1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
3 * Enhancements for sbcl running on the Windows operating system:
4 ** (user-homedir-pathname) and default initialization file
5 locations now know about the user's "Documents and Settings"
6 directory (thanks to Yaroslav Kavenchuk)
7 ** run-program is implemented (thanks to Mike Thomas)
8 ** sockets support (thanks to Timothy Ritchey)
9 * new feature: command line options --no-sysinit, --no-userinit to
10 inhibit loading the corresponding init files
11 * bug fix: LISTEN sometimes returned T even in cases where no data was
12 immediately available from the stream
13 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
14 were reversed. (reported by Levente Mészáros)
15 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
16 when given a vector displaced to an adjustable array. (reported
18 * fixed some bugs revealed by Paul Dietz' test suite:
19 ** REMOVE-METHOD returns its generic function argument even when
20 the method is not one of the generic functions' methods.
21 ** objects declared as MEMBER types can be admissible arguments to
23 ** printing characters should simply be printed by the FORMAT ~:C
26 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
27 * new platform: experimental support for SBCL x86/Darwin, including
28 MacOS X 10.4.5 on Intel.
29 * new feature: Unicode character names are now known to the system
30 (through CHAR-NAME and NAME-CHAR).
31 * new feature: the filesystem location of SBCL's core file is
32 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
33 * minor incompatible change: the contrib modules SB-POSIX and
34 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
35 this change is to make it easier to distribute
36 location-independent binaries.
37 * bug fix: as implied by AMOP, standardized classes no longer have
38 slots named by external symbols of public packages. (reported by
40 * optimization: calling functions via constant symbols -- as in
41 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
42 via the function object as in (FUNCALL #'FOO).
43 * optimization: on x86 and x86-64, the calling convention for the
44 case of dealing with an unknown number of values has been altered
45 to be friendlier to the prediction heuristics implemented,
46 particularly with reference to CALL and RET pairing. (thanks to
48 * optimization: on x86, the code for access to thread-local symbol
49 values and binding/unbinding of thread-local symbols is smaller.
50 (thanks to Alastair Bridgewater)
51 * enhancement: CONSTANTP is now able to determine constantness of
52 more complex forms, including calls to constant-foldable standardized
53 functions and some special forms beyond QUOTE.
54 * fixed bug: occasional GC crashes on Solaris/x86
55 * optimization: x86-64 supports stack allocation of results of simple
56 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
57 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
58 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
59 (thanks to James Knight)
60 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
61 by the standard (thanks to Stephen Compall)
63 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
64 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
65 be used for bundling the runtime and the core file into one
66 executable binary. This feature is not currently supported on all SBCL
67 platforms. (thanks to James Bielman and NIIMI Satoshi)
68 * new feature: a generational or ephemeral garbage collector is now
69 the default on the PowerPC platform (both Linux and Darwin). The
70 old Cheney (stop and copy) collector is a build-time option.
71 (thanks to Cyrus Harmon, after Raymond Toy)
72 * minor incompatible change: the method by which SBCL finds its
73 contributed modules has changed; it no longer relies on symbolic
74 links from an $SBCL_HOME/systems directory, but searches directly
75 in the subdirectories of $SBCL_HOME.
76 * enhancement: the dynamic heap size on the Linux/PPC platform is
77 markedly larger, even using the older Cheney garbage collector.
78 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
79 certain complicated string types. (reported by Gary King)
80 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
81 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
82 * fixed bug: DOCUMENTATION on structures no longer signals an
83 error if no structure type of the right name exists. (reported by
85 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
86 applicable non-standard primary or :AROUND method on
87 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
88 no longer cause unbound variable errors. (reported by Kevin Reid)
89 * fixed bug: metacircle resolution in cases where methods have slots
90 added before the slots from STANDARD-METHOD. (reported by Jean
92 * fixed bug: the Power PC assembly code for calling into lisp
93 sometimes computed its return address wrongly (depending on the
94 15th bit of the address, fixed at link time). This used to
95 manifest itself in a segmentation violation while building PCL.
96 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
97 many others over the years)
98 * fixed bug: ROOM no longer reports silly numbers for stack usage if
99 the stack pointer (treated as a signed integer) is negative.
100 (thanks to Peter van Eynde)
102 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
103 * new platform: experimental support for the Windows operating
104 system has been added. (thanks to Alastair Bridgewater)
105 * fixed several bugs in and robustified the PPC FFI (including
106 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
107 * bug fix: allow non-simple string symbol names (reported by Paul
109 * bug fix: interrupt handling on NetBSD (thanks to Richard M
111 * bug fix: saving a core corrupted callbacks on x86/x86-64
112 * bug fix: closed a loophole in metacircularity detection and
113 grounding in the PCL implementation of CLOS.
114 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
115 (reported by tomppa on #lisp)
116 * bug fix: clear the direction flag on Lisp -> C transitions, as
117 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
118 (reported by Andrej Grozin and Hendrik Maryns)
119 * optimization: major improvements to GC efficiency on GENCGC platforms
120 * optimization: faster implementation of EQUAL
121 * optimization: emit more efficient opcodes for some common
122 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
123 * optimization: several other minor code-generation improvements on x86-64
124 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
126 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
127 * minor incompatible change: (SETF CLASS-NAME) and (SETF
128 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
129 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
130 by ANSI as a generic function, and for consistency (SETF
131 GENERIC-FUNCTION-NAME) is treated likewise.
132 * fixed bug #233.b: make constraint propagation notice when a variable
133 value is changed after it is referenced but before it is used
134 * fixed bug #296: no more arbitrary behaviour from filenames with
135 odd characters as --load arguments. (reported by Adam Warner)
136 * fixed bug #390: :CHARACTER-SET pathname components now work as
137 expected. (reported by Tim Daly Jr)
138 * fixed bug #391: complicated :TYPE intersections in slot
139 definitions no longer cause an error in PCL internals.
140 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
141 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
142 returning the number of octets which would be written to the
143 file-stream. (thanks to Robert J. Macomber)
144 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
145 arguments to the internal stream functions. (thanks to David
147 * bug fix: stack exhaustion is now handled on the x86/SunOS
148 platform. (thanks to Daisuke Homma)
149 * bug fix: more accurate ROOM results on GENCGC platforms
150 * optimization: improved type inference for arithmetic-for
151 index variables in LOOP
152 * optimization: faster floating-point SQRT on x86-64
153 * fixed some bugs revealed by Paul Dietz' test suite:
154 ** DOCUMENTATION returns NIL instead of "" for method combinations
155 that don't have a docstring
157 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
158 * minor incompatible change: (SETF CLASS-NAME) and (SETF
159 GENERIC-FUNCTION-NAME) are no longer generic functions, and
160 therefore are not customizeable by user code (as seems to be at
161 least permitted and maybe required by AMOP). As a consolation,
162 however, the SBCL implementation of these functions now calls
163 REINITIALIZE-INSTANCE as specified by AMOP.
164 * bug fix: slot-definition documentation is propagated as per ANSI
165 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
166 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
167 argument, as do eslotd initialization methods. (from Pascal
168 Costanza's "Closer" project)
169 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
170 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
172 * bug fix: it is now possible to have more than one subclass of
173 STANDARD-GENERIC-FUNCTION without causing stack overflow.
174 (reported by Bruno Haible, Pascal Costanza and others)
175 * bug fix: the dependent update protocol now works for generic
176 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
178 * bug fix: condition-class instances corresponding to
179 DEFINE-CONDITION forms are now created eagerly. (reported by
180 Kalle Olavi Niemitalo on comp.lang.lisp)
181 * bug fix: floating point printing is more accurate in some
182 circumstances. (thanks to Simon Alexander)
183 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
184 merged with *DEFAULT-PATHNAME-DEFAULTS*.
185 * bug fix: callbacks on OS X now preserve stack-alignment.
186 (thanks to Cyrus Harmon)
187 * enhancement: the x86-64 disassembler is much better at
188 disassembling SSE instructions. (thanks to Lutz Euler)
189 * enhancement: CHECK-TYPE on a variable now allows type-inference
190 (thanks to Vincent Arkesteijn)
191 * optimization: improved performance of EUC-JP external format.
192 (thanks to NIIMI Satoshi)
193 * optimization: performance improvements to IO on file streams of
194 :ELEMENT-TYPE CHARACTER
195 * optimization: much faster memory allocation on x86-64
196 * optimization: faster garbage collections (latency and throughput)
198 * optimization: faster bignum subtraction and division on x86 and x86-64
199 (thanks to Lutz Euler)
200 * optimization: more accurate type-inference for loops that use a
201 floating point index variable or a negative step.
203 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
204 * bug fix: add a workaround to SBCL looping infinitely at startup on
205 Linux kernels with apparently buggy implementations of personality().
206 (thanks to Svein Ove Aas)
207 * bug fix: Unicode symbols are correctly printed in LDB backtraces
208 (thanks to David Lichteblau)
209 * bug fix: local bindings shadow global symbol macros for
210 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
211 * optimization: non-open coded uses of numeric comparison operators
212 (e.g. >) no longer cons when called with more than one parameter
213 on platforms supporting dynamic-extent allocation.
214 * enhancement: saving cores with foreign code loaded is now supported
215 on MIPS/Linux in addition to the previously supported platforms.
216 * bug fix: FIND-RESTART now tests for activity, not applicability when given
217 a restart object as identifier. (reported by Helmut Eller for CMUCL)
218 * bug fix: division by zero in sb-sprof when no samples were collected
219 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
220 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
221 (thanks to Alexander Kjeldaas)
222 * new feature: ignoring the return values of destructive functions
223 like NREVERSE signals a compile-time style-warning.
224 (thanks to Kevin Reid)
225 * new feature: the alignment of alien structure fields can be
226 explicitly specified. (thanks to Cyrus Harmon)
227 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
228 and dump core on SIGQUIT
230 ** incompatible change: threads do not inherit values of specials
231 from their parents (see manual)
232 ** bug fix: threads stacks belonging to dead threads are freed by the
233 next exiting thread, no need to gc to collect thread stacks anymore
234 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
235 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
236 ** bug fix: dynamic variable and thread start related gc lossage
237 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
239 * fixed some bugs revealed by Paul Dietz' test suite:
240 ** SUBTYPEP is slightly more accurate on heinously complicated
241 CONS types where some of the members have uncertain (in the
242 NIL, NIL sense) type relationships to each other.
244 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
245 no longer purified unless :PURIFY T is explicitly specified.
246 ** Non-purified cores are significantly smaller than before
248 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
249 * new feature: timers based on Zach Beane's excellent timer package
250 * added support for the following external formats: koi8-u,
251 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
252 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
253 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
254 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
255 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
256 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
257 * incompatible change: a threaded SBCL will no longer revert to
258 non-threaded mode on non-NPTL systems, but refuse to start entirely.
259 * bug fix: interrupts are disabled until startup is complete; no
260 more sigsegvs when receiving a signal to soon
261 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
263 * bug fix: add a workaround for the memory randomization features in
264 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
265 workaround will only be in effect on systems with the proc filesystem
267 * bug fix: printing objects of type HASH-TABLE signals a
268 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
270 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
271 have been running for over 50 days. (reported by Gilbert Baumann)
272 * bug fix: the logic for getting names of functions gets less
273 confused when confronded with alternate-metaclass
274 funcallable-instances. (reported by Cyrus Harmon)
275 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
276 consistent, even on internal alternate-metaclass objects.
277 * bug fix: flush closure information collected by physical
278 environment analysis prepass before the main pass. (bug reported
280 * bug fix: compiler pack phase does not modify a hashtable, which is
281 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
282 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
283 may be expanded. (reported by Alexander <malishev> on c.l.l)
284 * bug fix: no more style-warnings when compiling code using LAST
285 under high-SPEED low-SPACE settings. (reported by David Wallin
287 * fixed some incompatibilities between SBCL's MOP and the MOP
289 ** the METAOBJECT class is now implemented;
290 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
291 STANDARD-OBJECT, as required;
292 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
293 now compatible (as required by VALIDATE-SUPERCLASS); there
294 remains a consistency requirement over the presence or absence
295 of the FUNCTION class in the superclass list of finalized
296 classes; see the manual for more details;
297 ** the :AROUND method for COMPUTE-SLOTS on
298 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
299 requested slot ordering.
301 ** bug fix: parent thread now can be gc'ed even with a live
303 ** bug fix: binding a special with PROGV to no value is not
304 the same as that symbol not having been bound (thanks to
306 ** bug fix: binding specials is thread safe (thanks to
308 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
309 and SB-SPROF do not die with 'no handler for signal XX in
310 interrupt_handle_now(..)' anymore
311 ** bug fix: WITH-TIMEOUT works with multiple running threads
312 * fixed some bugs revealed by Paul Dietz' test suite:
313 ** ENSURE-GENERIC-FUNCTION should take a method class object for
314 the :method-class keyword argument.
316 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
317 * new port: the Solaris operating system on x86 processors is now
318 mostly supported, though some rough edges in the environment
319 remain. (thanks to Daisuke Homma)
320 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
321 space for its heap. (thanks to Thiemo Seufer)
322 * optimization: SBCL on MIPS platforms now supports dynamic-extent
323 closures. (thanks to Thiemo Seufer)
324 * minor incompatible change: eof selects abort in the debugger.
325 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
326 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
328 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
329 deprecated, and will go away in a future revision of SBCL.
330 * minor incompatible change: GC-ON and GC-OFF are no longer
331 implemented with a counter, it does not matter how many times gc
332 is switched on or off
333 * bug fix: discriminating functions for generic function classes
334 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
335 make invalid assumptions about method precedence order. (reported
337 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
338 platforms. (thanks to Thiemo Seufer)
339 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
340 * bug fix: correct pathname printing with printer escaping is on.
341 (thanks to Kevin Reid)
342 * bug fix: complex VOP definitions in "user-space" no longer trigger
343 package locks. (reported by Zach Beane)
344 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
345 for funcallable-instance functions, and is no different from
347 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
348 characters in the current readtable. (reported by Nicholas Neuss)
349 * bug fix: SUBTYPEP on various CONS types returns more a more
350 accurate acknowledgment of its certainty.
351 * optimizations: REMOVE-DUPLICATES now runs in linear time on
352 lists in some cases. This partially fixes bug 384.
353 * flush all standard streams before prompting in the REPL and the
355 * bug fix: signal handling and triggering gc do not conflict
356 directly anymore, in particular a high frequency sb-sprof does
357 not prevent gc from running
358 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
359 approximation for timezone and DST information between the
360 universal time epoch and the smallest negative 32-bit time_t.
361 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
362 year 1899 so long as the encoded time is non-negative.
363 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
364 the correct answer for non-constant rotate amounts when performing
365 an inline 32-bit rotation.
367 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
368 there is only one thread in the session
369 ** bug fix: memory leak for streams created in one thread and
370 written to in another
371 ** bug fix: lockup when compiled with gcc4
372 ** bug fix: race that allows the gc to be triggered when gc is
374 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
375 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
377 ** bug fix: finished threads are gc'ed properly
378 * fixed some bugs revealed by Paul Dietz' test suite:
379 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
380 has arguments with a different set of applicable methods from
381 the orignal arguments.
382 ** The type error thrown by MAP now has a correct expected-type
384 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
385 name a compiled function.
386 ** Three MISC tests where a large bignum was improperly coerced to
387 a float that couldn't represent that bignum during type
388 derivation were fixed.
389 ** SUBTYPEP can now handle types involving the negation of a
390 list-form FUNCTION type.
391 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
392 as (COMPLEX (AND RATIO (NOT FIXNUM))).
393 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
395 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
396 * New feature: Experimental support for bivalent streams: streams
397 opened with :element-type :default now allow character and binary
398 (unsigned-byte 8) I/O
399 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
400 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
402 * bug fix: no more highly sporadic "couldn't check whether ~S is
403 readable" when reading a stream and an interrupt hits in the middle
404 of a select system call
405 * compiler better recognizes complex arrays (reported by Eduardo
407 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
408 for variables on SBCL built with linkage-tables. (reported by Luis
410 * various error reporting improvements.
411 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
412 (Thanks to Thiemo Seufer)
413 * bug fix: nested reader invokations work correctly
414 * bug fix: it is possible to have simultaneous references to foreign
415 code and foreign data with the same name.
417 ** added x86-64 support
418 ** incompatible change: the threading api now works with thread
419 objects instead of thread ids
420 ** bug fix: threads are protected from signals and interruption when
421 starting up or going down
422 ** bug fix: a race where an exiting thread could lose its stack to gc
423 ** bug fix: don't halt on infinite error in threads if possible
424 ** fixed numerous gc deadlocks introduced in the pthread merge
425 ** bug fix: fixed thread safety issues in read and print
426 ** bug fix: debugger doesn't hang on session lock if interrupted at
427 an inappropriate moment
428 ** bug fix: run-program is now thread safe(r)
429 ** bug fix: inner with-recursive-lock no longer releases the mutex
430 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
431 * fixed some bugs revealed by Paul Dietz' test suite:
432 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
433 EXPECTED-TYPE slots filled.
434 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
436 ** MAKE-CONDITION accepts classes as type-designators.
437 ** COMPILE may never return NIL.
438 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
439 range before calling Unix time functions
441 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
442 * numerous signal handling fixes to increase stability
443 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
444 * minor incompatible change: we now correctly canonize default
445 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
446 as per the MOP, rather than the historical (INITARG INITFUNCTION
447 INITFORM). (reported by Bruno Haible)
448 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
449 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
450 for more information.
451 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
452 pathname is a directory pathname.
453 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
454 * dynamic space size on PPC has been increased to 768Mb. (thanks to
456 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
457 :METACLASS argument in addition to a class name. (reported by
458 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
459 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
460 (reported by Bruno Haible; patch from Gerd Moellmann)
461 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
463 * bug fix: more cleanups to the floating point exception handling on
464 x86-64 (thanks to James Knight)
465 * bug fix: the compiler does not try to constant fold calls of
466 COERCE and BIT-* functions when they return freshly constructed
467 objects. (reported by Nikodemus Siivola and Paul Dietz)
468 * optimization: improved the MIPS versions of generic (in the
469 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
470 * optimization: direct conversion of (unsigned-byte 32) to floats on
471 the PowerPC platform.
