1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
3 * bug fix: LISTEN sometimes returned T even in cases where no data was
4 immediately available from the stream
5 * fixed some bugs revealed by Paul Dietz' test suite:
6 ** REMOVE-METHOD returns its generic function argument even when
7 the method is not one of the generic functions' methods.
8 ** objects declared as MEMBER types can be admissible arguments to
10 ** printing characters should simply be printed by the FORMAT ~:C
13 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
14 * new platform: experimental support for SBCL x86/Darwin, including
15 MacOS X 10.4.5 on Intel.
16 * new feature: Unicode character names are now known to the system
17 (through CHAR-NAME and NAME-CHAR).
18 * new feature: the filesystem location of SBCL's core file is
19 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
20 * minor incompatible change: the contrib modules SB-POSIX and
21 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
22 this change is to make it easier to distribute
23 location-independent binaries.
24 * bug fix: as implied by AMOP, standardized classes no longer have
25 slots named by external symbols of public packages. (reported by
27 * optimization: calling functions via constant symbols -- as in
28 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
29 via the function object as in (FUNCALL #'FOO).
30 * optimization: on x86 and x86-64, the calling convention for the
31 case of dealing with an unknown number of values has been altered
32 to be friendlier to the prediction heuristics implemented,
33 particularly with reference to CALL and RET pairing. (thanks to
35 * optimization: on x86, the code for access to thread-local symbol
36 values and binding/unbinding of thread-local symbols is smaller.
37 (thanks to Alastair Bridgewater)
38 * enhancement: CONSTANTP is now able to determine constantness of
39 more complex forms, including calls to constant-foldable standardized
40 functions and some special forms beyond QUOTE.
41 * fixed bug: occasional GC crashes on Solaris/x86
42 * optimization: x86-64 supports stack allocation of results of simple
43 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
44 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
45 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
46 (thanks to James Knight)
47 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
48 by the standard (thanks to Stephen Compall)
50 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
51 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
52 be used for bundling the runtime and the core file into one
53 executable binary. This feature is not currently supported on all SBCL
54 platforms. (thanks to James Bielman and NIIMI Satoshi)
55 * new feature: a generational or ephemeral garbage collector is now
56 the default on the PowerPC platform (both Linux and Darwin). The
57 old Cheney (stop and copy) collector is a build-time option.
58 (thanks to Cyrus Harmon, after Raymond Toy)
59 * minor incompatible change: the method by which SBCL finds its
60 contributed modules has changed; it no longer relies on symbolic
61 links from an $SBCL_HOME/systems directory, but searches directly
62 in the subdirectories of $SBCL_HOME.
63 * enhancement: the dynamic heap size on the Linux/PPC platform is
64 markedly larger, even using the older Cheney garbage collector.
65 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
66 certain complicated string types. (reported by Gary King)
67 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
68 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
69 * fixed bug: DOCUMENTATION on structures no longer signals an
70 error if no structure type of the right name exists. (reported by
72 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
73 applicable non-standard primary or :AROUND method on
74 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
75 no longer cause unbound variable errors. (reported by Kevin Reid)
76 * fixed bug: metacircle resolution in cases where methods have slots
77 added before the slots from STANDARD-METHOD. (reported by Jean
79 * fixed bug: the Power PC assembly code for calling into lisp
80 sometimes computed its return address wrongly (depending on the
81 15th bit of the address, fixed at link time). This used to
82 manifest itself in a segmentation violation while building PCL.
83 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
84 many others over the years)
85 * fixed bug: ROOM no longer reports silly numbers for stack usage if
86 the stack pointer (treated as a signed integer) is negative.
87 (thanks to Peter van Eynde)
89 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
90 * new platform: experimental support for the Windows operating
91 system has been added. (thanks to Alastair Bridgewater)
92 * fixed several bugs in and robustified the PPC FFI (including
93 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
94 * bug fix: allow non-simple string symbol names (reported by Paul
96 * bug fix: interrupt handling on NetBSD (thanks to Richard M
98 * bug fix: saving a core corrupted callbacks on x86/x86-64
99 * bug fix: closed a loophole in metacircularity detection and
100 grounding in the PCL implementation of CLOS.
101 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
102 (reported by tomppa on #lisp)
103 * bug fix: clear the direction flag on Lisp -> C transitions, as
104 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
105 (reported by Andrej Grozin and Hendrik Maryns)
106 * optimization: major improvements to GC efficiency on GENCGC platforms
107 * optimization: faster implementation of EQUAL
108 * optimization: emit more efficient opcodes for some common
109 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
110 * optimization: several other minor code-generation improvements on x86-64
111 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
113 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
114 * minor incompatible change: (SETF CLASS-NAME) and (SETF
115 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
116 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
117 by ANSI as a generic function, and for consistency (SETF
118 GENERIC-FUNCTION-NAME) is treated likewise.
119 * fixed bug #233.b: make constraint propagation notice when a variable
120 value is changed after it is referenced but before it is used
121 * fixed bug #296: no more arbitrary behaviour from filenames with
122 odd characters as --load arguments. (reported by Adam Warner)
123 * fixed bug #390: :CHARACTER-SET pathname components now work as
124 expected. (reported by Tim Daly Jr)
125 * fixed bug #391: complicated :TYPE intersections in slot
126 definitions no longer cause an error in PCL internals.
