intertwingled
3624 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3625 * removed bogus "support" for compiler macros named (SETF FOO), and
3626 removed the compiler macro for SETF INFO (but only after making a fool
3627 of myself on the cmucl-imp mailing list by posting a bogus patch for
3628 DEFINE-COMPILER-MACRO..)
3629 * Compiled files containing forms which have side effects on the Lisp
3630 reader (such as DEFPACKAGE forms) are now handled more correctly.
3631 (Compiler queuing of top level lambdas has been suppressed by setting
3632 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3633 * deleted various currently-unused source files, e.g. gengc.lisp. They
3634 may be added back at some point e.g. when porting to other architectures,
3635 but until they are it's distracting to distribute them and to try to
3637 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3638 there were so many of them they're just distractions, and UNCROSS is
3639 known to be able to handle the current sources
3640 * moved PROFILE functionality into TRACE, so that it will be clear
3641 how the wrapping and unwrapping of functions when you profile them
3642 interacts with the wrapping and unwrapping of functions when you
3643 trace them. (Actually, the functionality isn't there yet, but at least
3644 the interface specification is there. Hopefully, the functionality will
3645 arrive with some future maintenance release.)
3646 * removed host-oops.lisp
3647 * changed signature of QUIT function to allow UNIX-CODE argument
3648 * fixed READ-SEQUENCE bug
3649 * tweaked verbose GC output so that it looks more like the progress
3650 output that ANSI specifies for functions like LOAD
3651 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3652 * added <http://sbcl.sourceforge.com> to the banner information printed by