472 * optimization: structure instances with raw slots now use less
473 memory, and probably show better memory locality. (thanks to
475 * optimization: DYNAMIC-EXTENT declarations for lists and closures
476 are treated as requests for stack allocation on the x86-64,
477 Alpha-32, PPC and SPARC platforms.
478 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
479 the wrong file descriptor; implementation of SOCKET-OPEN-P.
480 (thanks to Tony Martinez)
482 ** gcing a dead thread can no longer lead to lockups
483 ** threads block signals until they are set up properly
484 ** errno is no longer shared by threads
485 ** interrupt-thread restores the eflags register on x86
486 ** fixed some lockups due to gc/thread interaction
487 * fixed some bugs revealed by Paul Dietz' test suite:
488 ** invalid dotted lists no longer raise a read error when
490 ** signal an error if a symbol that names a declaration is used
491 as the name of a type, or vice versa
492 ** allow using the (declare (typespec var*)) abbreviation for
493 (declare (type typespec var*)) with all type specifiers
494 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
495 ** always return NIL from PPRINT-POP when OBJECT is NIL
496 ** don't signal errors when pretty-printing malformed LABELS,
497 FLET or MACROLET forms
498 ** declarations in a DOLIST body are also in scope for the
500 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
502 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
505 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
506 * fixed cross-compiler leakages that prevented building a 32-bit
507 target with a 64-bit host compiler.
508 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
509 opened with :IF-EXISTS :OVERWRITE.
510 * fixed bug 281: error for an invalid qualifier in a short-form method
511 combination method is not signalled until the faulty method is called.
512 * bug fix: iteration variable type inferrer failed to deal with open
513 intervals. (reported by Alan Shields)
514 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
515 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
516 or more runtime options were provided to the sbcl binary.
517 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
518 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
519 * optimization: there's now a fast-path for fixnum arguments in the
520 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
521 * optimization: the code generated on x86-64 is more compact thanks
522 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
523 * bug fix: floating-point exception handling now partly works on
524 x86-64. (Thanks to James Knight)
525 * improvement to the MIPS backend from Thiemo Seufer: C-style
526 64-bit long long arguments and return values to alien functions
528 * fixed some bugs revealed by Paul Dietz' test suite:
529 ** the type-error signalled from WARN has a filled-in DATUM slot.
530 ** the type-error required when a stream is not associated with
531 a file has the stream as its datum.
532 ** type-errors on single-floats on x86-64 no longer have
533 :INVALID-OBJECT as the datum
534 ** the type-errors signalled for invalid function names now have
535 a correct expected type
536 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
537 for structures defined with a :TYPE. Documentation strings for
538 typed structures are no longer immediately discarded
539 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
540 broadcast streams and synonym streams. FILE-LENGTH now also works
541 on broadcast streams.
543 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
544 * incompatible change: the --noprogrammer option, deprecated since
545 version 0.7.5, has been removed. Please use the equivalent
546 --disable-debugger option instead.
547 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
549 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
550 inoperational for a while now) has been completely removed.
551 * null lexical environments are now printed as #<NULL-LEXENV>,
552 significantly reducing the amount of clutter in typical backtraces.
553 * documentation on weak pointers, finalization, and after GC hooks
554 has been added to the manual.
555 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
556 as other specialized array types, is much faster. SUBSEQ and
557 COPY-SEQ on such arrays have also been sped up.
558 * optimization: EQL is now more efficient when at least other argument
559 is known to be of type (OR FIXNUM (NOT NUMBER)).
560 * fixed bug: the runtime is now less vulnerable to changes in the
561 size of the SBCL object on OS X, and virtual memory is reserved for
562 all spaces that need to be at a fixed address.
563 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
564 * fixed bug: finalizers and after GC hooks that cause consing are now
566 * fixed bug: compiler error messages and summaries are now printed to
567 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
568 * fixed inference of the upper bound of an iteration variable.
569 (reported by Rajat Datta).
570 * fixed bug 211e: calling local functions with duplicated constant
571 keyword argument no longer causes a bogus style warning about an
573 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
574 ftype declaration to be disregarded. (reported by Dave Roberts)
575 * fixed bug 373: caused by erronous compilation of references to alien
576 variables in the runtime on ppc/darwin.
577 * fixed bug 376: CONJUGATE type deriver.
578 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
579 (reported by Baughn on #lisp)
580 * a cross-compiler bug on non-x86 platforms has been identified and
581 fixed. (thanks to Bruno Haible)
582 * improvements to the MIPS runtime code for increased stability.
583 (thanks to Thiemo Seufer)
584 * increased the maximimum compact environment size to allow
585 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
586 * partial workaround for bug 135: don't name the function we're
587 calling for hairy cases of EVAL, so as not to accumulate one environment
588 entry per call to EVAL. (reported by Kevin Reid)
589 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
590 * optimization: type testing for non-vector arrays should be faster.
591 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
593 * bug fix: setting 31st element of a bit vector to zero did not work
595 * bug fix: redefining a class definition which failed due to a
596 previous accessor / function clash now works (but see BUGS entry
597 #380 for more problems in this area). (thanks to Zach Beane)
598 * the long form of DEFINE-METHOD-COMBINATION disables method group
599 checking when given a single method group with pattern *. (thanks
601 * on x86 compiler supports stack allocation of results of simple
602 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
603 * support for building and running on Mac OS X 10.4 "Tiger" has been added
604 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
605 * fixed some bugs related to Unicode integration:
606 ** the restarts for recovering from input and output encoding
607 errors only appear when there is in fact such an error to
609 * fixed some bugs revealed by Paul Dietz' test suite:
610 ** MISC.549 and similar: late transformation of unsafe type
611 assertions into derived types caused unexpected code
613 ** SCALE-FLOAT type deriver is less wrong.
614 ** type derivers for EXP, LOG and similar functions compute result
615 types for complex arguments better.
616 ** (MISC.563) CONJUGATE type deriver works for very restricted
618 ** out-of-line type testers for character strings are available.
619 ** EQUAL compiler transform understands specialness of objects
621 ** accessing double-floats stored on the stack now works on x86-64.
622 ** debugger internals could sometimes create invalid lispobjs,
623 resulting in GC crashes.
624 ** MISC.548: type check weakening can convert required type into
626 ** initialization forms for bindings are not in scope of free special
629 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
630 * incompatible change: thread support for non-NPTL systems has
631 been removed - locking is buggy and unreliable. A threaded
632 SBCL build will now warn at startup and refuse to create
633 new threads, unless futex support is detected in the kernel
634 * incompatible change: the top level REPL now has only an ABORT
635 restart associated with it, not TOPLEVEL and ABORT as it used to.
636 TOP and TOPLEVEL are now available as debugger commands for
637 returning to the top level.
638 * incompatible change: forms evaluated in the REPL now use the
639 global optimization policy.
640 * incompatible change: user- and system-initialization files are
641 no longer processed with LOAD, but by READ and EVAL; hence the
642 global optimization policy, startup package, readtable, etc,
644 * The .fasl file format number has been incremented because of
645 various incompatible changes.
646 * internal entry point details and argument counts no longer appear
647 in backtraces unless explicitly requested by setting
648 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
649 * built-in and standard functions no longer have names like "top
650 level local call to FOO".
651 * fixed bug 32: functions defined in non-null lexical environments
652 now have more legible printed representation
653 * fixed bug 33: functions defined in non-null lexical environemnts
654 are now more amenable to inspection by INSPECT.
655 * workaround for bug 354: XEPs no longer appear in backtraces unless
656 explicitly requested.
657 * fixed bug: receiving the signal which results from attempting to
658 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
659 SIGBUS on other *BSDs) on architectures where the C stack is also
660 the Lisp stack (x86 and x86-64) from foreign code no longer leads
661 to debugger confusion or wild instability. (reported by Cheuksan
663 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
664 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
665 (reported by Lutz Euler)
666 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
667 compiled code. (reported by Paul Dietz)
668 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
669 specialized parameter to a method to something that is not TYPEP
670 the specializer is now possible.
671 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
672 face of package deletion.
673 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
674 pathname. (reported by David Tolpin on c.l.l)
675 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
676 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
677 STRING-STREAM and FILE-STREAM, respectively.
678 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
679 than previously; in addition, its test suite is now run on build.
680 (thanks to Luke Gorrie)
681 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
682 * optimization: added a immediate representation for single-floats
684 * fixed some bugs revealed by Paul Dietz' test suite:
685 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
686 ** attempting to create a package with a colliding nickname causes
687 correctable errors to be signalled.
688 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
689 ** several x86-64 backend bugs related to sign-extension of immediate
692 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
693 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
694 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
696 * fixed loading of multiply forward-referenced layouts.
697 (thanks to Cheuksan Wang)
698 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
699 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
700 (see COMPILE-FILE documentation for details of :PRINT option)
701 * fixed bugs 19 and 317: fixed-format floating point printing is
702 more accurate. This also fixes a bug reported by Adam Warner
703 related to the ~@F format directive.
704 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
706 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
707 dispatch table if the from-char is a dispatch macro character.
708 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
709 by the wordsize no longer ignore the last word. (reported by Lutz
711 * fixed bug in type checking of dynamic-extent variables. (reported
713 * optimization: sequence traversal functions use their freedom to
714 coerce function designators to functions.
715 * optimization: code with many calls to CLOS methods specialized on
716 CLOS classes has had redundant type checks removed.
717 * optimization: added declarations to speed up operations that access
718 the internal character database (for example STRING-UPCASE)
719 * optimization: comparison operations between floats and sufficiently small
720 fixnums no longer create extra rationals
721 * fixed some bugs related to Unicode integration:
722 ** portions of multibyte characters at the end of buffers for
723 character-based file input are correctly transferred to the
724 start of the buffer at the next read.
725 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
726 passing it through to OPEN.
727 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
728 argument given to internal calls to OPEN.
729 ** the built-in comment readers (introduced by character sequences
730 ";" and "#|") are more forgiving to encoding errors; they will
731 STYLE-WARN and then attempt to resync the stream at a character
732 boundary. (thanks to Teemu Kalvas)
733 * fixed some bugs in the x86-64 port:
734 ** Negative short int return values from c-calls are sign-extended
736 ** The stack is aligned to 16-bytes for c-calls, as required by
737 the ABI. (thanks to Cheuksan Wang)
738 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
739 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
740 for enums. (thanks to Vincent Arkesteijn)
741 ** Multiple small optimizations and bugfixes for floating point
743 * fixed some bugs revealed by Paul Dietz' test suite:
744 ** Space, Tab, Linefeed, Return and Page have the invalid
745 secondary constituent character trait.
746 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
748 ** WITH-INPUT-FROM-STRING should only update the index place on
750 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
752 ** Bit-array operations (BIT-AND and similar) worked incorrectly
753 with one-dimensional arrays with fill pointers.
754 ** TYPE-OF failed on a complex with an integer realpart and a
756 ** compiler failure during type inference for the code of form
757 (IF (EQL X (THE ...)) ...) (MISC.535).
759 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
760 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
761 platform. The port supports 61-bit fixnums, large memory spaces
762 and reloading shared object files.
763 * enhancement: saving cores with foreign code loaded is now
764 supported on ppc/Darwin in addition to the previously supported
766 * enhancement: the statistical profiler now walks deeper into the
767 call stack for more meaningful call-graphs and accrued time
768 reports (x86/x86-64 only). It also now reports time spent in
770 * enhancement: it is now possible to trace most individual methods
771 of a generic function in addition to tracing the generic function
773 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
775 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
776 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
777 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
778 produces an error. (thanks to Vincent Arkesteijn)
779 * bug fix: NAMESTRING on pathnames with :WILD components in their
780 directories works correctly. (thanks to Artem V. Andreev)
781 * fixed bug 125: compiler preserves identity of closures. (reported
783 * bug fixed: functions with &REST arguments sometimes failed with
784 "Undefined function" when compiled with (DEBUG 3). (reported by
786 * bug fix: overflow during compiling of setting element of a bit
787 vector with constant index and value. (reported by Timmy Douglas)
788 * build fix: fixed the dependence on *LOAD-PATHNAME* and
789 *COMPILE-FILE-PATHNAME* being absolute pathnames.
790 * on x86 compiler partially supports stack allocation of dynamic-extent
792 * GO and RETURN-FROM do not check the extent of their exit points
793 when compiled with SAFETY 0.
794 * fixed some bugs related to Unicode integration:
795 ** encoding and decoding errors are now much more robustly
796 handled; it should now be possible to recover even from invalid
797 input or output to the terminal. (thanks to Teemu Kalvas)
798 ** provided a first cut at implementing STRING-TO-OCTETS and
799 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
800 ** altered the SB-MD5 contributed module slightly, changing the
801 interface just enough for it to be supportable for builds where
802 lisp characters are not eight bits.
803 * fixed some bugs revealed by Paul Dietz' test suite:
804 ** the FORMATTER-generated functions for ~V[ conditionals require
805 the correct number of arguments.
806 ** READ-FROM-STRING returns the mandated second value when applied
807 to displaced strings.
808 ** the #\Rubout and #\Backspace characters are treated as invalid
809 constituent characters by the tokenizer.
811 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
812 * new feature: reloading changed shared object files with
813 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
814 * new feature: references to foreign variables and functions
815 can now be compiled and loaded before the corresponding shared
816 object file is loaded, as long as the foreign definitions are
817 available at runtime.
818 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
819 supported. (thanks to Dan Debertin)
820 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
821 just keywords. (thanks to Vincent Arkesteijn)
822 * fixed bug #331: structure-class instances corresponding to
823 DEFSTRUCT forms are now created eagerly.
824 * fixed bug #345: backtraces from calls to undefined functions work
825 on x86 as well. Related bug #61 is now also partially fixed on x86
826 (backtraces from throws to unknown catch tags.)
827 * bug fix: lambda-list parsing is now stricter vrt. order and number
828 of lambda-list keywords.
829 * bug fix: as specified by AMOP, an error is signalled if a
830 class-option appears multiple times in a DEFCLASS form. (reported
832 * bug fix: RANDOM can be compiled when the compiler derives the type
833 of its numeric argument as a disjoint set of small integers.
834 (reported by Paul Dietz)
835 * bug fix: starting a core saved with shared objects loaded when
836 those objects are not available no longer causes threaded SBCL to
837 hang. (reported by Sean Ross)
838 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
840 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
841 argument, which must always be NIL. (reported by Kalle Niemitalo)
842 * bug fix: printing 1.0d+23 no longer results in an error.
843 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
844 * bug fix: structure slot setters preserve evaluation order. (thanks
846 * fixed some bugs related to Unicode integration:
847 ** RUN-PROGRAM can allow its child to take input from a Lisp
848 stream. (reported by Stefan Scholl)
849 ** ASDF-INSTALL successfully downloads .tar.gz files in all
850 locales. (reported by Ken Causey)
851 * fixed some bugs revealed by Paul Dietz' test suite:
852 ** INCF, DECF and REMF evaluate their place form as specified in
854 ** FORMATTER expands ~{ ~} iteration directives with V or #
855 parameters correctly.
856 ** FORMATTER deals with the ~@[ ~] conditional directive where the
857 consequent uses no arguments correctly.
858 ** the system has a partial understanding of the (COMPLEX RATIO)
860 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
863 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
864 * new feature: a build-time option (controlled by the :SB-UNICODE
865 keyword feature, enabled by default) for building the system with
866 support for the entire 21-bit character space defined by the
868 * new feature: the system now has rudimentary external-format
869 support; the primary user-visible change at this time is that
870 characters with the high bit set (such as lower-case-e-acute) will
871 print correctly to a terminal in a UTF-8 environment.
872 * minor incompatible change: BASE-CHAR no longer names a class;
873 however, CHARACTER continues to do so, as required by ANSI.
874 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
875 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
876 has been moved to the SB-EXT package (temporarily re-exported from
878 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
879 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
881 * fixed bug #351: better error-handlind and reporting for malformed
883 * fixed bug #350: bignum-printing is now more memory-efficient,
884 allowing printing of very large bignums, eg. (expt 2 10000000).
885 (reported by Bruno Haible)
886 * fixed bug #302: better primitive-type selection for intersection
888 * fixed bug #308: non-graphic characters now all have names, as
889 required. (reported by Bruno Haible)
890 * bug fix: redefining a class with different superclasses now correctly
891 removes it from the direct-subclasses of its previous superclasses.
892 (reported by David Morse)
893 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
894 new value now works. (reported by Bruno Haible)
895 * bug fix: correct canonicalization of multiple non-standard slot
896 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
897 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
898 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
899 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
900 now exists, an signals an error.
901 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
902 during macroexpansion for non-iteration variables as well. (reported
903 by Bruno Haible for CMUCL)
904 * bug fix: Cyclic structures and unprintable objects in compiler
905 messages no longer cause errors. (reported by Bruno Haible)
906 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
907 and READ-FROM-STRING all now return a primary value of NIL if
908 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
909 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
910 (reported by Bruno Haible for CMUCL)
911 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
912 do not designate a condition. (reported by Bruno Haible for
914 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
915 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
916 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
917 permitting resolution in favour of any of the conflicting symbols.
918 (reported by Bruno Haible for CMUCL)
919 * bug fix: EQUAL compiler optimizations is less aggressive on
920 strings which can potentially compare true despite having distinct
921 specialized array element types.
922 * bug fix: unit enumerations can be defined without dividing by
923 zero. (thanks to Vincent Arkesteijn)
924 * FORMAT compile-time argument count checking has been enhanced.
925 (report from Bruno Haible for CMUCL)
926 * a partial workaround for the bug 262: the compiler does not try to
927 inline-expand a local function doing RETURN-FROM from a deleted
928 BLOCK. (thanks to Peter Denno for the bug report and to David
929 Wragg for the simple test case)
930 * fixed some bugs revealed by Paul Dietz' test suite:
931 ** RENAME-PACKAGE allows all package designators as new package
933 ** constraint propagation and lambda variable substitution are
934 more cautious in dealing with partially deleted code.
935 ** compiler sometimes lost reoptimization passes.