127 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
128 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
129 returning the number of octets which would be written to the
130 file-stream. (thanks to Robert J. Macomber)
131 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
132 arguments to the internal stream functions. (thanks to David
134 * bug fix: stack exhaustion is now handled on the x86/SunOS
135 platform. (thanks to Daisuke Homma)
136 * bug fix: more accurate ROOM results on GENCGC platforms
137 * optimization: improved type inference for arithmetic-for
138 index variables in LOOP
139 * optimization: faster floating-point SQRT on x86-64
140 * fixed some bugs revealed by Paul Dietz' test suite:
141 ** DOCUMENTATION returns NIL instead of "" for method combinations
142 that don't have a docstring
144 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
145 * minor incompatible change: (SETF CLASS-NAME) and (SETF
146 GENERIC-FUNCTION-NAME) are no longer generic functions, and
147 therefore are not customizeable by user code (as seems to be at
148 least permitted and maybe required by AMOP). As a consolation,
149 however, the SBCL implementation of these functions now calls
150 REINITIALIZE-INSTANCE as specified by AMOP.
151 * bug fix: slot-definition documentation is propagated as per ANSI
152 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
153 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
154 argument, as do eslotd initialization methods. (from Pascal
155 Costanza's "Closer" project)
156 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
157 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
159 * bug fix: it is now possible to have more than one subclass of
160 STANDARD-GENERIC-FUNCTION without causing stack overflow.
161 (reported by Bruno Haible, Pascal Costanza and others)
162 * bug fix: the dependent update protocol now works for generic
163 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
165 * bug fix: condition-class instances corresponding to
166 DEFINE-CONDITION forms are now created eagerly. (reported by
167 Kalle Olavi Niemitalo on comp.lang.lisp)
168 * bug fix: floating point printing is more accurate in some
169 circumstances. (thanks to Simon Alexander)
170 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
171 merged with *DEFAULT-PATHNAME-DEFAULTS*.
172 * bug fix: callbacks on OS X now preserve stack-alignment.
173 (thanks to Cyrus Harmon)
174 * enhancement: the x86-64 disassembler is much better at
175 disassembling SSE instructions. (thanks to Lutz Euler)
176 * enhancement: CHECK-TYPE on a variable now allows type-inference
177 (thanks to Vincent Arkesteijn)
178 * optimization: improved performance of EUC-JP external format.
179 (thanks to NIIMI Satoshi)
180 * optimization: performance improvements to IO on file streams of
181 :ELEMENT-TYPE CHARACTER
182 * optimization: much faster memory allocation on x86-64
183 * optimization: faster garbage collections (latency and throughput)
185 * optimization: faster bignum subtraction and division on x86 and x86-64
186 (thanks to Lutz Euler)
187 * optimization: more accurate type-inference for loops that use a
188 floating point index variable or a negative step.
190 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
191 * bug fix: add a workaround to SBCL looping infinitely at startup on
192 Linux kernels with apparently buggy implementations of personality().
193 (thanks to Svein Ove Aas)
194 * bug fix: Unicode symbols are correctly printed in LDB backtraces
195 (thanks to David Lichteblau)
196 * bug fix: local bindings shadow global symbol macros for
197 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
198 * optimization: non-open coded uses of numeric comparison operators
199 (e.g. >) no longer cons when called with more than one parameter
200 on platforms supporting dynamic-extent allocation.
201 * enhancement: saving cores with foreign code loaded is now supported
202 on MIPS/Linux in addition to the previously supported platforms.
203 * bug fix: FIND-RESTART now tests for activity, not applicability when given
204 a restart object as identifier. (reported by Helmut Eller for CMUCL)
205 * bug fix: division by zero in sb-sprof when no samples were collected
206 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
207 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
208 (thanks to Alexander Kjeldaas)
209 * new feature: ignoring the return values of destructive functions
210 like NREVERSE signals a compile-time style-warning.
211 (thanks to Kevin Reid)
212 * new feature: the alignment of alien structure fields can be
213 explicitly specified. (thanks to Cyrus Harmon)
214 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
215 and dump core on SIGQUIT
217 ** incompatible change: threads do not inherit values of specials
218 from their parents (see manual)
219 ** bug fix: threads stacks belonging to dead threads are freed by the
220 next exiting thread, no need to gc to collect thread stacks anymore
221 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
222 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
223 ** bug fix: dynamic variable and thread start related gc lossage
224 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
226 * fixed some bugs revealed by Paul Dietz' test suite:
227 ** SUBTYPEP is slightly more accurate on heinously complicated
228 CONS types where some of the members have uncertain (in the
229 NIL, NIL sense) type relationships to each other.
231 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
232 no longer purified unless :PURIFY T is explicitly specified.