936 ** CERROR, given a condition as condition designator, passes its
937 remaining arguments to the continue format control without
939 ** Case-altering FORMAT directives work correctly on non-ASCII
941 ** The REAL type specifier handles bounds outside the floating
942 point ranges without signalling FLOATING-POINT-OVERFLOW.
943 ** Functions with IR1-transformations can create intercomponent
944 references to global functions.
945 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
947 ** FORMAT ~R treats a nil value for its first parameter correctly.
949 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
950 * enhancement: saving cores with foreign code loaded is now
951 supported on x86/NetBSD and sparc/Linux in addition to the previously
953 * bug fix: on some platforms repeated installations caused multiple
954 copies of HTML documentation to be installed -- should not happen
955 any more. (reported by Stefan Scholl)
956 * bug fix: parsing self-recursive alien record types multiple times
957 no longer causes infinite recursion. (reported by Thomas F. Burdick,
958 original patch by Helmut Eller for CMUCL)
959 * bug fix: stack-exhaustion detection works now on NetBSD as well.
960 (thanks to Richard Kreuter)
961 * bug fix: defining classes whose accessors are methods on existing
962 generic functions in other (locked) packages no longer signals
963 bogus package lock violations. (reported by François-René Rideau)
964 * bug fix: special variables as DEFMETHOD parameters no longer have
965 associated bogus type declarations. (reported by David Wragg and
967 * bug fix: read-write consistency on streams of element-type
968 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
969 * bug fix: redefiniton of the only method of a generic function with
970 no DEFGENERIC no longer emits a full WARNING. In addition,
971 redefinition of generic functions with no DEFGENERIC to an
972 incompatible lambda list now signals an error. (thanks to Zach
974 * bug fix: DEFGENERIC now works even when there's a function of the
975 same name in an enclosing lexical environment. (thanks to Zach
977 * fixed compiler failure, caused by instrumenting code during
978 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
979 * optimization: added loop analysis and improved register allocation
980 * fixed some bugs revealed by Paul Dietz' test suite:
981 ** POSITION on displaced vectors with non-zero displacement
982 returns the right answer.
983 ** (SIMPLE-STRING) is a valid type specifier for sequence
985 ** *PRINT-LEVEL* handling for slotless structures is pedantically
987 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
988 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
990 ** The justification version of the FORMAT ~< directive treats
991 non-zero minpad parameter correctly.
993 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
994 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
995 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
996 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
997 the supported interface.
998 * new feature: Single-stepping of code compiled with DEBUG 2 or
999 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1000 * new feature: saving cores with foreign code loaded is now
1001 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1002 Timothy Moore's work for CMUCL)
1003 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1004 parameters to * instead of NIL if no initform is supplied.
1005 (reported by Johan Bockgaard)
1006 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1007 correctly when FUNCALL appears as the car of the form. Note:
1008 despite this FUNCALL forms are not currently subject to
1009 compiler-macro expansion. (port of Raymond Toy's fix for the
1010 same from CMUCL, reported by Johan Bockgaard)
1011 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1012 (thanks for Teemu Kalvas)
1013 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1014 case instead of terminating the iteration (thanks for Julian
1015 Squires, Sean Champ and Raymond Toy)
1016 * bug fix: incorrect expansion of defgeneric that caused a style
1017 warning. (thanks for Zach Beane)
1018 * bug fix: slot accessor effective method computation works properly
1019 for classes with multiple non-standard applicable methods on
1020 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1021 * on x86 compiler supports stack allocation of results of LIST and
1022 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1023 CMUCL implementation by Gerd Moellmann)
1024 * fixed some bugs revealed by Paul Dietz' test suite:
1025 ** FORMAT strings with both the ~<~:;~> form of the justification
1026 directive and pretty-printing directives cause an error.
1028 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1029 * incompatible change: the internal functions
1030 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1031 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1032 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1033 instead of the old functions.
1034 * new feature: on platforms where "dladdr" is available foreign
1035 function names now appear in backtraces. (based on Helmut Eller's
1037 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1038 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1040 * documentation: ASDF manual is now distributed with SBCL.
1041 * bug fix: SBCL can now load its contributed modules with REQUIRE
1042 even if the system-provided entries have been removed from
1043 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1045 * bug fix: backtraces involving undefined functions or assembly
1046 routines are more informative. (thanks to Brian Downing)
1047 * bug fix: mutually referent alien structures now work correctly.
1048 (reported by Rick Taube)
1049 * bug fix: structures defined by WITH-ALIEN can be referred to
1050 within other definitions in the same WITH-ALIEN.
1051 * bug fix: division operators (MOD, TRUNCATE and the like) with
1052 constant zero divisors and integer dividends no longer generate
1054 * bug fix: provide default methods for INPUT-STREAM-P and
1055 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1056 * bug fix: improve the detection and resolution of MOP metacycles
1057 (where effective method computation depends on the generic function
1058 having its effective method computed). (reported by Bruno Haible)
1059 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1060 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1061 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1062 * bug fix: pathnames with relative directory components can be
1063 represented relative to default pathnames.
1064 * optimization: in taking the GCD of bignums, reduce the two bignums
1065 to approximately the same size (using Euclid's algorithm) before
1066 applying the more sophisticated binary GCD. (thanks to Juho
1068 * optimization: COUNT on bitvectors now operates word-at-a-time.
1069 * optimization: ASH with a positive, but not necessarily constant,
1070 (leftwards) shift, when in a modular context, is compiled to a
1072 * fixed some bugs revealed by Paul Dietz' test suite:
1073 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1074 if the corresponding argument is NIL.
1075 ** FORMAT directives accepting a minpad parameter treat negative
1077 ** literal commas as character parameters to FORMAT directives are
1079 ** literal spaces directly after ~<Newline> directives within a
1080 format-logical-block (~:< ~@:>) do not induce :FILL-style
1081 conditional newlines.
1082 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1083 as specified: it no longer includes conditional newlines).
1084 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1086 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1087 :PER-LINE-PREFIX argument does not evaluate to a string.
1089 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1090 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1091 the manual for details; for now, package locks can be disabled by
1092 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1093 if no major problems are found then it is likely that they will be
1094 compiled in unconditionally.
1095 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1096 unsupported operators on all platforms. To load a shared library
1097 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1098 object file, link it into a shared library outside of SBCL and
1099 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1101 * bug workaround: ROOM T and the GC were not working together
1102 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1103 reporting facility aren't preserved. That reporting has been
1104 disabled, so now until and unless someone figures out how to make it
1105 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1106 * minor incompatible change: as threatened around sbcl-0.8.0, the
1107 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1108 The symbols are also exported from SB-PCL for backwards
1109 compatibility, but more so than before SB-PCL should be treated as
1110 an implementation-internal package.
1111 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1113 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1114 It is known as of this release not to work on the Alpha, however.
1115 * fixed bug #167: errors signalled due to illegal syntax in method
1116 bodies are now more legible.
1117 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1118 designators and can hence be used with TYPEP.
1119 * fixed bug #333: CHECK-TYPE now ensures that the type error
1120 signalled, if any, has the right object to be accessed by
1121 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1122 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1123 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1125 * fixed bug #334: programmatic addition of slots using specialized
1126 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1127 and :ALLOCATION :CLASS slots.
1128 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1129 float exponents. (rereported by Peter Seibel)
1130 * fixed a bug: #\Space (and other whitespace characters) are no
1131 longer considered to be macro characters in standard syntax by
1132 GET-MACRO-CHARACTER.
1133 * fixed bug: initialization of condition class metaobjects no longer
1134 causes an instance of the condition to be created. (reported by
1136 * fixed bug: it is now possible to have slots such that
1137 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1138 description is neither :INSTANCE nor :CLASS.
1139 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1140 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1141 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1142 non-standard SLOT-DEFINITION-ALLOCATION values.
1143 * fixed bugs: various race conditions handling exiting threads.
1144 CL-PPCRE's thread torture test now passes.
1145 * fixed arguably-a-bug: GC time proportional to number of threads in
1146 system even when most of them are idle
1147 * optimization: improved performance of BIT and SBIT on bit-vectors.
1148 * .fasl file incompatibility: The fasl file version number has
1149 been incremented because of changes associated with package locks.
1151 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1152 * minor incompatible change: the system no longer provides
1153 optimization hints (conditions which are TYPEP
1154 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1155 mode; these hints are still emitted when higher SPEED optimization
1157 * new contrib module: a sampling profiler (profiling by statistical
1158 sampling, rather than by instrumenting functions) is available as
1159 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1160 * the behaviour of the standard function ED is now customizeable by
1161 third parties through a hook variable: see ED's documentation
1162 string for information on the protocol.
1163 * the compiler no longer emits efficiency notes for (FUNCALL X)
1164 when the type of X is uncertain under default optimization
1166 * fixed bug 276: mutating a binding of a specialized parameter to a
1167 method to something that is not TYPEP the specializer is now
1169 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1170 exhibit the required behaviour on the x86 platform. (thanks to
1171 Peter van Eynde, Eric Marsden and Bruno Haible)
1172 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1173 even for difficult arguments. (reported by Peter Graves)
1174 * fixed bug 141a: the backquote printer now descends quoted
1176 * fixed another bug in backquote printing: no more destructive
1177 modification of the form's list structure. (reported by Brian
1179 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1180 move between its address being taken and the call to
1181 interrupt_thread, fixing a crashing race condition.
1182 * the SB-POSIX contrib implementation has been adjusted so that it
1183 no longer exhibits ridiculously poor performance when constructing
1184 instances corresponding to C structs.
1186 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1187 * minor incompatible change: the sb-grovel contrib now treats C
1188 structures as alien (in the sense of SB-ALIEN) objects rather than
1189 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1190 has implications for memory management of client code
1191 (sb-grovel-returned objects must now be manually managed) and for
1192 type safety (alien objects now have full types).
1193 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1194 used to control emission of compiler diagnostics, rather than the
1195 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1196 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1197 quality should be considered deprecated.
1198 * install.sh now installs the user manual as well
1199 * (not quite a new documentable feature, but worth considering in
1200 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1201 beginnings of a semantically meaningful condition hierarchy is
1202 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1204 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1205 designator as the defaults argument.
1206 * fixed bug: Displaced arrays whose displaced-to array has become
1207 too small now cause ARRAY-DIMENSION to signal an error, providing
1208 for safer bounds-checking. (reported by Bruno Haible)
1209 * fixed bug: DEFCLASS slot definitions with identical :READER and
1210 :WRITER names now signal a reasonable error. (reported by Thomas
1212 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1214 * fixed bug: Invalid slot specification errors now print correctly.
1215 (thanks to Zach Beane)
1216 * fixed bug 320: Shared to local slot value transfers in class
1217 redefinitions now happen corrently from superclasses as
1218 well. (reported by Bruno Haible)
1219 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1221 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1222 improper lists correctly. (reported by Bruno Haible)
1223 * fixed bug 313: source-transform for <fun-name> was erroneously
1224 applied to a call of a value of a variable with name <fun-name>.
1225 (reported by Antonio Menezes Leitao)
1226 * fixed bug 307: The obsolete instance protocol ensures that
1227 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1228 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1229 remove dead unknown-values globs from the middle of the stack.
1230 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1231 easier to resume long computations after using *BREAK-ON-SIGNALS*
1232 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1233 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1234 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1235 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1236 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1237 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1239 * fixed a (fairly theoretical) bug in string printing: if
1240 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1241 does not have array-element-type equal to the most general string
1243 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1244 function name. (thanks to Nikodemus Siivola)
1245 * fixed bug: compile-time format string checker failed on
1246 non-closed ~{. (reported by Thomas F Burdick)
1247 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1248 don't warn on legal loop code involving more than one aggregate
1249 boolean. (thanks to Nikodemus Siivola)
1250 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1251 in keyword position to MAKE-PACKAGE was not being reported as
1253 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1254 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1256 * fixed bugs 280 and 312: the checking for multiple definitions in a
1257 file is less likely to become confused by uses of inline
1259 * fixed bug: the #S reader macro performs the keyword coercion
1260 specified for slot names. (reported by Kalle Niemitalo)
1261 * fixed bug: lambda lists may contain symbols whose names start with
1262 & but are not lambda-list-keywords; their occurrence triggers a
1264 * fixed bug 321: define-method-combination argument lambda lists do
1265 not cause invalid code to be generated when &optional or &aux
1266 variables have default values. (reported by Bruno Haible)
1267 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1268 have CLOS classes; this appears also to have fixed reported
1269 instability in stack exhaustion detection.
1270 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1271 MUFFLE-WARNING when no associated restart is present is now
1273 * optimization: rearranged the expansion of various defining macros
1274 so that each expands into only one top-level form in a
1275 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1277 * optimization: used a previously little-used slot in symbols to
1278 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1280 * fixed some bugs revealed by Paul Dietz' test suite:
1281 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1283 ** Fixed an optimization bug related to inheritance of initforms
1284 from local to shared slots.
1285 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1286 ** CAST optimizer forgot to flush argument derived type.
1287 ** print/read consistency on floats is now orders of magnitude
1288 more likely. (thanks also to Bruno Haible for a similar report
1290 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1291 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1292 for objects of type REAL. Make it so.
1293 ** SXHASH is specified (infelicitously) to respect similarity,
1294 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1295 it so. (thanks to Markus Ziegler)
1296 ** on the Alpha, the compiler succeeds in compiling functions
1297 returning a known number of arguments greater than 63.
1298 ** fixed handling of invalid NIL arguments in keyword position
1300 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1302 ** provided a readably-printable representation for RANDOM-STATE
1304 ** ensured that pathnames, where they have a namestring, always
1305 print using #P"..." syntax.
1307 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1308 * Support for the forthcoming 2.0 version of the NetBSD kernel
1309 running on x86 hardware has been added. (thanks to Perry
1310 E. Metzger most immediately, and others for their past work)
1311 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1312 current release version; SBCL's OpenBSD support had been broken
1313 since about the time of OpenBSD's switch to ELF binary format.)
1314 * [placeholder for DX summary]
1315 ** user code with &REST lists declared dynamic-extent, under high
1316 speed or space and low safety and debug optimization policy.
1317 * The manual has been converted to Texinfo format and the debugger
1318 chapter from the cmucl manual has been added.
1319 * A facility has been added to extract documentation strings from
1320 sbcl and store them as Texinfo-formatted snippets for inclusion in
1321 the manual (via Texinfo's @include directive)
1322 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1323 the test case to Dave Roberts)
1324 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1325 pointers. (reported by Sean Ross)
1326 * bug fix: PROFILE output is printed nicely even for large numerical
1327 values. (thanks to Zach Beane)
1328 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1329 greater than 32 handle EOF correctly.
1330 * bug fix: on X86 an immediate argument of the IMUL instruction is
1331 correctly printed in disassembly. (thanks to Lutz Euler)
1332 * bug fix: class slots in redefined classes preserve their old
1333 values. (thanks to Bruno Haible and Nikodemus Siivola)
1334 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1335 succeeds. (reported by Marco Baringer)
1336 * fixed some bugs revealed by Paul Dietz' test suite:
1337 ** READ-SEQUENCE now works on ECHO-STREAMs.
1338 ** RATIONALIZE works more according to its specification. (thanks
1341 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1342 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1343 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1344 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1345 should matter to you only if you rebind the printer control
1346 variables and then find you want different bindings in the
1347 debugger than in the ordinary execution of your program.)
1348 * The runtime build system has been tweaked to support building
1349 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1350 assembler and linker. (thanks to Nikodemus Siivola)
1351 * Unbound, undefined, undeclared variables now trigger full
1352 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1353 more often programmer error than deliberate exploitation of undefined
1355 * optimization: the hash algorithm for strings has changed to one
1356 that is less vulnerable to spurious collisions. (thanks to Juho
1358 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1359 needless bounds checking. (thanks to Juho Snellman)
1360 * optimization: implemented multiplication as a modular
1361 (UNSIGNED-BYTE 32) operation on the PPC backend.
1362 * fixed some bugs revealed by Paul Dietz' test suite:
1363 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1365 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1366 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1367 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1369 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1371 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1373 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1374 their output stream on EOF from read.
1375 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1376 have been read to end-of-file.
1377 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1379 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1380 description of determination of which consecutive characters
1382 ** Printing the "Space" character with escaping on now yields "#\\ ",
1383 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1384 ** Reading floating-point numbers with *READ-BASE* set to a number
1385 less than 10 works correctly.
1386 ** Reading floating-point numbers with *READ-BASE* set to a number
1387 more than 10 works correctly.
1388 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1389 the readtable currently in effect.
1391 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1392 * minor incompatible change: parsing of namestrings on a physical
1393 (Unix) host has changed; numbers after the final #\. in a
1394 namestring are no longer interpreted as a version field. This is
1395 intented to be largely invisible to the user, except that the
1396 meaning of the namestring "*.*.*" has changed: it now refers to a
1397 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1398 should usually be replaced by
1399 (make-pathname :name :wild :type :wild :version :wild)
1400 with the added benefit that this is more likely to be portable.
1401 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1402 option to OPEN now signals an error if the file being opened
1403 exists; this may have an impact on existing code.
1404 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1405 signals received, on the PowerPC platforms (both Linux and
1406 Darwin). (thanks to Pierre Mai for pointing out the location of
1408 * several fixes on OS X: The system now builds and runs cleanly on
1409 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1410 (thanks to Brian Mastenbrook)
1411 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1412 second-resolution: integer multiples of 1/3600 between -24 and 24.
1413 (thanks to Vincent Arkesteijn)
1414 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1415 type when called with 1 argument; PEEK-CHAR checked type of
1416 PEEK-TYPE only after having read first character from a
1417 stream. (reported by Peter Graves)
1418 * bug fix: the garbage collector now has much better locality
1419 behaviour, and in particular no longer treats all memory as being
1420 exhausted when many small objects point to each other in a deeply
1422 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1423 recognized as being TYPEP their class.
1424 * bug fix: the PUSHNEW documentation string has been corrected.
1425 (thanks to Vincent Arkesteijn)
1426 * bug fix: defaulting of the value for the last of an atypically
1427 large number of multiple values being bound was not being
1428 performed correctly on the Alpha or PPC platforms
1429 * optimization: implemented multiplication as a modular
1430 (UNSIGNED-BYTE 32) operation on the x86 backend.
1431 * optimization: SEARCH on simple-base-strings can now be open-coded.
1432 (see also contrib/compiler-extras.lisp for inspiration for
1433 teaching the compiler about the Boyer-Moore algorithm).
1434 * value, returned by MAX (and MIN) called with several EQUALP, but
1435 not EQL, arguments now does not depend on compiler settings.
1436 * fixed some bugs revealed by Paul Dietz' test suite:
1437 ** in stack analysis liveness information is propagated from
1438 non-local entry points.
1439 ** pathwise CAST removing failed when the CAST node did not start
1441 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1442 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1444 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1445 host is already defined.
1446 ** RENAME-FILE works on streams instead of signalling an internal
1448 ** PEEK-CHAR uses the current readtable when determining whether
1449 or not a character is whitespace.
1450 ** MERGE-PATHNAMES handles the case when the pathname does not
1451 specify a name while the default-pathname specifies a version.
1452 ** Pathnames now stand a better chance of respecting print/read
1454 ** Attempting to use standardized file system operators with a
1455 pathname with invalid :DIRECTORY components signals a
1457 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1459 ** DIRECTORY on logical pathnames is more correct.
1460 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1461 signal a TYPE-ERROR if their argument is not a stream.
1462 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1463 designator argument does not designate a stream.
1464 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1465 examining the synonym.
1466 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1468 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1469 element-type larger than ([UN]SIGNED-BYTE 32).
1471 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1472 * When built with the :SB-FUTEX feature, threaded builds now take
1473 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1474 for faster/more reliable mutex and condition variable support.
1475 * Incompatible change (but one you probably shouldn't have been using
1476 anyway): the interface and code for arbitrating between multiple
1477 threads in the same user session has been redesigned.
1478 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1479 error when called without an explicit environment argument.
1480 (thanks to Nikodemus Siivola)
1481 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1482 confused about their position. (thanks to Adam Warner and Gerd
1484 * bug fix: Pretty printing backquoted forms with unquotations in the
1485 argument list position of various code constructs such as LAMBDA
1486 now works correctly. (reported by Paul Dietz)
1487 * bug fix: Pretty printing unquotations no longer loses all
1488 stream position information.
1489 * optimization: performance of string output streams is now less
1490 poor for multiple small sequence writes.
1491 * optimization: performance of CSUBTYPEP in the presence of complex
1492 expressions involving CONS and NOT many times has been improved.
1493 (reported by Paul Dietz)
1494 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1496 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1497 streams for the REPL, for the debugger, and so on. (thanks to
1499 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1500 * fixed some bugs revealed by Paul Dietz' test suite:
1501 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1502 is ignored if it is too small, rather than propagating through
1503 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1504 ** extremely complex negations of CONS types were not being
1505 sufficiently canonicalized, leading to inconsistencies in
1507 ** VALUES tranformer lost derived type.
1509 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1510 * fixed a bootstrapping bug: the build process no longer assumes
1511 that the various BOOLE-related constants have the same value in
1512 host and target lisps. (noted by Paul Dietz' test suite on an
1513 SBCL binary built from CLISP)
1514 * The system can now be dynamically linked on the MIPS platform,
1515 which enables dynamic loading of foreign code from Lisp. (thanks
1516 to Ralf Baechle for discussions on the MIPS ABI)
1517 * The system now records debugging information for its own source
1518 files in a filesystem-position-independent manner, relative to
1519 the "SYS" logical host.
1520 * fixed a compiler bug: MV-LET convertion did not check references
1521 to the "max args" entry point. (reported by Brian Downing)
1522 * tweaked disassembly notes to be less confident about proclaiming
1523 some instruction as an LRA. (thanks to Brian Downing)
1524 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1525 now each have their own history, command character, and other
1526 characteristics. (thanks to David Lichteblau)
1527 * fixed some bugs revealed by Paul Dietz' test suite:
1528 ** compiler failure in compiling LOGAND expressions including a
1530 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1531 shift greater than 32.
1532 ** FLUSH-DEST did not mark blocks for type check regeneration.
1533 ** HANDLER-CASE failed to accept declarations in handler clauses
1534 in some circumstances.
1536 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1537 * New code in contrib/sb-introspect (still probably not entirely
1538 stable yet) provides some support for smart Lisp development
1539 environments like SLIME.
1540 * The conditions signalled for errors occurring when loading .fasl
1541 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1542 in a way which should help ASDF recover gracefully.
1543 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1544 stuff has been cleaned up. If you code contrib/ stuff, this might
1545 affect you, and you can look at contrib/README, contrib/STANDARDS,
1546 and/or the 0.8.4.27 diff to check.
1547 * In full calls the compiler now does not generate checks for declared
1548 argument types for all arguments.
1549 * various threading fixes
1550 ** and some experimental patches which didn't make it into
1551 the main tree for this release, but which are shipped in
1552 contrib/experimental-thread.patch as a possible fix for some
1553 failures (deadlock, spinning...) in GC-intensive multithreaded
1555 * fixed PPC build problem (source code incompatibility of different
1556 library versions): added offsetof() hackery which attempts to divine
1557 where glibc maintainers put uc_mcontext today
1558 * fixed bug 282: compiler does not trust type assertions while passing
1559 arguments to a full call.
1560 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1561 &OPTIONAL VALUES type parameter.
1562 * fix bug 214: algorithm for noting rejected templates is now more
1563 similar to that of template seletion. (also reported by rydis on
1565 * fixed bug 141b: printing backquoted information readably and prettily
1566 inserts a space where necessary.
1567 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1568 (CHARACTER 10) are now reported as errors, rather than propagated
1569 as unknown types. (reported by piso on #lisp)
1570 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1571 with values NIL and :ERROR. (thanks to Milan Zamazal)
1572 * fixed bug 191c: CLOS now does proper keyword argument checking as
1573 described in CLHS 7.6.5 and 7.6.5.1.
1574 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1575 longer raise an error; further, using a list as a for-as-arithmetic
1576 counter now raises a meaningful error.
1577 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1578 now understood by sequence creation functions such as MAKE-SEQUENCE
1580 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1581 a STRING-INPUT-STREAM.
1582 * compiler enhancement: SIGNUM is now better able to derive the type
1584 * type declarations inside WITH-SLOTS are checked. (reported by
1586 * fixed some bugs revealed by Paul Dietz' test suite:
1587 ** incorrect optimization of TRUNCATE for a positive first
1588 argument and negative second.
1589 ** compiler failure in let-convertion during flushing dead code.
1590 ** compiler failure while deriving type of TRUNCATE on an
1591 interval, containing 0.
1592 ** ASH of a negative bignum by a negative bignum count now returns
1594 ** intersection of CONS types now canonicalizes properly, fixing
1595 inconsistencies in SUBTYPEP.
1597 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1598 * incompatible change: The --disable-debugger command line
1599 option now clobbers the debugger at a more fundamental
1600 level, by redefining #'INVOKE-DEBUGGER instead of by
1601 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1602 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1603 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1604 enter the debugger REPL and then suffer recursive errors
1605 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1606 while under the new system, BREAK in --disable-debugger mode
1607 terminates the system just as an unhandled error would.
1608 * fixed compiler performance when processing loops with a step >1;
1609 * bug fix: DOCUMENTATION now retrieves generic function
1610 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1611 support has been systematized, and now supports the methods
1612 specified by ANSI, along with a default method and a method for
1613 slot documentation. (reported by Nathan Froyd)
1614 * bug fix: effective methods associated with a generic function are
1615 no longer cached over a change of that generic function's method
1616 combination. (reported by Andreas Fuchs)
1617 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1618 names a non-existent directory. (thanks to Andreas Fuchs)
1619 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1620 the PPC platform, lead to stack corruption; this has been fixed.
1621 (reported by Rainer Joswig)
1622 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1623 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1624 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1625 arithmetic in general, is now much more reliable on the PPC
1627 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1628 platform now returns the right answer.
1629 * optimization: restored some effective method precomputation in
1630 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1631 precomputation is now tunable.
1632 * optimization: compiler-internal data structure use has been
1633 reviewed, and changes have been made that should improve the
1634 performance of the compiler by about 20%.
1635 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1636 simple-base-strings and simple-bit-vectors is improved.
1637 * optimization: the optimization of 32-bit logical and arithmetic
1638 functions introduced in version 0.8.3 on the x86 has been
1639 implemented on the mips, ppc and sparc platforms; an
1640 implementation of the same facility, but for 64-bit arithmetic,
1641 has been added for the alpha.
1642 * microoptimization: the compiler is better able to make use of the
1643 x86 LEA instruction for multiplication by constants.
1644 * bug fix: in some situations compiler did not report usage of
1645 generic arithmetic in (SPEED 3) policy.
1646 * bug 145b fix: compiler used wrong type specifier while converting
1647 MEMBER-types to numeric.
1648 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1650 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1651 index argument now works properly on the Alpha platform.
1652 * bug fix: floating point exception treatment on the Alpha platform
1654 * bug fix: FILE-POSITION works much better on string input and
1655 output streams. (thanks to Nikodemus Siivola)
1656 * bug fix: many threading/garbage collection symptoms sorted.
1657 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1658 might be pseudo-atomic.
1659 * internal change: Stopping for GC is now done with signals not
1660 ptrace. GC is now done in whichever thread wanted it, instead of
1662 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1664 * bug fix: result form in DO is not contained in the implicit
1666 * incompatible change: ICR structure is changed; the value part of
1667 CONTINUATION is now called LVAR; corresponding functions are
1668 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1669 * added type deriver for ISQRT (thanks to Robert E. Brown).
1670 * bug fix: better support for loading from the command line when an
1671 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1673 * fixed some bugs revealed by Paul Dietz' test suite:
1674 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1675 ** ROUND and FROUND now give the right answer when given very
1676 small float arguments.
1677 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1679 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1680 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1681 ** LCM should return a non-negative integer.
1682 ** PARSE-INTEGER returned the index of a terminator instead of the
1683 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1684 ** PARSE-INTEGER returned an incorrect index being applied to a
1686 ** LCM with two arguments of 0 returns 0 rather than signalling
1688 ** unsigned addition of a 32-bit constant with the high bit set no
1689 longer causes an internal compiler error on the x86.
1690 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1691 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1692 with negative last argument.
1693 ** byte specifiers with zero size and position no longer cause
1694 an error during type derivation.
1695 ** bignum multiplication on the Alpha platform now returns the
1697 * porting: The system now builds on SuSE AMD64, although it still
1698 generates a 32-bit binary.
1699 * .fasl file incompatibility: The fasl file version number has
1700 been incremented (because of the changes to internal compiler
1701 data structures referred to above).
1703 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1704 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1705 more accurately, on the Darwin kernel running on PowerPC hardware.
1706 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1707 * Compiler code deletion notes now signal a condition of type
1708 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1709 an associated MUFFLE-WARNING restart.
1710 * The compiler now performs limited argument count validation of
1711 constant format strings in FORMAT, and where appropriate in ERROR,
1712 CERROR and WARN. (thanks to Gerd Moellmann)
1713 * New ASDF-INSTALL contrib can be used for automatic download and
1714 installation of third-party Lisp code from CCLAN or other sites
1716 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1717 which forces another thread to execute a function supplied by the
1719 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1720 now accept and act upon their :ELEMENT-TYPE keyword argument.
1721 (reported by Martin Atzmueller, Edi Weitz)
1722 * bug fix: FILE-POSITION now accepts position designators up to
1723 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1724 is the greater. (thanks to Patrik Nordebo)
1725 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1726 Kalle Olavi Niemitalo)
1727 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1728 printable. (reported by Eric Marsden)
1729 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1730 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1731 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1732 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1733 (thanks to Stig Sandoe)
1734 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1735 UNREAD-CHAR now correctly includes the unread character in the
1736 target sequence. (thanks to Gerd Moellmann)
1737 * bug fix in threaded builds: the system can now be suspended and
1738 resumed by shell job control with minimal disruption.
1739 * bug fixes in times and timezones >2038AD
1740 * better handling of "where is GNU make?" problem in build scripts
1741 (thanks to Nikodemus Siivola)
1742 * new optimization: inside a named function any reference to a
1743 function with the same name is considered to be a self-reference;
1744 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1745 optimization quality.
1746 * new optimization on x86: logical functions and + now have
1747 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1748 used when the result is truncated to 32 bits.
1749 * VALUES declaration is partially enabled.
1750 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1751 * fixed some bugs revealed by Paul Dietz' test suite:
1752 ** The system now obeys the constraint imposed by
1753 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1754 types form a lattice under type intersection.
1755 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1756 ** ASSOC now ignores NIL elements in an alist.
1757 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1758 and (1+ MOST-POSITIVE-FIXNUM) answers.
1759 ** The addition of a method with invalid qualifiers to a generic
1760 function does not cause an error to be signalled immediately;
1761 a warning is signalled, and the error is generated only on
1762 calling the generic function.
1763 * changed .fasl file version number, in part to add type codes for
1764 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1765 obscure ANSI requirements
1767 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1768 * fixed bug 148: failure to inline-expand a local function left
1769 garbage, confusing the compiler.
1770 * fixed bugs 3cd: structure slot readers perform type check if the
1771 slot can have an invalid value (i.e. it is either not initialized
1772 or can be written with a less specific slot writer).
1773 * bug fix: the compiler now traps array references to elements off
1774 the end of an array; previously, the bounds checking in some
1775 circumstances could go off-by-one.
1776 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1778 * type declarations for array element types now obey the description
1779 on the CLHS page "Declaration TYPE", as per discussions on
1780 sbcl-help around 2003-05-08. This means that a declaration
1781 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1782 declaration, all references to BAR will be asserted or assumed
1783 (with THE, so dependent on compiler policy) to involve objects of
1784 type FOO. Note that no such declaration is implied in
1785 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1786 * declared types of functions from the "Conditions"
1787 chapter. (reported by Paul Dietz)
1788 * bug fix: CERROR accepts a function as its first argument.
1789 * bug fix: NTH an NTHCDR accept a bignum as index
1790 arguments. (reported by Adam Warner)
1791 * optimization: character compare routines now optimize comparing
1792 against a constant character. (reported by Gilbert Baumann)
1793 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1794 argument now works properly on the MIPS platform.
1795 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1796 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1797 declared to have a functional type, does not cause an error
1799 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1800 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1801 * fixed bug: if last continuation of a deleted block has a
1802 destination, this destination should be deleted too. (reported by
1804 * fixed a bug in the bootstrap process: the host compiler's values
1805 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1806 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1807 test case from Patrik Nordebo)
1808 * improved the ability of the disassembler on the PPC platform to
1809 provide helpful disassembly notes.
1810 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1811 the class in more cases than previously.
1812 * bug fix: FILE-POSITION now understands :START and :END for
1813 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1814 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1815 8) data. (thanks to David Lichteblau)
1816 * bug fix: it is possible to add a method to a generic function
1817 without lambda list.
1818 * bug fix: reader failed to signal END-OF-FILE inside an
1819 object representation. (reported by Nikodemus Siivola)
1820 * fixed some bugs revealed by Paul Dietz' test suite:
1821 ** LAST and [N]BUTLAST should accept a bignum.
1822 ** condition slot accessors are methods.
1823 ** (VECTOR NIL) is a subtype of STRING.
1825 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1826 * minor incompatible change: some nonsensical specialized lambda
1827 lists (used in DEFMETHOD) which were previously ignored now signal
1829 * minor incompatible change: the system is now aware of the types of
1830 variables in the COMMON-LISP package, and will signal errors for
1831 most violations of these type constraints (where previously they
1832 were silently accepted).
1833 * minor incompatible change: COMPILE-FILE now uses the freedom
1834 afforded (ANSI 3.2.2.3) to use derived function types for
1835 functions defined in the same file. This also permits the system
1836 to warn on static type mismatches and function
1837 redefinition. (Currently it does not work with high DEBUG level.)
1838 * minor incompatible change: VALUES declaration is disabled.
1839 * When issuing notes, the compiler now signals a condition of type
1840 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1841 restart for use in user handlers. It is expected that the
1842 COMPILER-NOTE condition will eventually become a condition
1843 supertype to a hierarchy of note types, which will then be
1844 handleable in a similar fashion. However, at the moment, no such
1845 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1846 but it's an implementation detail, not a classification for the
1848 * Changes in type checking closed the following bugs:
1849 ** type checking of unused values (192b, 194d, 203);
1850 ** template selection based on unsafe type assertions (192c, 236);
1851 ** type checking in branches (194bc).
1852 * A short form of VALUES type specifier has ANSI meaning (it has
1853 increased the number of situations when SBCL cannot perform type
1855 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1856 keywords or constants is permissible.
1857 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1858 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1859 classes. (thanks to Antonio Martinez)
1860 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1861 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1862 argument to a pretty-print stream. Instead, it leaves any such
1863 operations to DESCRIBE-OBJECT methods.
1864 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1866 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1867 lambda lists. (thanks to David Lichteblau)
1868 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1869 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1870 (thanks to Gerd Moellmann)
1871 * fixed bug 246: increased compilation speed of long
1872 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1874 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1876 * DEFCONSTANT now throws a condition of type
1877 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1878 constant to a non-EQL value; CONTINUE and ABORT restarts
1879 respectively change and preserve the value.
1880 * fixed bug 63: The code walker, part of the implementation of CLOS,
1881 is now better at handling symbol macros.
1882 * bug fix: There is no longer an internal implementation type named
1883 CL:LENGTH. (reported by Raymond Toy)
1884 * bug fix: In macro-like defining macros/special operators the
1885 implicit block does not enclose the lambda list.
1886 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1887 enclosed in parentheses) are not suitable as type specifiers, and
1888 their use properly signals an error now.
1889 * bug fix: An argument count mismatch for a type specifier in code
1890 being compiled no longer causes an unhandled error at compile
1891 time, but signals a compile-time warning.
1892 * fixed simple vector readable printing
1893 * bug fix: DESCRIBE takes more care over whether the class
1894 precedence list slot of a class is bound before accessing it.
1895 (reported by Markus Krummenacker)
1896 * bug fix: FORMATTER can successfully compile pretty-printer format
1897 strings which use variants of the ~* directive inside.