233 ** Non-purified cores are significantly smaller than before
235 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
236 * new feature: timers based on Zach Beane's excellent timer package
237 * added support for the following external formats: koi8-u,
238 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
239 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
240 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
241 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
242 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
243 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
244 * incompatible change: a threaded SBCL will no longer revert to
245 non-threaded mode on non-NPTL systems, but refuse to start entirely.
246 * bug fix: interrupts are disabled until startup is complete; no
247 more sigsegvs when receiving a signal to soon
248 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
250 * bug fix: add a workaround for the memory randomization features in
251 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
252 workaround will only be in effect on systems with the proc filesystem
254 * bug fix: printing objects of type HASH-TABLE signals a
255 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
257 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
258 have been running for over 50 days. (reported by Gilbert Baumann)
259 * bug fix: the logic for getting names of functions gets less
260 confused when confronded with alternate-metaclass
261 funcallable-instances. (reported by Cyrus Harmon)
262 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
263 consistent, even on internal alternate-metaclass objects.
264 * bug fix: flush closure information collected by physical
265 environment analysis prepass before the main pass. (bug reported
267 * bug fix: compiler pack phase does not modify a hashtable, which is
268 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
269 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
270 may be expanded. (reported by Alexander <malishev> on c.l.l)
271 * bug fix: no more style-warnings when compiling code using LAST
272 under high-SPEED low-SPACE settings. (reported by David Wallin
274 * fixed some incompatibilities between SBCL's MOP and the MOP
276 ** the METAOBJECT class is now implemented;
277 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
278 STANDARD-OBJECT, as required;
279 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
280 now compatible (as required by VALIDATE-SUPERCLASS); there
281 remains a consistency requirement over the presence or absence
282 of the FUNCTION class in the superclass list of finalized
283 classes; see the manual for more details;
284 ** the :AROUND method for COMPUTE-SLOTS on
285 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
286 requested slot ordering.
288 ** bug fix: parent thread now can be gc'ed even with a live
290 ** bug fix: binding a special with PROGV to no value is not
291 the same as that symbol not having been bound (thanks to
293 ** bug fix: binding specials is thread safe (thanks to
295 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
296 and SB-SPROF do not die with 'no handler for signal XX in
297 interrupt_handle_now(..)' anymore
298 ** bug fix: WITH-TIMEOUT works with multiple running threads
299 * fixed some bugs revealed by Paul Dietz' test suite:
300 ** ENSURE-GENERIC-FUNCTION should take a method class object for
301 the :method-class keyword argument.
303 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
304 * new port: the Solaris operating system on x86 processors is now
305 mostly supported, though some rough edges in the environment
306 remain. (thanks to Daisuke Homma)
307 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
308 space for its heap. (thanks to Thiemo Seufer)
309 * optimization: SBCL on MIPS platforms now supports dynamic-extent
310 closures. (thanks to Thiemo Seufer)
311 * minor incompatible change: eof selects abort in the debugger.
312 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
313 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
315 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
316 deprecated, and will go away in a future revision of SBCL.
317 * minor incompatible change: GC-ON and GC-OFF are no longer
318 implemented with a counter, it does not matter how many times gc
319 is switched on or off
320 * bug fix: discriminating functions for generic function classes
321 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
322 make invalid assumptions about method precedence order. (reported
324 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
325 platforms. (thanks to Thiemo Seufer)
326 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
327 * bug fix: correct pathname printing with printer escaping is on.
328 (thanks to Kevin Reid)
329 * bug fix: complex VOP definitions in "user-space" no longer trigger
330 package locks. (reported by Zach Beane)
331 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
332 for funcallable-instance functions, and is no different from
334 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
335 characters in the current readtable. (reported by Nicholas Neuss)
336 * bug fix: SUBTYPEP on various CONS types returns more a more
337 accurate acknowledgment of its certainty.
338 * optimizations: REMOVE-DUPLICATES now runs in linear time on
339 lists in some cases. This partially fixes bug 384.
340 * flush all standard streams before prompting in the REPL and the
342 * bug fix: signal handling and triggering gc do not conflict
343 directly anymore, in particular a high frequency sb-sprof does
344 not prevent gc from running
345 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
346 approximation for timezone and DST information between the
347 universal time epoch and the smallest negative 32-bit time_t.
348 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
349 year 1899 so long as the encoded time is non-negative.
350 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
351 the correct answer for non-constant rotate amounts when performing
352 an inline 32-bit rotation.
354 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
355 there is only one thread in the session
356 ** bug fix: memory leak for streams created in one thread and
357 written to in another
358 ** bug fix: lockup when compiled with gcc4
359 ** bug fix: race that allows the gc to be triggered when gc is
361 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
362 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
364 ** bug fix: finished threads are gc'ed properly
365 * fixed some bugs revealed by Paul Dietz' test suite:
366 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
367 has arguments with a different set of applicable methods from
368 the orignal arguments.
369 ** The type error thrown by MAP now has a correct expected-type
371 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
372 name a compiled function.
373 ** Three MISC tests where a large bignum was improperly coerced to
374 a float that couldn't represent that bignum during type
375 derivation were fixed.
376 ** SUBTYPEP can now handle types involving the negation of a
377 list-form FUNCTION type.
378 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
379 as (COMPLEX (AND RATIO (NOT FIXNUM))).
380 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
382 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
383 * New feature: Experimental support for bivalent streams: streams
384 opened with :element-type :default now allow character and binary
385 (unsigned-byte 8) I/O
386 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
387 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
389 * bug fix: no more highly sporadic "couldn't check whether ~S is
390 readable" when reading a stream and an interrupt hits in the middle
391 of a select system call
392 * compiler better recognizes complex arrays (reported by Eduardo
394 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
395 for variables on SBCL built with linkage-tables. (reported by Luis
397 * various error reporting improvements.
398 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
399 (Thanks to Thiemo Seufer)
400 * bug fix: nested reader invokations work correctly
401 * bug fix: it is possible to have simultaneous references to foreign
402 code and foreign data with the same name.
404 ** added x86-64 support
405 ** incompatible change: the threading api now works with thread
406 objects instead of thread ids
407 ** bug fix: threads are protected from signals and interruption when
408 starting up or going down
409 ** bug fix: a race where an exiting thread could lose its stack to gc
410 ** bug fix: don't halt on infinite error in threads if possible
411 ** fixed numerous gc deadlocks introduced in the pthread merge
412 ** bug fix: fixed thread safety issues in read and print
413 ** bug fix: debugger doesn't hang on session lock if interrupted at
414 an inappropriate moment
415 ** bug fix: run-program is now thread safe(r)
416 ** bug fix: inner with-recursive-lock no longer releases the mutex
417 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
418 * fixed some bugs revealed by Paul Dietz' test suite:
419 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
420 EXPECTED-TYPE slots filled.
421 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
423 ** MAKE-CONDITION accepts classes as type-designators.
424 ** COMPILE may never return NIL.
425 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
426 range before calling Unix time functions
428 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
429 * numerous signal handling fixes to increase stability
430 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
431 * minor incompatible change: we now correctly canonize default
432 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
433 as per the MOP, rather than the historical (INITARG INITFUNCTION
434 INITFORM). (reported by Bruno Haible)
435 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
436 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
437 for more information.
438 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
439 pathname is a directory pathname.
440 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
441 * dynamic space size on PPC has been increased to 768Mb. (thanks to
443 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
444 :METACLASS argument in addition to a class name. (reported by
445 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
446 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
447 (reported by Bruno Haible; patch from Gerd Moellmann)
448 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
450 * bug fix: more cleanups to the floating point exception handling on
451 x86-64 (thanks to James Knight)
452 * bug fix: the compiler does not try to constant fold calls of
453 COERCE and BIT-* functions when they return freshly constructed
454 objects. (reported by Nikodemus Siivola and Paul Dietz)
455 * optimization: improved the MIPS versions of generic (in the
456 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
457 * optimization: direct conversion of (unsigned-byte 32) to floats on
458 the PowerPC platform.
459 * optimization: structure instances with raw slots now use less
460 memory, and probably show better memory locality. (thanks to
462 * optimization: DYNAMIC-EXTENT declarations for lists and closures
463 are treated as requests for stack allocation on the x86-64,
464 Alpha-32, PPC and SPARC platforms.
465 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
466 the wrong file descriptor; implementation of SOCKET-OPEN-P.
467 (thanks to Tony Martinez)
469 ** gcing a dead thread can no longer lead to lockups
470 ** threads block signals until they are set up properly
471 ** errno is no longer shared by threads
472 ** interrupt-thread restores the eflags register on x86
473 ** fixed some lockups due to gc/thread interaction
474 * fixed some bugs revealed by Paul Dietz' test suite:
475 ** invalid dotted lists no longer raise a read error when
477 ** signal an error if a symbol that names a declaration is used
478 as the name of a type, or vice versa
479 ** allow using the (declare (typespec var*)) abbreviation for
480 (declare (type typespec var*)) with all type specifiers
481 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
482 ** always return NIL from PPRINT-POP when OBJECT is NIL
483 ** don't signal errors when pretty-printing malformed LABELS,
484 FLET or MACROLET forms
485 ** declarations in a DOLIST body are also in scope for the
487 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
489 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
492 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
493 * fixed cross-compiler leakages that prevented building a 32-bit
494 target with a 64-bit host compiler.
495 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
496 opened with :IF-EXISTS :OVERWRITE.
497 * fixed bug 281: error for an invalid qualifier in a short-form method
498 combination method is not signalled until the faulty method is called.
499 * bug fix: iteration variable type inferrer failed to deal with open
500 intervals. (reported by Alan Shields)
501 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
502 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
503 or more runtime options were provided to the sbcl binary.