1898 * bug fix: SEARCH now applies its TEST predicate to the elements of
1899 the arguments in the correct order. (thanks to Wolfhard Buss)
1900 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1901 in type inference. (thanks to Robert E. Brown)
1902 * bug fix: Reading in symbols with an explicit package name of ""
1903 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1904 * fixed some bugs revealed by Paul Dietz' test suite:
1905 ** NIL is now allowed as a structure slot name.
1906 ** Arbitrary numbers, not just REALs, are allowed in certain
1907 circumstances in LOOP for-as-arithmetic clauses.
1908 ** Multiple class redefinitions before slot access no longer
1909 causes a type error.
1910 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1911 association between the name and a class.
1912 ** Generic functions with non-standard method-combination and over
1913 five methods all of which return constants no longer return NIL
1914 after the first few invocations. (thanks to Gerd Moellmann)
1915 ** CALL-NEXT-METHOD with no arguments now passes the original
1916 values of the arguments, even in the presence of assignment.
1917 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1919 ** DISASSEMBLE works with closures and funcallable instances.
1920 ** ADD-METHOD now returns the generic function, not the new method.
1921 ** FIND-METHOD signals an error if the lengths of the specializers
1922 is incompatible with the generic function, even if the ERRORP
1924 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1925 which its argument is a member.
1926 ** DEFCLASS only redefines the class named by its class-name
1927 argument if that name is the proper name of the class;
1928 otherwise, it creates a new class.
1929 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1930 of the UNBOUND-SLOT condition to the name of the slot.
1931 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1932 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1933 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1934 SLOT-MAKUNBOUND in the specified fashion.
1936 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1937 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1938 cross-compilation host. As a consequence, we can now bootstrap our
1939 way up to SBCL starting with a bare gcc toolchain and human-readable
1940 source code (first the source to CLISP, then the source to SBCL).
1941 * A contributed module containing a partial implementation of the
1942 simple-streams interface has been included. (thanks to Rudi
1944 * A contributed module implementing the RFC1321 Message Digest
1945 Algorithm, known as MD5, has been included.
1946 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1947 no longer has any effect, as the code controlled by this feature
1948 has been deleted. (As far as we know, no-one has ever built using
1949 this feature, and its semantics were confused in any case).
1950 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1951 work on conditions (as required by the ANSI specification),
1952 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1953 expected behaviour on conditions. Users should note, however,
1954 that such behaviour is not required by the ANSI specification,
1955 and so use of this behaviour may render their code unportable.
1956 * fixed some bugs revealed by Paul Dietz' test suite:
1957 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1959 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1960 ** SXHASH on condition objects no longer returns NIL.
1961 ** :ALLOCATION :CLASS slots are better treated; their values are
1962 updated on class redefinition, and initforms inherited from
1963 superclasses are applied.
1964 ** REMOVE-METHOD returns its generic function argument even when
1965 no method was removed.
1966 ** SHARED-INITIALIZE now initializes the values of the requested
1967 slots, including those with :ALLOCATION :CLASS.
1968 ** ALLOCATE-INSTANCE now works on structure classes defined via
1969 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1971 ** SLOT-EXISTS-P now works on conditions, as well as structures
1973 ** MAKE-LOAD-FORM now has the required methods on
1974 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1975 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1976 keyword, and now implements the SLOT-NAMES argument.
1977 ** methods with &OPTIONAL arguments no longer allow too many
1978 arguments to be passed in the call without error.
1979 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1980 option is consistent with the required arguments of the generic
1981 function lambda list.
1982 * bug fix: REQUIRE accepts a string designator. (Thanks to
1984 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1985 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1986 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1987 * bug fix: The compiler now checks for duplicated variables in macro
1989 * bug fix: SETQ on globals returns the correct value.
1990 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1991 not a COMPILER-ERROR (followed by some other strange error on
1992 choosing the CONTINUE restart).
1993 * bug fix: make.sh and friends are now more consistent in the way that
1994 they look for GNU "make".
1996 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1997 * experimental native threads support (on x86 Linux >=2.4 only).
1998 This is not compiled in by default: you need to add :SB-THREAD to
1999 the target features. See the "Beyond ANSI" chapter of the manual
2001 * fix for longstanding nonANSIism: The old distinction between
2002 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2003 The return value from CL:FIND-CLASS is now a CLOS class, and
2004 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2005 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2006 * An interface to the MetaObject Protocol, as described in Kiczales,
2007 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2008 MIT Press, 1991, is available from the SB-MOP package.
2009 * incompatible change: the SB-PCL package should now be considered
2010 a private implementation detail, and no longer a semi-private MOP
2012 * minor incompatible change: due to rearrangement for threads, the
2013 control stack and binding stack are now allocated at arbitrary
2014 addresses instead of being hardcoded per-port. Users affected by
2015 this probably have to be doing advanced things with shared
2016 libraries, and will know who they are.
2017 * minor incompatible change: Previously, all --eval forms used were
2018 processed with READ before any of them were processed with EVAL.
2019 Now each --eval form is processed with both READ and EVAL before
2020 the next --eval form is processed. (Thus package operations like
2021 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2022 work as the user might reasonably expect.)
2023 * minor incompatible change: *STANDARD-INPUT* is now only an
2024 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2026 * minor incompatible change: Y-OR-N-P is now character-oriented,
2027 not line oriented. Also, YES-OR-NO-P now works without errors.
2028 (thanks to Antonio Martinez)
2029 * sb-aclrepl module improvements: an integrated inspector, added
2030 repl features, and a bug fix to :trace command.
2031 * Known functions, which cannot be open coded by the backend, are
2032 considered to be able to check types of their arguments. (fixing
2033 a bug report by Nathan J. Froyd)
2034 * fixed a bug in computing method discriminating functions: It is
2035 now possible to define methods specialized on classes which have
2036 forward-referenced superclasses. (thanks to Gerd Moellmann)
2037 * fixed evaluation order in optional entries (reported by Gilbert
2039 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2040 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2041 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2042 required-by-AMOP NAME argument, as well as CLASS and
2043 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2044 * fixed bug 20: DEFMETHOD can define methods using names that are
2045 not the proper names of classes to designate class specializers.
2046 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2047 Unix file descriptors, instead of blowing up. (thanks to Antonio
2049 * Garbage collection refactoring: user-visible change is that a
2050 call to the GC function during WITHOUT-GCING will not do garbage
2051 collection until the end of the WITHOUT-GCING. If you were doing
2052 this you were probably losing anyway.
2053 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2054 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2055 TYPEP the latter but not the former.
2056 * The compiler issues a full WARNING for calls to undefined functions
2057 with names from the CL package.
2058 * MAP-INTO for a vector destination is open coded. (reported by
2059 Brian Downing on c.l.l)
2060 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2061 documentation string.
2062 * fixed some bugs revealed by Paul Dietz' test suite:
2063 ** COPY-ALIST now signals an error if its argument is a dotted
2065 ** Condition slots are now accessed more correctly in the presence
2066 of multiple initargs for a given slot.
2067 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2068 exclude restarts of the same name associated with a different
2070 ** DEFCLASS of forward-referenced classes with another
2071 forward-referenced class in the superclasses list no longer
2073 ** Condition slots are now initialized once each, not multiple
2074 times. (thanks to Gerd Moellmann)
2075 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2077 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2078 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2079 arguments contain duplicated elements.
2080 ** RESTART-CASE understands local macros.
2081 ** RESTART-CASE associates exactly its own restarts with a condition.
2082 ** ENDP in safe mode checks its argument to be of type LIST.
2083 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2084 in question is unbound.
2085 ** Optimized MAKE-INSTANCE functions no longer cause internal
2086 assertion failures in the presence of duplicate initargs.
2087 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2088 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2089 propagated even in the ctor optimized implementation of
2091 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2093 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2094 * a better implementation of SXHASH on (simple) bit vectors,
2095 measured both in execution speed and in distribution of results
2096 over the positive fixnums, has been installed. Likewise, a better
2097 implementation of EQUAL for simple bit vectors is now available.
2098 * fixed CEILING optimization for a divisor of form 2^k.
2099 * fixed bug 240 (emitting extra style warnings "using the lexical
2100 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2101 by Antonio Martinez)
2102 * fixed SXHASH, giving different results for NIL depending on type
2103 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2104 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2105 variable. (found by Rolf Wester)
2106 * SBCL does not ignore type declarations for special
2107 variables. (reported by rif on c.l.l 2003-03-05)
2108 * some bug fixes in contrib/sb-aclrepl/
2109 * fixed some bugs revealed by Paul Dietz' test suite:
2110 ** a bug in the CONS type specifier, whereby the CAR and CDR
2111 types got intertwined, has been fixed;
2112 ** the type system is now able to reason about the interaction
2113 between INTEGER and RATIO types more completely;
2114 ** APPEND, [N]REVERSE and NRECONC check that those their
2115 arguments, which must be proper lists, are really so;
2116 ** An array specialized to be unable to hold elements has been
2117 implemented, as required -- yes, really -- by ANSI;
2118 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2119 on malformed property lists;
2121 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2122 * incompatible packaging change: in line with Unix convention,
2123 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2124 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2125 * REQUIRE and PROVIDE are now optionally capable of doing something
2126 useful. See the documentation string for REQUIRE.
2127 * infrastructure for a managed SBCL contrib system: contributed
2128 modules in this release include:
2129 ** the ASDF system definition facility;
2130 ** an interface to the BSD Sockets API;
2131 ** an ACL-like convenience interface to the repl;
2132 (thanks to Kevin Rosenberg)
2133 ** an implementation of ROTATE-BYTE, with efficient implementation
2135 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2136 gives a closer approximation to the right answer than previously.
2137 (thanks to Raymond Toy)
2138 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2139 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2140 optional environment arguments, as required by ANSI.
2141 * fixed bugs in other functions taking environment objects, allowing
2142 calls with an explicit NIL environment argument to be compiled
2144 * fixed bug 228: primary return values from
2145 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2146 COMPILE or FUNCTION.
2147 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2148 :TYPE will no longer signal errors on innocuous objects.
2149 * fixed bug 231b: SETQ is better at respecting type declarations in
2150 the lexical environment.
2151 * fixed a bug in DEFCLASS: classes named by symbols with no or
2152 unprintable packages can now be defined.
2153 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2154 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2155 * fixed compiler failure related to checking types of functions.
2156 (reported by Robert E. Brown)
2157 * the compiler is now much more consistent in its error-checking
2158 treatment of bounding index arguments to sequence functions: in
2159 (SAFETY 3) code, errors will be signalled in almost all cases if
2160 invalid sequence bounding indices are passed to functions defined
2161 by ANSI to operate on sequences.
2162 * fixed a bug in the build procedure: documentation of SBCL-specific
2163 packages is now preserved and available in the final Lisp image.
2164 * lifted FDEFINITION lookup out of loops in the implementation of
2165 many list operations. (thanks to Robert E. Brown)
2166 * fixed a bug in the reader: the #n# reader macro now works for
2167 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2168 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2169 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2170 E. Brown and Tony Martinez respectively)
2171 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2172 SLOT-BOUNDP: in particular, it is now able to optimize them much
2173 better, and is now not vulnerable to having packages renamed.
2174 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2175 always called when a slot is not present in an instance. (thanks
2177 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2178 longer signal an error on unreading a character following EOF on
2179 the previous constituent stream. (thanks to Tony Martinez)
2180 * fixed some bugs revealed by Paul Dietz' test suite:
2181 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2182 not just nonnegative fixnums;
2183 ** the logical bit-array operators such as BIT-AND now accept an
2184 explicit NIL for their "opt-arg" argument (to indicate a
2185 freshly-consed result bit-array);
2186 ** ELT now signals an error on an invalid sequence index in safe
2188 ** the type system is now cleverer about negations of numeric
2189 types, and consequently understands the BIGNUM and RATIO types
2191 ** the type system is now cleverer about the interaction between
2192 INTEGER and RATIO types: while bugs still remain, many more
2193 cases are accurately computed;
2194 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2195 if it is in the last clause;
2196 ** CONSTANTLY now correctly returns a side-effect-free function in
2198 ** DECLARE is no longer treated as a special-operator; in
2199 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2200 * incremented fasl file version number due to the change in the
2201 DEFSTRUCT-SLOT-DESCRIPTION structure.
2203 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2204 * minor incompatible change: code processed by the "interpreter" or
2205 EVAL now has a compilation optimization policy of (DEBUG 2)
2206 (changed from (DEBUG 1)) to improve debuggability of interactive
2207 development, and to allow the use of the debug RETURN command in
2209 * an experimental implementation of the RETURN command for the
2210 debugger has been included. (thanks to Frederik Kuivinen)
2211 * fixed bug 62: constraints were not propagated into a loop.
2212 * fixed bug in embedded calls of SORT (reported and investigated by
2214 * fixed some bugs revealed by Paul Dietz' test suite:
2215 ** printing and reading of arrays with some dimensions having
2216 length 0 (thanks to Gerd Moellmann);
2217 ** BOA constructor with &AUX argument without a default value does
2218 not cause a type error;
2219 ** CONSTANTP now returns true for all self-evaluating objects.
2221 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2222 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2223 accessors that are related by inheritance, as specified in the
2224 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2225 Valtteri Vuorikoski)
2226 * The compiler is now able to inline functions that were defined in
2227 a complex lexical environment (e.g. inside a MACROLET).
2228 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2230 * Support for the upcoming FreeBSD-5.0 release has been included.
2231 (thanks to Dag-Erling Smorgrav)
2232 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2233 effect when it is not in a toplevel context.
2234 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2235 stand a better chance of being correct. (thanks to Gerd
2237 * fixed bug in COERCE, which now signals an error on coercing a
2238 rational to a bounded real type which excludes the expected
2240 * The compiler is now able to derive types more accurately from the
2241 COERCE and COMPILE functions.
2242 * fixed bug 223: functional binding is considered to be constant
2243 only for symbols in the CL package.
2244 * fixed bug 231: SETQ did not check the type of a variable being set
2245 (reported by Robert E. Brown)
2246 * A new optimization for MAKE-INSTANCE has been included, fixing
2247 various bugs (including relating to :ALLOCATION :CLASS slots and
2248 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2250 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2251 ** As required by ANSI, LOOP now disallows anonymous collection
2252 clauses such as COLLECT I in conjunction with aggregate boolean
2253 clauses such as THEREIS (= I 1);
2254 ** LOOP now signals an error when any variable is reused in the
2255 same loop (including the potentially useful construct analogous
2256 to WITH A = 1 WITH A = (1+ A);
2257 ** IT is only a special loop symbol within the first clause of a
2258 conditional loop clause;
2259 ** LOOP with a typed iteration variable over a hashtable now
2260 signals a type error iff it should.
2261 * fixed some other bugs revealed by Paul Dietz' test suite:
2262 ** FILE-STREAM now names the class previously known as FD-STREAM;
2263 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2264 argument) no longer signals an error;
2265 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2266 of the slot symbol, rather than using the current package
2267 ((:CONC-NAME "") continues to intern the slot's name in the
2269 * incremented fasl file version number, because of the incompatible
2270 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2271 of the new implementation of DEFINE-COMPILER-MACRO.
2273 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2274 * Support for building SBCL for MIPS platforms running in
2275 little-endian mode has now been checked in, and basic
2276 functionality on said platforms verified.
2277 * minor incompatible change: PCL now records the pathname of a file
2278 in which methods and the like are defined, rather than its
2280 * minor incompatible change: TRUENAME now considers the truename of
2281 a file naming a directory to be the pathname with :DIRECTORY
2282 component indicating that directory.
2283 * minor incompatible change: a NAMED clause in the extended form of
2284 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2285 reason for the previous behaviour is unclear.
2286 * more systematization and improvement of CLOS and MOP conformance
2287 in PCL (thanks to Gerd Moellman and Pierre Mai):
2288 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2290 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2291 primary methods with no specializers;
2292 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2294 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2295 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2296 FORWARD-REFERENCED-CLASSes; error reporting on
2297 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2299 ** SXHASH on CLOS instances now uses a slot internal to the
2300 instance to return different numbers on distinct instances,
2301 while preserving the same return value through invocations of
2303 ** DEFMETHOD signals errors when methods with longer incongruent
2304 lambda lists are added to generic functions;
2305 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2306 CLASS, as specified in AMOP;
2307 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2308 on the order returned by the primary method for classes of
2309 class STANDARD-CLASS;
2310 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2311 * fixed some bugs shown by Paul Dietz' test suite:
2312 ** DOLIST puts its body in TAGBODY;
2313 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2315 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2316 value producing form;
2317 ** if more variables are given to PROGV than values, extra
2318 variables are bound and made to have no value;
2319 ** NSUBSTITUTE on list arguments gets the right answer with
2321 ** ELT signals an error of type TYPE-ERROR when the index argument
2322 is not a valid sequence index;
2323 ** LOOP signals (at macroexpansion time) an error of type
2324 PROGRAM-ERROR when duplicate variable names are found;
2325 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2326 ** FUNCALL of special-operators now cause an error of type
2328 ** PSETQ now works as required in the presence of side-effecting
2329 symbol-macro places;
2330 ** NCONC accepts any object as its last argument;
2331 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2333 ** loop-for-as-package does not require a package to be explicitely
2335 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2337 * fixed bug 166: compiler preserves "there is a way to go"
2338 invariant when deleting code.
2339 * fixed bug 172: macro lambda lists with required arguments after
2340 &REST arguments now cause an error to be signalled. (thanks to
2342 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2343 bug. (thanks to Matthew Danish)
2344 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2346 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2347 arguments are lexically rebound. (thanks to Gerd Moellmann and
2349 * fixed bug 194: error messages are now more informative when there
2350 is no primary method applicable in a call to a generic
2351 function. (thanks to Gerd Moellmann)
2352 * fixed bug in command line argument checking (thanks to Julian
2354 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2355 (thanks to Matthew Danish)
2356 * incremented fasl file version number, because of the
2357 SXHASH-related changes in the layout of CLOS data structures
2359 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2360 * minor incompatible change: The runtime (the Unix executable named
2361 "sbcl") is now much pickier about the .core files it will load.