504 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
505 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
506 * optimization: there's now a fast-path for fixnum arguments in the
507 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
508 * optimization: the code generated on x86-64 is more compact thanks
509 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
510 * bug fix: floating-point exception handling now partly works on
511 x86-64. (Thanks to James Knight)
512 * improvement to the MIPS backend from Thiemo Seufer: C-style
513 64-bit long long arguments and return values to alien functions
515 * fixed some bugs revealed by Paul Dietz' test suite:
516 ** the type-error signalled from WARN has a filled-in DATUM slot.
517 ** the type-error required when a stream is not associated with
518 a file has the stream as its datum.
519 ** type-errors on single-floats on x86-64 no longer have
520 :INVALID-OBJECT as the datum
521 ** the type-errors signalled for invalid function names now have
522 a correct expected type
523 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
524 for structures defined with a :TYPE. Documentation strings for
525 typed structures are no longer immediately discarded
526 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
527 broadcast streams and synonym streams. FILE-LENGTH now also works
528 on broadcast streams.
530 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
531 * incompatible change: the --noprogrammer option, deprecated since
532 version 0.7.5, has been removed. Please use the equivalent
533 --disable-debugger option instead.
534 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
536 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
537 inoperational for a while now) has been completely removed.
538 * null lexical environments are now printed as #<NULL-LEXENV>,
539 significantly reducing the amount of clutter in typical backtraces.
540 * documentation on weak pointers, finalization, and after GC hooks
541 has been added to the manual.
542 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
543 as other specialized array types, is much faster. SUBSEQ and
544 COPY-SEQ on such arrays have also been sped up.
545 * optimization: EQL is now more efficient when at least other argument
546 is known to be of type (OR FIXNUM (NOT NUMBER)).
547 * fixed bug: the runtime is now less vulnerable to changes in the
548 size of the SBCL object on OS X, and virtual memory is reserved for
549 all spaces that need to be at a fixed address.
550 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
551 * fixed bug: finalizers and after GC hooks that cause consing are now
553 * fixed bug: compiler error messages and summaries are now printed to
554 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
555 * fixed inference of the upper bound of an iteration variable.
556 (reported by Rajat Datta).
557 * fixed bug 211e: calling local functions with duplicated constant
558 keyword argument no longer causes a bogus style warning about an
560 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
561 ftype declaration to be disregarded. (reported by Dave Roberts)
562 * fixed bug 373: caused by erronous compilation of references to alien
563 variables in the runtime on ppc/darwin.
564 * fixed bug 376: CONJUGATE type deriver.
565 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
566 (reported by Baughn on #lisp)
567 * a cross-compiler bug on non-x86 platforms has been identified and
568 fixed. (thanks to Bruno Haible)
569 * improvements to the MIPS runtime code for increased stability.
570 (thanks to Thiemo Seufer)
571 * increased the maximimum compact environment size to allow
572 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
573 * partial workaround for bug 135: don't name the function we're
574 calling for hairy cases of EVAL, so as not to accumulate one environment
575 entry per call to EVAL. (reported by Kevin Reid)
576 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
577 * optimization: type testing for non-vector arrays should be faster.
578 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
580 * bug fix: setting 31st element of a bit vector to zero did not work
582 * bug fix: redefining a class definition which failed due to a
583 previous accessor / function clash now works (but see BUGS entry
584 #380 for more problems in this area). (thanks to Zach Beane)
585 * the long form of DEFINE-METHOD-COMBINATION disables method group
586 checking when given a single method group with pattern *. (thanks
588 * on x86 compiler supports stack allocation of results of simple
589 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
590 * support for building and running on Mac OS X 10.4 "Tiger" has been added
591 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
592 * fixed some bugs related to Unicode integration:
593 ** the restarts for recovering from input and output encoding
594 errors only appear when there is in fact such an error to
596 * fixed some bugs revealed by Paul Dietz' test suite:
597 ** MISC.549 and similar: late transformation of unsafe type
598 assertions into derived types caused unexpected code
600 ** SCALE-FLOAT type deriver is less wrong.
601 ** type derivers for EXP, LOG and similar functions compute result
602 types for complex arguments better.
603 ** (MISC.563) CONJUGATE type deriver works for very restricted
605 ** out-of-line type testers for character strings are available.
606 ** EQUAL compiler transform understands specialness of objects
608 ** accessing double-floats stored on the stack now works on x86-64.
609 ** debugger internals could sometimes create invalid lispobjs,
610 resulting in GC crashes.
611 ** MISC.548: type check weakening can convert required type into
613 ** initialization forms for bindings are not in scope of free special
616 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
617 * incompatible change: thread support for non-NPTL systems has
618 been removed - locking is buggy and unreliable. A threaded
619 SBCL build will now warn at startup and refuse to create
620 new threads, unless futex support is detected in the kernel
621 * incompatible change: the top level REPL now has only an ABORT
622 restart associated with it, not TOPLEVEL and ABORT as it used to.
623 TOP and TOPLEVEL are now available as debugger commands for
624 returning to the top level.
625 * incompatible change: forms evaluated in the REPL now use the
626 global optimization policy.
627 * incompatible change: user- and system-initialization files are
628 no longer processed with LOAD, but by READ and EVAL; hence the
629 global optimization policy, startup package, readtable, etc,
631 * The .fasl file format number has been incremented because of
632 various incompatible changes.
633 * internal entry point details and argument counts no longer appear
634 in backtraces unless explicitly requested by setting
635 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
636 * built-in and standard functions no longer have names like "top
637 level local call to FOO".
638 * fixed bug 32: functions defined in non-null lexical environments
639 now have more legible printed representation
640 * fixed bug 33: functions defined in non-null lexical environemnts
641 are now more amenable to inspection by INSPECT.
642 * workaround for bug 354: XEPs no longer appear in backtraces unless
643 explicitly requested.
644 * fixed bug: receiving the signal which results from attempting to
645 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
646 SIGBUS on other *BSDs) on architectures where the C stack is also
647 the Lisp stack (x86 and x86-64) from foreign code no longer leads
648 to debugger confusion or wild instability. (reported by Cheuksan
650 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
651 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
652 (reported by Lutz Euler)
653 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
654 compiled code. (reported by Paul Dietz)
655 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
656 specialized parameter to a method to something that is not TYPEP
657 the specializer is now possible.
658 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
659 face of package deletion.
660 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
661 pathname. (reported by David Tolpin on c.l.l)
662 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
663 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
664 STRING-STREAM and FILE-STREAM, respectively.
665 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
666 than previously; in addition, its test suite is now run on build.
667 (thanks to Luke Gorrie)
668 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
669 * optimization: added a immediate representation for single-floats
671 * fixed some bugs revealed by Paul Dietz' test suite:
672 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
673 ** attempting to create a package with a colliding nickname causes
674 correctable errors to be signalled.
675 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
676 ** several x86-64 backend bugs related to sign-extension of immediate
679 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
680 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
681 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
683 * fixed loading of multiply forward-referenced layouts.
684 (thanks to Cheuksan Wang)
685 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
686 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
687 (see COMPILE-FILE documentation for details of :PRINT option)
688 * fixed bugs 19 and 317: fixed-format floating point printing is
689 more accurate. This also fixes a bug reported by Adam Warner
690 related to the ~@F format directive.
691 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
693 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
694 dispatch table if the from-char is a dispatch macro character.
695 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
696 by the wordsize no longer ignore the last word. (reported by Lutz
698 * fixed bug in type checking of dynamic-extent variables. (reported
700 * optimization: sequence traversal functions use their freedom to
701 coerce function designators to functions.
702 * optimization: code with many calls to CLOS methods specialized on
703 CLOS classes has had redundant type checks removed.
704 * optimization: added declarations to speed up operations that access
705 the internal character database (for example STRING-UPCASE)
706 * optimization: comparison operations between floats and sufficiently small
707 fixnums no longer create extra rationals
708 * fixed some bugs related to Unicode integration:
709 ** portions of multibyte characters at the end of buffers for
710 character-based file input are correctly transferred to the
711 start of the buffer at the next read.
712 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
713 passing it through to OPEN.
714 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
715 argument given to internal calls to OPEN.
716 ** the built-in comment readers (introduced by character sequences
717 ";" and "#|") are more forgiving to encoding errors; they will
718 STYLE-WARN and then attempt to resync the stream at a character
719 boundary. (thanks to Teemu Kalvas)
720 * fixed some bugs in the x86-64 port:
721 ** Negative short int return values from c-calls are sign-extended
723 ** The stack is aligned to 16-bytes for c-calls, as required by
724 the ABI. (thanks to Cheuksan Wang)
725 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
726 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
727 for enums. (thanks to Vincent Arkesteijn)
728 ** Multiple small optimizations and bugfixes for floating point
730 * fixed some bugs revealed by Paul Dietz' test suite:
731 ** Space, Tab, Linefeed, Return and Page have the invalid
732 secondary constituent character trait.
733 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
735 ** WITH-INPUT-FROM-STRING should only update the index place on
737 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
739 ** Bit-array operations (BIT-AND and similar) worked incorrectly
740 with one-dimensional arrays with fill pointers.
741 ** TYPE-OF failed on a complex with an integer realpart and a
743 ** compiler failure during type inference for the code of form
744 (IF (EQL X (THE ...)) ...) (MISC.535).
746 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
747 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
748 platform. The port supports 61-bit fixnums, large memory spaces
749 and reloading shared object files.
750 * enhancement: saving cores with foreign code loaded is now
751 supported on ppc/Darwin in addition to the previously supported
753 * enhancement: the statistical profiler now walks deeper into the
754 call stack for more meaningful call-graphs and accrued time
755 reports (x86/x86-64 only). It also now reports time spent in
757 * enhancement: it is now possible to trace most individual methods
758 of a generic function in addition to tracing the generic function
760 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
762 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
763 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
764 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
765 produces an error. (thanks to Vincent Arkesteijn)
766 * bug fix: NAMESTRING on pathnames with :WILD components in their
767 directories works correctly. (thanks to Artem V. Andreev)
768 * fixed bug 125: compiler preserves identity of closures. (reported
770 * bug fixed: functions with &REST arguments sometimes failed with
771 "Undefined function" when compiled with (DEBUG 3). (reported by
773 * bug fix: overflow during compiling of setting element of a bit
774 vector with constant index and value. (reported by Timmy Douglas)
775 * build fix: fixed the dependence on *LOAD-PATHNAME* and
776 *COMPILE-FILE-PATHNAME* being absolute pathnames.