2362 Essentially it now requires .core files to descend from the same
2363 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2364 as the runtime does. (The intent is to prevent the crashes which
2365 can occur, and which can even be reported as mysterious failures,
2366 when people patch the sources or change the build parameters
2367 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2368 sbcl and .core files.)
2369 * fixed bug: VALUES-LIST is no longer optimized away.
2370 * fixed bug 142: The FFI conversion of C string values to Lisp
2371 string values no longer conses excessively. (thanks to Nathan
2372 Froyd porting Raymond Toy's fix to CMU CL)
2373 * began to systematize and improve MOP conformance in PCL (thanks to
2374 Nathan Froyd, Gerd Moellman and Pierre Mai):
2375 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2377 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2378 ** FINALIZE-INHERITANCE is now called on class finalization;
2379 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2380 argument precedence order.
2381 * fixed bug 202: The compiler no longer fails on functions whose
2382 derived types contradict their declared type.
2383 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2384 so it can be non-toplevel.
2385 * The fasl file version number has changed (because of the new
2386 implementation of DEFMACRO).
2387 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2388 safe code, that any length requirement by their type-specifier
2389 argument is valid. The exceptions to this are described in bug
2391 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2392 functions now check the types of their inputs as required by ANSI.
2393 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2394 introduced symbol is DECLAREd to be SPECIAL.
2395 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2396 * fixed bug: :COUNT argument to sequence functions may be negative
2397 * fixed bug: body of DO-SYMBOLS may contain declarations
2398 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2399 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2400 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2401 symbol macro only once
2402 * fixed printing of call frame when argument list is unavailable
2403 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2404 * compiler no longer signals WARNING on unknown keyword
2407 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2408 * A beta-quality port to the mips architecture running Linux,
2409 based on the old CMUCL backend, has been made. It has been tested
2410 on a big-endian kernel, and works sufficiently well to be able to
2411 rebuild itself; it has not been tested in little-endian mode.
2412 * fixed an inconsistency between gencgc.c and purify.c which made
2413 dumping/loading .core files unreliable
2414 * fixed bug 120a: The compiler now deals correctly with IFs where
2415 the consequent is the same as the alternative, instead of
2416 misderiving the return type. (thanks to Alexey Dejneka)
2417 * fixed bug 113: Logical pathnames are now dumpable (the logical
2418 host is resolved at load-time, throwing an error if it is not
2420 * fixed bug 174: FORMAT's error message is slightly clearer when a
2421 non-printing character is used in a format directive.
2422 * fixed several bugs in compiler checking of type declarations, i.e.
2423 violations of the Python "declarations are assertions" principle
2424 (thanks to Alexey Dejneka)
2425 * fixed several bugs in PCL's error checking (thanks to Gerd
2427 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2429 * fixed bug in compilation of functions as first class values
2430 (thanks to Antonio Martinez-Shotton)
2431 * The compiler's handling TYPE-ERRORs which it can prove will
2432 inevitably happen at runtime has been cleaned up and corrected
2433 in several ways. (thanks to Alexey Dejneka)
2434 * improved argument type checking for various basic arithmetic
2435 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2436 lavished on them in the past that they can be compiled in many
2437 ways in different special cases
2438 * fixed bug 181: compiler checks validity of user supplied type
2440 * cleaned up code flushing in optimization: Function calls which
2441 should signal errors for safety purposes (e.g. which ANSI says
2442 should signal errors when their arguments are of incorrect type)
2443 are no longer optimized away.
2444 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2445 * incremented fasl file version number, because changes in the
2446 implementation of sequence functions like COERCE caused
2447 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2448 in old inline expansions) to become undefined. (Actually these
2449 changes were later undone, so we might very well be binary
2450 compatible with 0.7.7 after all, but leaving the version number
2451 incremented seemed like the simplest and most conservative
2454 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2455 * An alpha-quality port to the parisc architecture running Linux,
2456 based on the old CMUCL backend, has been made. This, even more so
2457 than the other backends, should be considered still a work in
2458 progress; known problems include that the Linux kernel in 64-bit
2459 mode does not propagate the correct sigcontext structure to
2460 userspace, and consequently SBCL on a parisc64 kernel will not
2462 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2463 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2464 suppressed.) Also now that the compiler is looking at declarations
2465 in the environment, it checks optimization declarations as well,
2466 and suppresses inlining when (> DEBUG SPEED).
2467 * More fixes have been made to treatment of floating point exception
2468 treatment and other Unix signals. In particular, floating point
2469 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2470 * The detection and handling of control stack exhaustion (infinite
2471 or very deeply nested recursion) has changed. Stack exhaustion
2472 detection is now done by write-protecting pages at the OS level
2473 and applies at all optimization settings; when found, a
2474 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2475 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2476 be caught using IGNORE-ERRORS.
2477 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2478 correctly when transforming recursive LABELS functions to LETs.
2479 (thanks to Alexey Dejneka)
2480 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2481 that are names of constants or global variables.
2482 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2483 alien routines with docstrings.
2484 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2485 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2487 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2488 (thanks to Eric Marsden)
2489 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2490 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2491 * Bug fix: The SPARC backend can now compile functions involving
2492 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2493 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2494 to a functions expecting a PATHNAME-DESIGNATOR.
2495 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2496 lambda lists. (thanks to Alexey Dejneka)
2497 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2498 functions so that EXPORTing the name of the class doesn't cause
2499 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2500 to Antonio Martinez for reporting this)
2501 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2502 constant-folded without failing an assertion. (thanks to Einar
2503 Floystad Dorum for reporting this)
2504 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2505 types are now dealt with more correctly by the compiler.
2506 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2507 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2508 behaviour analogous to RENAME-FILE. This puts its behaviour more
2509 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2511 * The fasl file version number has changed again. (because of the
2512 bug fix involving the names of PCL MAKE-INSTANCE functions)
2514 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2515 * bug fix: Floating point exceptions are treated much more
2516 consistently on the x86/Linux and PPC/Linux platforms.
2517 * Array initialization with :INITIAL-ELEMENT is now much faster for
2518 cases when the compiler cannot open code the array creation, but
2519 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2520 array accesses have also seen a speed increase.
2521 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2522 specified. (This was at the root of some bad interactions between
2523 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2525 * bug fix: Internal error arguments for undefined functions are now
2526 computed correctly on the PPC/Linux platform.
2527 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2528 Raymond Toy's patch for CMU CL)
2529 * Support for the Solaris 9 operating environment has been included
2530 (thanks to Daniel Merritt)
2531 * A very ugly but hopefully complete draft of the missing FFI chapter
2532 of the manual has been created by reformatting the corresponding
2533 CMU CL manual chapter into (currently very ugly and incoherent)
2534 DocBook and bringing it up to date for SBCL behavior. Thus, the
2535 manual is now essentially complete, at least by my extreme
2536 once-and-only-once standards, whereby it's acceptable to refer to
2537 the doc strings of SB-EXT functions as the primary documentation.
2538 * The fasl file version number has changed again, due to cleanup of
2539 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2540 is no longer a static symbol.)
2542 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2543 * SBCL now builds with OpenMCL (version 0.12) as the
2544 cross-compilation host; also, more progress has been made toward
2545 bootstrapping under CLISP.
2546 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2548 * bug 158 fixed: The compiler can now deal with integer loop
2549 increments different from 1; fixing this turned out also to fix
2551 * bug 169 fixed: no more bogus warnings about using lexical bindings
2552 despite the presence of perfectly good SPECIAL declarations (thanks
2553 to David Lichteblau)
2554 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2555 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2556 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2558 * bug 180 fixed: Method combination specifications no longer ignore
2559 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2560 * bug fix: Structure type predicate functions now check their argument
2561 count as they should.
2562 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2563 correctly. (thanks to Pierre Mai)
2564 * minor incompatible change: The --noprogrammer option is deprecated
2565 in favor of the new --disable-debugger option, which is very similar.
2566 (The major difference is that it takes effect at a slightly different
2567 time at startup, causing handling of errors in --sysinit and
2568 --userinit files will be affected differently.) The
2569 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2570 been added to allow this functionality to be controlled from ordinary
2571 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2572 maintainers, who might want to run non-interactive scripts to
2573 build SBCL cores which will later be used interactively.)
2574 * minor incompatible change: The LOAD function no longer, when given
2575 a wild pathname to load, loads all files matching that pathname.
2576 Instead, an error of type FILE-ERROR is signalled.
2578 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2579 * bug 147 fixed: The compiler preserves its block link/count
2580 invariants more correctly now so that it doesn't crash. (thanks
2582 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2584 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2585 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2586 no-value-for-this-character as NIL (as specified by ANSI).
2587 * HOST-NAMESTRING on physical pathnames now returns a string that is
2588 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2589 (thanks to Christophe Rhodes)
2590 * The Alpha port handles icache flushing more correctly. (thanks to
2592 * More progress has been made toward bootstrapping under CLISP. (thanks
2593 to Christophe Rhodes)
2594 * The fasl file format has changed again, because dynamic loading
2595 on OpenBSD (which has non-ELF object files) motivated some cleanups
2596 in the way that foreign symbols are transformed and passed around.
2597 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2598 is no longer treated as whitespace by the reader, but instead as
2599 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2600 |A
\7fB|, instead of A as it used to.
2602 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2603 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2604 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2605 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2606 this as of 0.7.1.45, but was left out of the previous news section
2607 (thanks to Dan Barlow)
2608 * SBCL now runs on the Solaris operating system on SPARC architectures
2609 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2610 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2611 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2612 * SPARC backend cleanups, allowing builds of cores optimized for V8
2613 and V9 SPARCS, and also emission of code targeted to a particular
2614 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2616 * SBCL is closer to bootstrapping under CLISP, thanks to various
2617 fixes by Christophe Rhodes.
2618 * The fasl file format has changed again, to allow the compiler's
2619 INFO database to support symbol macros.
2620 * The user manual (in doc/) is formatted into HTML more nicely.
2621 (thanks to coreythomas)
2622 * The system is smarter about SUBTYPEP relationships, especially
2623 those involving NOT types (including types such as ATOM which are
2624 represented internally using NOT types). Thus SUBTYPEP is less
2625 likely to return (VALUES NIL NIL) in general, and in particular
2626 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2628 * The fasl file format has changed again, because the internal
2629 representation of types now includes a new slot to support the new
2630 SUBTYPEP-of-NOT-types logic.
2631 * (not a change in the main branch of SBCL, but a related prototype
2632 which can hopefully be merged into the main branch of SBCL in the
2633 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2634 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2635 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2636 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2637 dispatch macro character combinations. (thanks to Alexey Dejneka)
2638 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2639 that superficially look like logical namestrings correctly.
2640 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2642 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2643 default supplied-p) for &optional and &key arguments. (thanks to
2646 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2647 * incompatible change: The compiler is now less aggressive about
2648 tail call optimization, doing it only when (> SPACE DEBUG) or
2649 (> SPEED DEBUG). (This is an incompatible change because there are
2650 programs which relied on the old CMU-CL-style behavior to optimize
2651 away their unbounded recursion which will now die of stack overflow.)
2652 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2653 for non-GENCGC systems has been increased to 20M (since that
2654 seems much closer to the likely performance optimum for modern
2655 systems than the old 4M value was)
2656 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2657 and *DEBUG-PRINT-LEVEL*
2658 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2659 of CMU CL's support for SPARC, and various endianness and other
2660 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2661 * new syntactic sugar for the Unix command line: --load foo.bar is now
2662 an alternate notation for --eval '(load "foo.bar")'.
2664 ** The system now detects stack overflow and handles it gracefully,
2665 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2666 optimization settings. (This is a good thing in general, and
2667 its introduction in this version should be particularly timely
2668 for anyone whose code fails because of suppression of tail
2670 ** The system now hunts for the C variable "environ" in a more
2671 devious way, to avoid segfaults when the C library version differs
2672 between compile time and run time. (thanks to Christophe Rhodes)
2673 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2674 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2675 ** The compiler no longer issues bogus style warnings for undefined
2676 classes in the same source file as the DEFCLASSes which defined
2677 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2679 ** fixes in CONDITION class precedence list for undefined function
2680 errors (thanks to Alexei Dejneka)
2681 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2682 correctly. (thanks to Dan Barlow)
2683 ** portability fixes aiming at bootstrapping under CLISP (thanks
2684 to Dave McDonald and Christophe Rhodes)
2685 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2686 ** fixes in type translation and and type inference (thanks to
2688 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2689 ** various fixes in the new ports (thanks to Dan Barlow)
2690 * several changes related to debugging:
2691 ** suppression of tail recursion, as noted above
2692 ** stack overflow detection, as noted above
2693 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2694 is now the default. (For some time encapsulation has been more
2695 reliable than the breakpoint-based :ENCAPSULATE NIL
2696 implementation, at least on X86 systems; and I just noticed that
2697 encapsulation also seems closer to the spirit of the ANSI
2700 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2702 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2703 up properly again. (There was a packaging bug in 0.7.0 which
2704 left their definitions in SB-SYS::LOAD-FOREIGN and
2705 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2706 vital for most things which interface to C-level interfaces,
2707 like extensions working with sockets or databases or
2708 Perl-compatible regexes or whatever, and the need to fix
2709 this bug is the main reason that 0.7.1 was released so
2711 ** DEFGENERIC is now choosier about the methods it redefines, so that
2712 reLOADing a previously-LOADed file containing DEFGENERICs does
2713 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2714 cycle now works as it should. (thanks to Alexey Dejneka)
2715 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2716 fixed by Christophe Rhodes. (He actually submitted this patch
2717 months ago, and I delayed until after 0.7.0.)
2718 ** Bug 111 (internal compiler confusion about runtime checks on
2719 FUNCTION types) was fixed by Alexey Dejneka.
2720 * Some internal cleanups (getting rid of variables which aren't
2721 needed now that the byte interpreter is gone) caused the fasl
2722 file format number to change again.
2724 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2725 * major incompatible change: The default fasl file extension, i.e. the
2726 default extension for files produced by COMPILE-FILE, has changed
2727 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2729 ** There are many changes in the implementation of the compiler.
2730 SBCL is now essentially a compiler-only implementation of ANSI
2731 Common Lisp. EVAL still "interprets" a few special cases, but
2732 almost all the interesting cases are handled by creating
2733 a LAMBDA expression, calling COMPILE on it, then calling
2734 FUNCALL on the result.
2735 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2736 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2737 Since the code is newer, there might still be some new bugs
2738 (though not as many as before Martin Atzmueller's fixes:-). But
2739 the new code is substantially simpler and clearer, and hopefully
2740 any remaining bugs will be simpler, less fundamental, and more
2741 fixable then the bugs in the old code.
2742 ** The revised compiler is still a little unsteady on its feet.
2744 *** The debugging information it produces (particularly the names
2745 of FUNCTION objects) is sometimes much less useful than what
2746 the old compiler produced.
2747 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2748 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2749 within a MACROLET) has been temporarily weakened.
2750 ** There are new compiler optimizations for various functions:
2751 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2752 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2753 *** the math functions TRUNCATE, FLOOR, and CEILING
2754 *** the function-of-all-trades COERCE
2755 Mostly these should be transparent, but there's one
2756 potentially-annoying problem (bug 117): when the compiler
2757 inline-expands a function and does type analysis on the result,
2758 it can create control paths which have type mismatches, and
2759 when it can't prove that those control paths aren't taken,
2760 it will issue WARNINGs about the type mismatches. This is
2761 a particular problem in practice for the new sequence functions.
2762 It's not clear how this should be fixed, and for now, a
2763 workaround is given in the entry for 117 in the BUGS file.
2764 ** (Because of the interaction between the two previous items --
2765 occasional inlining problems and new inline expansions -- some
2766 of the new sequence function optimizations won't really kick in
2767 completely until debugging information, and then inlining, are
2768 straightened out in some future version.)
2769 * minor incompatible changes:
2770 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2771 DIRECTORY no longer implicitly promotes NIL slots of its
2772 pathname argument to :WILD. In particular, when you ask for the
2773 contents of a directory (which you used to be able to do without
2774 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2775 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2776 ** changes in behavior that ANSI explicitly defines to be
2777 implementation dependent:
2778 *** The new compiler-only implementation still conforms with ANSI,
2779 but acts a little different than before. Besides the obvious
2780 changes in performance tradeoffs (that the cost per form passed
2781 to EVAL has gone up, and the cost per form executed by EVAL
2782 has gone down), the behavior of the system changes a little
2783 because there are no longer any interpreted function objects.
2784 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2785 e.g. doing COMPILE on the output of interactive DEFUN is
2787 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2789 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2790 has changed from (:CL) to NIL.
2791 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2792 e.g. #\Newline and #\Tab, don't have names specified in the
2793 ANSI Common Lisp standard, is now based on their ASCII symbolic
2794 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2795 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2796 are no longer used for output.
2797 ** changes in internal implementation constants:
2798 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2799 4 million. (If your application spends a lot of time GCing and
2800 you have a lot of RAM, you might want to experiment with
2801 increasing it even more.)
2802 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2803 However, almost all old code should still continue to work without
2804 immediate update, as SB-C-CALL is now a (deprecated) nickname
2806 ** Old operator names in the style DEF-FOO are now deprecated in
2807 favor of new corresponding names DEFINE-FOO, for consistency with
2808 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2809 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2810 internal symbols, but a few supported extensions like
2811 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2812 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2813 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2814 etc. as you get deeper into recursive calls to the debugger
2815 command loop, instead of the old "5]", "5]]", "5]]]"
2816 sequence. (I was motivated to do this when squabbles between
2817 ILISP and SBCL left me very deeply nested in the debugger. In the
2818 short term, this change will probably provoke more ILISP/SBCL
2819 squabbles, but hopefully it will be an improvement in the long run.)
2820 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2821 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2822 function, since declaiming return types involving aliens is
2823 (1) annoyingly messy to do by hand and (2) vital to efficient
2824 compilation of code which calls such functions.
2825 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2826 longer reexported by the SB-EXT package. They're solely useful
2827 for alien code, so it seems more logical that you should get
2828 them from the SB-ALIEN package, not in SB-EXT.
2829 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2830 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2831 features. Instead, the code that they used to control is always
2832 built into the system.