777 * on x86 compiler partially supports stack allocation of dynamic-extent
779 * GO and RETURN-FROM do not check the extent of their exit points
780 when compiled with SAFETY 0.
781 * fixed some bugs related to Unicode integration:
782 ** encoding and decoding errors are now much more robustly
783 handled; it should now be possible to recover even from invalid
784 input or output to the terminal. (thanks to Teemu Kalvas)
785 ** provided a first cut at implementing STRING-TO-OCTETS and
786 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
787 ** altered the SB-MD5 contributed module slightly, changing the
788 interface just enough for it to be supportable for builds where
789 lisp characters are not eight bits.
790 * fixed some bugs revealed by Paul Dietz' test suite:
791 ** the FORMATTER-generated functions for ~V[ conditionals require
792 the correct number of arguments.
793 ** READ-FROM-STRING returns the mandated second value when applied
794 to displaced strings.
795 ** the #\Rubout and #\Backspace characters are treated as invalid
796 constituent characters by the tokenizer.
798 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
799 * new feature: reloading changed shared object files with
800 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
801 * new feature: references to foreign variables and functions
802 can now be compiled and loaded before the corresponding shared
803 object file is loaded, as long as the foreign definitions are
804 available at runtime.
805 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
806 supported. (thanks to Dan Debertin)
807 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
808 just keywords. (thanks to Vincent Arkesteijn)
809 * fixed bug #331: structure-class instances corresponding to
810 DEFSTRUCT forms are now created eagerly.
811 * fixed bug #345: backtraces from calls to undefined functions work
812 on x86 as well. Related bug #61 is now also partially fixed on x86
813 (backtraces from throws to unknown catch tags.)
814 * bug fix: lambda-list parsing is now stricter vrt. order and number
815 of lambda-list keywords.
816 * bug fix: as specified by AMOP, an error is signalled if a
817 class-option appears multiple times in a DEFCLASS form. (reported
819 * bug fix: RANDOM can be compiled when the compiler derives the type
820 of its numeric argument as a disjoint set of small integers.
821 (reported by Paul Dietz)
822 * bug fix: starting a core saved with shared objects loaded when
823 those objects are not available no longer causes threaded SBCL to
824 hang. (reported by Sean Ross)
825 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
827 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
828 argument, which must always be NIL. (reported by Kalle Niemitalo)
829 * bug fix: printing 1.0d+23 no longer results in an error.
830 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
831 * bug fix: structure slot setters preserve evaluation order. (thanks
833 * fixed some bugs related to Unicode integration:
834 ** RUN-PROGRAM can allow its child to take input from a Lisp
835 stream. (reported by Stefan Scholl)
836 ** ASDF-INSTALL successfully downloads .tar.gz files in all
837 locales. (reported by Ken Causey)
838 * fixed some bugs revealed by Paul Dietz' test suite:
839 ** INCF, DECF and REMF evaluate their place form as specified in
841 ** FORMATTER expands ~{ ~} iteration directives with V or #
842 parameters correctly.
843 ** FORMATTER deals with the ~@[ ~] conditional directive where the
844 consequent uses no arguments correctly.
845 ** the system has a partial understanding of the (COMPLEX RATIO)
847 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
850 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
851 * new feature: a build-time option (controlled by the :SB-UNICODE
852 keyword feature, enabled by default) for building the system with
853 support for the entire 21-bit character space defined by the
855 * new feature: the system now has rudimentary external-format
856 support; the primary user-visible change at this time is that
857 characters with the high bit set (such as lower-case-e-acute) will
858 print correctly to a terminal in a UTF-8 environment.
859 * minor incompatible change: BASE-CHAR no longer names a class;
860 however, CHARACTER continues to do so, as required by ANSI.
861 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
862 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
863 has been moved to the SB-EXT package (temporarily re-exported from
865 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
866 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
868 * fixed bug #351: better error-handlind and reporting for malformed
870 * fixed bug #350: bignum-printing is now more memory-efficient,
871 allowing printing of very large bignums, eg. (expt 2 10000000).
872 (reported by Bruno Haible)
873 * fixed bug #302: better primitive-type selection for intersection
875 * fixed bug #308: non-graphic characters now all have names, as
876 required. (reported by Bruno Haible)
877 * bug fix: redefining a class with different superclasses now correctly
878 removes it from the direct-subclasses of its previous superclasses.
879 (reported by David Morse)
880 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
881 new value now works. (reported by Bruno Haible)
882 * bug fix: correct canonicalization of multiple non-standard slot
883 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
884 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
885 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
886 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
887 now exists, an signals an error.