2833 * many other bug fixes
2834 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2835 advantage of the new EVAL-WHEN stuff and to clean them up in
2836 general, and they are now more ANSI-compliant in a number of
2837 ways. Martin Atzmueller is responsible for a lot of this.
2838 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2840 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2841 *** correct ERROR type for various file operations
2842 *** some fixes for Lisp streams
2843 *** DEFMETHOD syntax checking
2844 *** changing old weird representation of debug information as
2845 strings (which, among their other deficiencies, don't transform
2846 correctly when you rename packages, and don't change their
2847 print representation when you change things like *PACKAGE*
2848 and *PRINT-LENGTH*) to symbols and lists of symbols
2849 He also made several improvements and fixed several bugs in DESCRIBE.
2850 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2852 *** misbehavior of WRITE-STRING/WRITE-LINE
2853 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2854 and several other LOOP problems as well
2855 *** DIRECTORY when similar filenames are present
2856 *** DEFGENERIC with :METHOD options
2857 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2858 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2859 *** argument ordering in FIND with :TEST option
2860 *** mishandled package designator argument in APROPOS-LIST
2861 *** various problems in the backquote readmacro
2862 *** a bug in APROPOS
2863 *** probably some others that I'm not describing very well here,
2864 since the CVS log documents them by reference to sbcl-devel
2865 messages, and the SourceForge archives aren't working well.:-(
2866 ** Dan Barlow improved the Alpha port (and is making progress on the
2867 PPC port, for those of you who think different).
2868 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2869 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2870 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2871 continues to work on the SPARC port (for those of you in a position
2872 to look down upon our little PC-compatible boxes from a great height).
2873 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2874 and uses that copy, rather than the current dynamic value, when
2875 it's trying to decide whether to truncate output. Thus e.g.
2876 (let ((*print-lines* 50))
2877 (pprint-logical-block (stream nil)
2879 (let ((*print-lines* 8))
2880 (print (aref possiblybigthings i) stream)))))
2881 should now truncate the logical block only at 50 lines, instead of
2882 often truncating it at 8 lines, as it did before.
2883 * The doc/cmucl/ directory, containing old CMU CL documentation
2884 from the time of the fork, is no longer part of the base system.
2885 SourceForge has shut down its anonymous FTP service, and with it
2886 my original plan for distributing the old CMU CL documentation
2887 there. For now, if you need these files you can download an old
2888 SBCL source release and extract them from it.
2889 * The fasl file version number changed again, for dozens of reasons,
2890 some of which are apparent above.
2892 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2893 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2894 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2895 patch, so that DISASSEMBLE output is much nicer.
2896 * The code in the SB-PROFILE package now seems reasonably stable.
2897 I still haven't decided what the final interface should look like
2898 (I'd like PROFILE to interact cleanly with TRACE, since both
2899 facilities use function encapsulation) but if you have a need
2900 for profiling now, you can probably use it successfully with
2901 the current CMU-CL-style interface.
2902 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2903 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2904 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2905 link now returns the dangling link itself, and for similar
2906 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2907 link itself. (In these cases the old code signalled an error and
2908 looped endlessly, respectively.) Thus, DIRECTORY now works even
2909 in the presence of dangling and cyclic symbolic links.
2910 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2911 is now a supported extension again, since the consensus on
2912 sbcl-devel was that it can be useful for ordinary development
2913 work, not just for debugging SBCL itself.
2914 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2915 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2916 that currently-defined functions might be redefined later with
2917 different return types.
2918 * Hash tables can be printed readably, as inspired by CMU CL code
2919 of Eric Marsden and SBCL code of Martin Atzmueller.
2920 * better error handling in CLOS method combination, thanks to
2921 Martin Atzmueller porting Pierre Mai's CMU CL patches
2922 * more overflow fixes for >16Mbyte I/O buffers
2923 * A bug in READ has been fixed, so that now a single Ctrl-D
2924 character suffices to cause end-of-file on character streams.
2925 In particular, now you only need one Ctrl-D at the command
2926 line (not two) to exit SBCL.
2927 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2929 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2930 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2931 does the right thing.
2932 * The implementation of some type tests, especially for CONDITION
2933 types, is now tidier and maybe faster, due to CMU CL code
2934 originally by Douglas Crosher, ported by Martin Atzmueller.
2935 * Some math functions have been fixed, and there are new
2936 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2937 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2938 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2939 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2940 likely to be built into sbcl-0.7.0.)
2941 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2942 CL, ported to SBCL by Martin Atzmueller)
2943 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2944 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2945 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2946 * There's a new slam.sh hack to shorten the edit/compile/debug
2947 cycle for low-level changes to SBCL itself, and a new
2948 :SB-AFTER-XC-CORE target feature to control the generation of
2949 the after-xc.core file needed by slam.sh.
2950 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2951 COMPILE-FILE is no longer supported, so that now every function
2952 gets an entry point, so that block compilation looks a little
2953 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2954 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2955 returns the number of bytes consed since the system started,
2956 rather than the number consed since the first time the function
2957 was called. (The new definition parallels ANSI functions like
2958 CL:GET-INTERNAL-RUN-TIME.)
2959 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2960 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2961 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2962 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2963 * Fasl file version numbers are now independent of the target CPU,
2964 since historically most system changes which required version
2965 number changes have affected all CPUs equally. Similarly,
2966 the byte fasl file version is now equal to the ordinary
2969 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2970 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2971 is no longer recognized. I apologize for this, because it was
2972 listed in SB-EXT as a supported extension, but I found that
2973 its existing behavior was poorly specified, as well as incorrectly
2974 specified, and it looked like too much of a mess to straighten it
2975 out. I have enough on my hands trying to get ANSI stuff to work..
2976 * many patches ported from CMU CL by Martin Atzmueller, with
2977 half a dozen bug fixes in pretty-printing and the debugger, and
2978 half a dozen others elsewhere
2979 * fixed bug 13: Floating point infinities are now supported again.
2980 They might still be a little bit flaky, but thanks to bug reports
2981 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2982 as flaky as they were.
2983 * The --noprogrammer command line option is now supported. (Its
2984 behavior is slightly different in detail from what the old man
2985 page claimed it would do, but it's still appropriate under the
2986 same circumstances that the man page talks about.)
2987 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2988 are now supported, and enabled by default. Thus, the compiler can
2989 handle many floating point and complex operations much less
2990 inefficiently. (Thus e.g. you can implement a complex FFT
2992 * The compiler now detects type mismatches between DECLAIM FTYPE
2993 and DEFUN better, and implements CHECK-TYPE more correctly, and
2994 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2995 thanks to patches from Martin Atzmueller.
2996 * various fixes to make the cross-compiler more portable to
2997 ANSI-conforming-but-different cross-compilation hosts (notably
2998 Lispworks for Windows, following bug reports from Arthur Lemmens)
2999 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3000 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3001 fixed thanks to Pierre Mai's CMU CL patches.
3002 * fixes to keep the system from overflowing internal counters when
3003 it tries to use i/o buffers larger than 16M bytes
3004 * fixed bug 45a: Various internal functions required to support
3005 complex special functions have been merged from CMU CL sources.
3006 (When I was first setting up SBCL, I misunderstood a compile-time
3007 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3008 * improved support for type intersection and union, fixing bug 12
3009 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3010 more obscure bugs as well
3011 * some steps toward byte-compiling non-performance-critical
3012 parts of the system, courtesy of patches from Martin Atzmueller
3013 * Christophe Rhodes has made some debian packages of sbcl at
3014 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3015 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3016 stable, but are nonetheless usable. When he's ready, I'd be happy
3017 to add them to the SourceForge "File Releases" section. (And if
3018 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3019 * new fasl file format version number (because of changes in
3020 internal representation of (OR ..) types to accommodate the new
3021 support for (AND ..) types, among other things)
3023 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3024 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3026 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3028 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3029 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3030 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3031 * DESCRIBE now works on structure objects again.
3032 * Most function call argument type mismatches are now handled as
3033 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3034 know whether the function will be redefined before the call is
3035 executed. (The compiler could flag local calls with full WARNINGs,
3036 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3037 it doesn't keep track of enough information to know whether calls
3038 are local in this sense.)
3039 * Compiler output is now more verbose, with messages truncated
3040 later than before. (There should be some supported way for users
3041 to override the default verbosity, but I haven't decided how to
3042 provide it yet, so this behavior is still controlled by the internal
3043 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3044 src/compiler/ir1util.lisp.)
3045 * Fasl file format version numbers have increased again, because
3046 support for the Gray streams extension changes the layout of the
3047 system's STREAM objects.
3048 * The Gray subclassable streams extension now works, thanks to a
3049 patch from Martin Atzmueller.
3050 * The full LOAD-FOREIGN extension (not just the primitive
3051 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3052 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3053 but like most other programs, it defaults to copying the Unix
3054 environment from the original process instead of starting the
3055 new process in an empty environment.
3056 * Extensions which manipulate the Unix environment now support
3057 an :ENVIRONMENT keyword option which doesn't smash case or
3058 do other bad things. The CMU-CL-style :ENV option is retained
3059 for porting convenience.
3060 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3061 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3063 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3065 * A patch from Martin Atzmueller seems to have solved the SIGINT
3066 problem, and as far as we know, signal-handling now works cleanly.
3067 (If you find any new bugs, please report them!)
3068 * The system no longer defaults Lisp source file names to types
3069 ".l", ".cl", or ".lsp", but only to ".lisp".
3070 * The compiler no longer uses special default file extensions for
3071 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3072 seems to expect a single default extension for all compiled code,
3073 and there's no compelling reason to try to stretch the standard
3074 to allow two different extensions.) Instead, byte-compiled files
3075 default to the same extension as native-compiled files.
3076 * Fasl file format version numbers have increased again, because
3077 a rearrangement of internal implementation packages made some
3078 dumped symbols in old fasl files unreadable in new cores.
3079 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3080 many fewer weird special cases.
3081 * Bug #17 (differing COMPILE-FILE behavior between logical and
3082 physical pathnames) has been fixed, and some related misbehavior too,
3083 thanks to a patch from Martin Atzmueller.
3084 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3085 by Tim Moore, ported to SBCL by Martin Atzmueller.
3086 * Martin Atzmueller fixed several filesystem-related problems,
3087 including bug #36, in part by porting CMU CL patches, which were
3088 written in part by Paul Werkowski.
3089 * More compiler warnings in src/runtime/ are gone, thanks to
3090 more patches from Martin Atzmueller.
3091 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3094 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3096 * DESCRIBE now works on CONDITION objects.
3097 * The debugger now handles errors which arise when trying to print
3098 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3100 * The build system now uses an additional file, customize-target-features.lisp,
3101 to allow local modifications to the target *FEATURES* list. (The point of
3102 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3103 debugging features enabled, without having to worry about propagating it
3104 into everyone's system when I do a "cvs update".) When no
3105 customize-target-features.lisp file exists, the target *FEATURES* list
3106 should be constructed the same way as before.
3107 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3108 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3109 warn about behavior which is undefined under the ANSI spec. Note: This
3110 is specified by ANSI, but it's not very popular with programmers.
3111 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3112 macro in the SBCL sources for an example of a workaround which you
3113 might use to make portable ANSI-standard code which does what you want.
3114 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3115 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3116 when it's used as a non-toplevel form. (This is required in order
3117 to implement the DEFCONSTANT-EQX macro.)
3118 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3119 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3120 funny things, due to bugs in the implementation of EVAL-WHEN
3121 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3122 non-toplevel DEFCONSTANTs are uncommon.)
3123 * The core file version number and fasl file version number have been
3124 incremented, because the old noncompliant DEFCONSTANT behavior involved
3125 calling functions which no longer exist, and because I also took the
3126 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3127 * fixed bug 1 (error handling before read-eval-print loop starts), and
3128 redid debugger restarts and related debugger commands somewhat while
3130 ** The QUIT debugger command is gone, since it did something
3131 rather different than the SB-EXT:QUIT command, and since it never
3132 worked properly outside the main toplevel read/eval/print loop.
3133 Invoking the new TOPLEVEL restart provides the same functionality.
3134 ** The GO debugger command is also gone, since you can just invoke
3135 the CONTINUE restart directly instead.
3136 ** The TOP debugger command is also gone, since it's redundant with the
3137 FRAME 0 command, and since it interfered with abbreviations for the
3139 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3141 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3142 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3143 some porting work by Martin Atzmueller)
3144 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3145 with a boatload of patches which he ported to SBCL. Now that those
3147 ** The system tries to make sure that its low-priority messages
3148 are prefixed by semicolons, to help people who like to use
3149 syntax highlighting in their ILISP buffer. (This patch
3150 was originally due to Raymond Toy.)
3151 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3152 patches originally written by Raymond Toy.
3153 ** The compiler understands coercion between single-value and
3154 multiple-VALUES type expressions better, getting rid of some very
3155 weird behavior, thanks to patches originally by Robert MacLachlan
3156 and Douglas Crosher.
3157 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3158 lambda lists, thanks to a patch originally by Pierre Mai.
3159 ** The system no longer bogusly warns about "abbreviated type
3161 ** The compiler gets less confused by inlining and RETURN-FROM,
3162 thanks to some patches originally by Tim Moore.
3163 ** The system no longer hangs when dumping circular lists to fasl
3164 files, thanks to a patch originally from Douglas Crosher.
3165 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3166 undefined function error.
3167 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3168 in structure slot initforms) for now, documented workaround instead:-|
3169 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3170 * fixed bug 5: added stubs for various Gray stream functions called
3171 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3172 installed, at least appropriate type errors are generated
3173 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3174 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3175 consistently in DEFMETHOD forms.
3176 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3177 it doesn't seem to affect SBCL after all
3178 * The C runtime system now builds with better optimization and many
3179 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3181 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3183 * The system is now under CVS at SourceForge (instead of the
3184 CVS repository on my home machine).
3185 * The new signal handling code has been tweaked to treat register
3186 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3187 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3188 downstream, e.g. in debug-int.lisp, has implicit dependencies
3189 on the unsignedness of integer representation of machine words,
3190 and that caused the system to bomb out with infinite regress
3191 when trying to recover from type errors involving signed values,
3192 e.g. (BUTLAST '(1 2 3) -1).)
3193 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3194 (This was one of the bugs Peter Van Eynde reported back in July.)
3195 * The system now uses code inspired by Colin Walters' O(N)
3196 implementation of MAP (from the cmucl-imp@cons.org mailing
3197 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3198 inline the MAP operation, and there is more than one
3199 sequence argument to the MAP call (so that it can't just
3200 do ETYPECASE once and for all based on the type of the
3201 single sequence argument). (The old non-inline implementation
3202 of the general M-argument sequence-of-length-N case required
3203 O(M*N*N) time when any of the sequence arguments were LISTs.)
3204 * The QUIT :UNIX-CODE keyword argument has been renamed to
3205 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3206 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3208 * Raymond Wiker's patches to port dynamic loading from Linux to
3209 FreeBSD have been added.
3210 * The BUGS file is now more nearly up to date, thanks in large part
3211 to Martin Atzmueller's review of it.
3212 * The debugger now flushes standard output streams before it begins
3213 its output ("debugger invoked" and so forth).
3214 * The core version number and fasl file version number have both
3215 been incremented, because of incompatible changes in the layout
3217 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3218 used to not be called for a saved Lisp image.)
3219 * Martin Atzmueller's version of a patch to fix a compiler crash,
3220 as posted on sbcl-devel 13 September 2000, has been installed.
3221 * Instead of installing Martin Atzmueller's patch for the
3222 compiler transform for SUBSEQ, I deleted the compiler transform,
3223 and transforms for some similar consing operations.
3224 * A bug in signal handling which kept TRACE from working on OpenBSD
3226 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3227 away by constant folding
3228 * The system now defines its address space constants in one place
3229 (in the Lisp sources), and propagates them automatically elsewhere
3230 (through GENESIS and the sbcl.h file). Therefore, patching the
3231 address map is less unnecessarily tedious and error-prone. The
3232 Lisp names of address space constants have also been systematized.
3233 * CVS tags like dollar-Header-dollar have been removed from
3234 the sources, because they have never saved me trouble and
3235 they've been source of trouble working with patches and other
3236 diff-related operations.
3237 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3238 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3240 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3242 * The system has been ported to OpenBSD.
3243 * The system now compiles with a simple "sh make.sh" on the systems
3244 that it's supported on. I.e., now you no longer need to tweak
3245 text in the target-features.lisp-expr and symlinks in src/runtime/
3246 by hand, the make.sh takes care of it for you.
3247 * The system is no longer so grossly inefficient when compiling code
3248 involving vectors implemented as general (not simple) vectors (VECTOR T),
3249 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3250 which dares to use the various sequence functions on non-simple
3251 vectors, takes less of a performance hit.
3252 * There is now a primitive type predicate VECTOR-T-P
3253 to test for the (VECTOR T) type, so that e.g.
3254 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3255 can now be compiled with some semblance of efficiency. (The old code
3256 turned the type declaration into a full call to %TYPEP at runtime!)
3257 * AREF on (VECTOR T) is still not fast, since it's still compiled
3258 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3259 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3260 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3261 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3262 without making full calls to SUBTYPEP at runtime.
3263 (Some analogous efficiency issues for non-simple vectors specialized to
3264 element types other than T, or for non-simple multidimensional arrays,
3265 have not been addressed. They could almost certainly be handled the
3266 same way if anyone is motivated to do so.)
3267 * The changes in array handling break binary compatibility, so
3268 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3269 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3270 instead of (VALUES T T).
3271 * By following the instructions that Dan Barlow posted to sbcl-devel
3272 on 2 July 2000, I was able to enable primitive dynamic object
3273 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3274 functionality is not implemented (since it calls ld to resolve
3275 library references automatically, requiring RUN-PROGRAM for its
3276 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3277 try to resolve library references) is now supported.
3278 * The system now flushes the standard output streams when it terminates,
3279 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3280 them at several other probably-convenient times, e.g. in each pass of
3281 the toplevel read-eval-print loop, and after evaluating a form given
3282 as an "--eval" command-line option. (These changes were motivated by a
3283 discussion of stream flushing issues on cmucl-imp in August 2000.)
3284 * The source transform for TYPEP of array types no longer assumes
3285 that an array whose element type is a not-yet-defined type
3286 is implemented as an array of T, but instead punts, so that the
3287 type will be interpreted at runtime.
3288 * There is now some support for cross-compiling in make.sh: each of
3289 the phases of make.sh has its own script. (This should be transparent
3290 to people doing ordinary, non-cross-compile builds.)
3291 * Since my laptop doesn't have hundreds of megabytes of memory like
3292 my desktop machine, I became more motivated to do some items on
3293 my to-do list in order to reduce the size of the system a little:
3294 ** Arrange for various needed-only-at-cold-init things to be
3295 uninterned after cold init. To support this, those things have
3296 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3297 symbols with such names are now uninterned after cold init).
3298 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3299 fasl files for cold load.
3300 ** Remove the old compiler structure pooling code (which used to
3301 be conditional on the target feature :SB-ALLOC) completely.
3302 ** Redo the representation of some data in cold init to be more compact.
3303 (I also looked into supporting byte compiled code at bootstrap time,
3304 which would probably reduce the size of the system a lot, but that
3305 looked too complicated, so I punted for now.)
3306 * The maximum signal nesting depth in the src/runtime/ support code has
3307 been reduced from 4096 to 256. (I don't know any reason for the very
3308 large old value. If the new smaller value turns out to break something,
3309 I'll probably just bump it back up.)
3310 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3312 * Many, many bugs reported by Peter Van Eynde have been added to
3313 the BUGS list; some have even been fixed.
3314 * While enabling dynamic object file loading, I tried to make the
3315 code easier to understand, renaming various functions and variables
3316 with less ambiguous names, and changing some function calling
3317 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3318 * While trying to figure out how to do the OpenBSD port, I tried to
3319 clean up some of the code in src/runtime/. In particular, I dropped
3320 support for non-POSIX signal handling, added various comments,
3321 tweaked the code to reduce the number of compilation warnings, and
3322 renamed some files to increase consistency.
3323 * To support the new automatic configuration functionality in make.sh,
3324 the source file target-features.lisp-expr has been replaced with the
3325 source file base-target-features.lisp-expr and the machine-generated
3326 file local-target-features.lisp-expr.
3327 * fixed a stupid quoting error in make.sh so that using CMU CL
3328 "lisp -batch" as cross-compilation host works again
3330 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3332 * DESCRIBE no longer tries to call itself recursively to describe
3333 bound/fbound values, so that it no longer fails on symbols which are
3334 bound to themselves (like keywords, T, and NIL).
3335 * DESCRIBE now works on generic functions.
3336 * The printer now prints less-screwed-up representations of closures
3337 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3338 * A private symbol is used instead of the :EMPTY keyword previously
3339 used to mark empty slots in hash tables. Thus
3340 (DEFVAR *HT* (MAKE-HASH-TABLE))
3341 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3342 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3343 now does what ANSI says that it should. (You can still get
3344 similar noncompliant behavior if bang on the hash table
3345 implementation with all the symbols you get back from
3346 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3347 This breaks binary compatibility, since tests for equality to
3348 :EMPTY are wired into things like the macroexpansion of
3349 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3351 * There's now a minimal placeholder implementation for CL:STEP,
3352 as required by ANSI.
3353 * An obscure bug in the interaction of the normal compiler, the byte
3354 compiler, inlining, and structure predicates has been patched
3355 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3356 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3358 * Missing ordinary arguments in a macro call are now detected even
3359 when the macro lambda list contains &KEY or &REST.
3360 * The debugger no longer complains about encountering the top of the
3361 stack when you type "FRAME 0" to explicitly instruct it to go to
3362 the top of the stack. And it now prints the frame you request even
3363 if it's the current frame (instead of saying "You are here.").
3364 * As specified by ANSI, the system now always prints keywords
3365 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3367 * The default initial SIZE of HASH-TABLEs is now smaller.
3368 * Type information from CLOS class dispatch is now propagated
3369 into DEFMETHOD bodies, so that e.g.
3370 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3372 is now basically equivalent to
3373 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3374 (DECLARE (TYPE SINGLE-FLOAT X))
3376 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3377 operation, without having to do run-time type dispatch.
3378 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3379 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3380 * Since I don't seem to be making big file reorganizations very often
3381 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3382 is overflowing my ability to conveniently back them up), I've finally
3383 checked the system into CVS. (The CVS repository is on my home system,
3384 not at SourceForge -- putting it on SourceForge might come later.)
3385 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3386 high-level GC-NOTIFY-FOO functions send their output. (There's
3387 still very little control of where low-level verbose GC functions
3388 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3389 less than it used to -- the GC-NOTIFY-FOO functions are now under
3390 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3391 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3392 in only one place in the source code, and propagates it automatically
3393 everywhere that it's needed. Thus e.g. when I bump the version from
3394 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3395 * The C source files now include boilerplate legalese and documentation
3396 at the head of each file (just as the Lisp source files already did).
3397 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3398 to his page will be replaced with a link to his new CLiki service.
3400 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3402 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3403 * The build process now looks for GNU make under the default name "gmake",
3404 instead of "make" as it used to. If GNU make is not available as "gmake"
3405 on your system, you can change this default behavior by setting the
3406 GNUMAKE environment variable.
3407 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3408 can build without error under CMU CL.
3410 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3412 * There is now a partial SBCL user manual (with some new text and some
3413 text cribbed from the CMU CL manual).
3414 * The beginnings of a profiler have been added (starting with the
3415 CMU CL profiler and simplifying and cleaning up). Eventually the
3416 main interface should be through the TRACE macro, but for now,
3417 it's still accessed through vaguely CMU-CL-style functions and macros
3418 exported from the package SB-PROFILE.
3419 * Some problems left over from porting CMU CL to the new
3420 cross-compilation bootstrap process have been cleaned up:
3421 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3422 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3423 part when getting rid of the compiler *BACKEND* stuff.)
3424 ** The value of *NULL-TYPE* was screwed up, because it was
3425 being initialized before the type system knew the final
3426 definition of the 'NULL type. This screwed up several key
3427 optimizations in the compiler, causing inefficiency in all sorts
3428 of places. (I found it because I wanted to understand why
3429 GET-INTERNAL-RUN-TIME was consing.)
3430 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3431 PROCLAIM FTYPE information. Unfortunately this broke binary
3432 compatibility again, since now the forms output by DEFGENERIC
3433 to refer to functions which didn't exist in 0.6.3.
3434 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3435 can use the new (as of 0.6.3) transform for SOME into MAP into
3437 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3438 values to (INTEGER 0 1000000), so that the time code will no longer
3439 occasionally get blown up by Linux returning 1000000 microseconds
3440 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3441 its output conform to the ANSI spec. (Alas, this makes its output
3442 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3443 * A full call to MAP NIL with a single sequence argument no longer conses.
3444 * fixes to problems pointed out by Martin Atzmueller:
3445 * The manual page no longer talks about multiprocessing as though
3446 it were currently supported.
3447 * The ILISP support patches have been removed from the distribution,
3448 because as of version 5.10.1, ILISP now supports SBCL without us
3449 having to maintain patches.
3450 * added a modified version of Raymond Toy's recent CMU CL patch for
3451 EQUALP comparison of HASH-TABLE
3453 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3455 * The system still can't cross-compile itself with
3456 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3457 entails), but at least it can compile more of itself that way
3458 than it used to be able to, and various buglets which were uncovered
3459 by trying to cross-compile itself that way have now been fixed.
3460 * This release breaks binary compatibility again. This time
3461 at least I've incremented the FASL file format version to 2, so that the
3462 problem can be detected reliably instead of just causing weird errors.
3463 * various new style warnings:
3464 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3465 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3466 ** using the lexical binding of a variable named in the *FOO* style
3467 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3468 as specified by ANSI.
3469 * *RANDOM-STATE* is no longer automatically initialized from
3470 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3471 default behavior of the system is to repeat its behavior every time
3472 it's run. If you'd like to change this behavior, you can always
3473 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3474 old convention there was no comparably easy way to get the system to
3475 repeat its behavior every time it was run.
3476 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3477 FTYPE declarations has been removed, in favor of their ANSI
3478 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3479 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3480 the types of their sequence arguments can be determined at compile time.
3481 This is done through a new open code expansion for MAP which eliminates
3482 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3483 argument types can be determined at compile time.
3484 * The optimizer now transforms COERCE into an identity operation when it
3485 can prove that the coerced object is already of the correct type. (This
3486 can be a win for machine generated code, including the output of other
3487 optimization transforms, such as the MAP transform above.)
3488 * Credit information has been moved from source file headers into CREDITS.
3489 * Source file headers have been made more standard.
3490 * The CASE macro now compiles without complaining even when it has
3493 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3495 * (Note that the way that the PCL macroexpansions were rewritten
3496 to accommodate the change in DEFGENERIC below breaks binary
3497 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3498 not run under sbcl-0.6.2. Once we get out of alpha releases,
3499 i.e. hit release 1.0.0, we'll probably try to maintain binary
3500 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3501 and sbcl-1.4.4. Until then, however, it might be fairly common
3502 for maintenance releases to break binary compatibility.)
3503 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3505 * The reporting of unhandled signals has been changed to print some
3506 explanatory text as well as the report form. (Previously only
3507 the report form was printed.)
3508 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3509 it defines, so that the compiler no longer issues undefined function
3510 warnings for compiled-but-not-yet-loaded generic functions.
3511 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3512 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3513 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3514 section of the ANSI spec are supported.
3515 * The "" nickname for the "KEYWORD" package has been removed.
3516 The reader still handles symbol tokens which begin with a package marker
3517 as keywords, but it doesn't expose its mechanism for doing so in the
3518 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3519 * The system now issues STYLE-WARNINGs for contradictory TYPE
3520 proclamations. (Warnings for contradictory FTYPE proclamations would
3521 be nice too, but those can't be done usefully unless the type system
3522 is made smarter about FUNCTION types.)
3523 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3524 systematized, so that "*target-*.lisp is supposed to exist only on the
3525 target and imply that there's a related file which exists on the
3526 host, and *host-*.lisp is supposed to exist only on the host and imply
3527 that there's a related file which exists on the target. This involves a
3528 lot of renaming. Hopefully the acute confusion caused by the renaming
3529 will be justified by the reduction in chronic confusion..
3530 ** runtime-type.lisp -> early-target-type.lisp
3531 ** target-type.lisp -> late-target-type.lisp
3532 ** early-host-format.lisp -> early-format.lisp
3533 ** late-host-format.lisp -> late-format.lisp
3534 ** host-error.lisp -> misc-error.lisp
3535 ** early-error.lisp -> early-target-error.lisp
3536 ** late-error.lisp -> late-target-error.lisp
3537 ** host-defboot.lisp -> early-defboot.lisp
3538 ** code/misc.lisp -> code/target-misc.lisp
3539 ** code/host-misc.lisp -> code/misc.lisp
3540 ** code/numbers.lisp -> code/target-numbers.lisp
3541 ** code/early-numbers.lisp -> numbers.lisp
3542 ** early-host-type.lisp -> early-type.lisp
3543 ** late-host-type.lisp -> late-type.lisp
3544 ** host-typep.lisp -> typep.lisp
3545 ** load.lisp -> target-load.lisp
3546 ** host-load.lisp -> load.lisp
3547 ** host-disassem.lisp -> disassem.lisp
3548 ** host-insts.lisp -> insts.lisp
3549 ** byte-comp.lisp -> target-byte-comp.lisp
3550 ** host-byte-comp.lisp -> byte-comp.lisp
3551 ** host-signal.lisp -> signal.lisp
3552 ** host-defstruct.lisp -> defstruct.lisp
3553 ** late-target-type.lisp -> deftypes-for-target.lisp
3554 Furthermore, several other previously target-only files foo.lisp (e.g.
3555 hash-table.lisp and random.lisp) have been split into a target-and-host
3556 foo.lisp file and a target-only target-foo.lisp file, with their key type
3557 definitions in the target-and-host part, so that the cross-compiler will
3558 know more about target types.
3559 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3560 gone away. In their place are various *BACKEND-FOO* variables
3561 corresponding to the slots of the old structure.
3562 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3563 into the target SBCL has been fixed.
3564 * The chill.lisp system for loading cold code into a running SBCL
3566 * Support for the CMU CL "scavenger hook" extension has been removed.
3567 (It was undocumented and unused in the CMU CL sources that SBCL was
3568 derived from, and stale in sbcl-0.6.1.)
3569 * Various errors in the cross-compiler type system were detected
3570 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3571 (enabling various consistency checks). Many of them were fixed,
3572 but some hard problems remain, so the compiler is back to
3573 running without *TYPE-SYSTEM-INITIALIZED* for now.
3574 * As part of the cross-compiler type system cleanup, I implemented
3575 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3576 * I have started adding UNCROSS calls throughout the type system
3577 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3578 on cross-compiler input files will be able to go away, and various
3580 * CONSTANTP now returns true for quoted forms (as explicitly required
3583 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3585 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3586 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3587 (MAKE-LIST -1). In the longer term, it ought to become true
3588 that declarations are assertions even at SAFETY 1. For now, it's not
3589 quite true even at SAFETY 3, but it's at least more nearly true..
3590 (Note that this change seems to increases the size of the system by
3591 O(5%) and to decrease the speed of the compiler by 20% or more.)
3592 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3593 for the problem of printing dozens of lines of distracting information
3594 about low-level system machinery as part of the top stack frame
3595 on entry to the debugger when an undefined function was called.
3596 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3597 is not reset to COMMON-LISP-USER.
3598 * Compilation of stuff related to dyncount.lisp has been made conditional
3599 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3600 smaller. The various dyncount-related symbols have been moved into
3601 a new "SB-DYNCOUNT" package.
3602 * tty-inspect.lisp has been renamed to inspect.lisp.
3603 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3604 feature has gone away. (When we eventually port to other flavors of
3605 libc and/or Unix, we'll try to make the differences between flavors
3606 invisible at the user level.)
3607 * Various other *FEATURES* tags, and/or their associated conditionals,
3608 have been removed if obsolescent, or given better documentation, or
3609 sometimes given more-mnemonic names.
3611 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3613 * tidied up "make.sh" script
3614 * tidied up system directory structure
3615 * better "clean.sh" behavior
3616 * added doc/FOR-CMUCL-DEVELOPERS
3617 * many many small tweaks to output format, e.g. removing possibly-confusing
3618 trailing #\. character in DESCRIBE-INSTANCE
3619 * (EQUALP #\A 'A) no longer signals an error.
3620 * new hashing code, including EQUALP hashing
3621 * tidied up Lisp initialization and toplevel
3622 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3623 * command line argument processing
3624 * added POSIX-GETENV function to deal with Unix-ish environment variables
3625 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3626 terminating SBCL on EOF
3627 * non-verbose GC by default
3628 * There is no more "sbcl" shell script; the sbcl file is now the C
3629 runtime executable (just like CMU CL).
3630 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3632 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3633 symbol and package information as Lisp native symbol and package objects
3634 instead of strings naming symbols and strings naming packages. This way,
3635 whenever packages are renamed (as in warm init), debug information is
3636 transformed along with everything else.
3637 * tweaked the optimization policy declarations which control the building
3638 of SBCL itself. Now, among other things, the system no longer saves
3639 source location debugging information. (This helps two problems at once
3640 by reducing SBCL size and by keeping SBCL from trying to look for its
3641 sources -- which may not exist -- when reporting errors.)
3642 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3643 debugging and testing purposes
3644 * cleaned up printing, making the printer call PRINT-OBJECT for
3645 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3646 methods, giving nearly-ANSI behavior
3647 * converted almost all special variables to use *FOO* naming convention
3648 * deleted PARSE-TIME functionality, since it can be done portably
3649 * moved some files out of cold init into warm init
3650 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3652 * regularized formatting of source files
3653 * added an install.sh script
3654 * fixed ridiculous memory usage of cross-compiler by making
3655 compiler/alloc.lisp not try to do pooling unless it can hook
3656 itself into the GC of the cross-compilation host. Now the system
3657 builds nicely on my old laptop.
3658 * added :SB-ALLOC in target-features.lisp-expr
3659 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3660 was not implemented)
3661 * re-did condition handling and note reporting in the compiler. Notes
3662 are no longer handled by signalling conditions. Style warnings
3663 and warnings are handled more correctly and reported in such a way
3664 that it's easy to find one or the other in your output (so that you
3665 can e.g. figure out which of many problems caused COMPILE-FILE to
3667 * changed the severity of several compiler warnings from full WARNING
3668 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3669 compiler note reporting so that it doesn't use the condition system
3670 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3671 COMPILE-FILE command)
3672 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3673 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3674 no longer supported, since their implementation was deeply intertwingled
3675 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3676 * removed bogus "support" for compiler macros named (SETF FOO), and
3677 removed the compiler macro for SETF INFO (but only after making a fool
3678 of myself on the cmucl-imp mailing list by posting a bogus patch for
3679 DEFINE-COMPILER-MACRO..)
3680 * Compiled files containing forms which have side effects on the Lisp
3681 reader (such as DEFPACKAGE forms) are now handled more correctly.
3682 (Compiler queuing of top level lambdas has been suppressed by setting
3683 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3684 * deleted various currently-unused source files, e.g. gengc.lisp. They
3685 may be added back at some point e.g. when porting to other architectures,
3686 but until they are it's distracting to distribute them and to try to
3688 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3689 there were so many of them they're just distractions, and UNCROSS is
3690 known to be able to handle the current sources
3691 * moved PROFILE functionality into TRACE, so that it will be clear
3692 how the wrapping and unwrapping of functions when you profile them
3693 interacts with the wrapping and unwrapping of functions when you
3694 trace them. (Actually, the functionality isn't there yet, but at least
3695 the interface specification is there. Hopefully, the functionality will
3696 arrive with some future maintenance release.)
3697 * removed host-oops.lisp
3698 * changed signature of QUIT function to allow UNIX-CODE argument
3699 * fixed READ-SEQUENCE bug
3700 * tweaked verbose GC output so that it looks more like the progress
3701 output that ANSI specifies for functions like LOAD
3702 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3703 * added <http://sbcl.sourceforge.com> to the banner information printed by