888 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
889 during macroexpansion for non-iteration variables as well. (reported
890 by Bruno Haible for CMUCL)
891 * bug fix: Cyclic structures and unprintable objects in compiler
892 messages no longer cause errors. (reported by Bruno Haible)
893 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
894 and READ-FROM-STRING all now return a primary value of NIL if
895 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
896 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
897 (reported by Bruno Haible for CMUCL)
898 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
899 do not designate a condition. (reported by Bruno Haible for
901 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
902 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
903 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
904 permitting resolution in favour of any of the conflicting symbols.
905 (reported by Bruno Haible for CMUCL)
906 * bug fix: EQUAL compiler optimizations is less aggressive on
907 strings which can potentially compare true despite having distinct
908 specialized array element types.
909 * bug fix: unit enumerations can be defined without dividing by
910 zero. (thanks to Vincent Arkesteijn)
911 * FORMAT compile-time argument count checking has been enhanced.
912 (report from Bruno Haible for CMUCL)
913 * a partial workaround for the bug 262: the compiler does not try to
914 inline-expand a local function doing RETURN-FROM from a deleted
915 BLOCK. (thanks to Peter Denno for the bug report and to David
916 Wragg for the simple test case)
917 * fixed some bugs revealed by Paul Dietz' test suite:
918 ** RENAME-PACKAGE allows all package designators as new package
920 ** constraint propagation and lambda variable substitution are
921 more cautious in dealing with partially deleted code.
922 ** compiler sometimes lost reoptimization passes.
923 ** CERROR, given a condition as condition designator, passes its
924 remaining arguments to the continue format control without
926 ** Case-altering FORMAT directives work correctly on non-ASCII
928 ** The REAL type specifier handles bounds outside the floating
929 point ranges without signalling FLOATING-POINT-OVERFLOW.
930 ** Functions with IR1-transformations can create intercomponent
931 references to global functions.
932 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
934 ** FORMAT ~R treats a nil value for its first parameter correctly.
936 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
937 * enhancement: saving cores with foreign code loaded is now
938 supported on x86/NetBSD and sparc/Linux in addition to the previously
940 * bug fix: on some platforms repeated installations caused multiple
941 copies of HTML documentation to be installed -- should not happen
942 any more. (reported by Stefan Scholl)
943 * bug fix: parsing self-recursive alien record types multiple times
944 no longer causes infinite recursion. (reported by Thomas F. Burdick,
945 original patch by Helmut Eller for CMUCL)
946 * bug fix: stack-exhaustion detection works now on NetBSD as well.
947 (thanks to Richard Kreuter)
948 * bug fix: defining classes whose accessors are methods on existing
949 generic functions in other (locked) packages no longer signals
950 bogus package lock violations. (reported by François-René Rideau)
951 * bug fix: special variables as DEFMETHOD parameters no longer have
952 associated bogus type declarations. (reported by David Wragg and
954 * bug fix: read-write consistency on streams of element-type
955 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
956 * bug fix: redefiniton of the only method of a generic function with
957 no DEFGENERIC no longer emits a full WARNING. In addition,
958 redefinition of generic functions with no DEFGENERIC to an
959 incompatible lambda list now signals an error. (thanks to Zach
961 * bug fix: DEFGENERIC now works even when there's a function of the
962 same name in an enclosing lexical environment. (thanks to Zach
964 * fixed compiler failure, caused by instrumenting code during
965 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
966 * optimization: added loop analysis and improved register allocation
967 * fixed some bugs revealed by Paul Dietz' test suite:
968 ** POSITION on displaced vectors with non-zero displacement
969 returns the right answer.
970 ** (SIMPLE-STRING) is a valid type specifier for sequence
972 ** *PRINT-LEVEL* handling for slotless structures is pedantically
974 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
975 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
977 ** The justification version of the FORMAT ~< directive treats
978 non-zero minpad parameter correctly.
980 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
981 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
982 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
983 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
984 the supported interface.
985 * new feature: Single-stepping of code compiled with DEBUG 2 or
986 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
987 * new feature: saving cores with foreign code loaded is now
988 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
989 Timothy Moore's work for CMUCL)
990 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
991 parameters to * instead of NIL if no initform is supplied.
992 (reported by Johan Bockgaard)
993 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
994 correctly when FUNCALL appears as the car of the form. Note:
995 despite this FUNCALL forms are not currently subject to
996 compiler-macro expansion. (port of Raymond Toy's fix for the
997 same from CMUCL, reported by Johan Bockgaard)
998 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
999 (thanks for Teemu Kalvas)
1000 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1001 case instead of terminating the iteration (thanks for Julian
1002 Squires, Sean Champ and Raymond Toy)
1003 * bug fix: incorrect expansion of defgeneric that caused a style
1004 warning. (thanks for Zach Beane)
1005 * bug fix: slot accessor effective method computation works properly
1006 for classes with multiple non-standard applicable methods on
1007 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1008 * on x86 compiler supports stack allocation of results of LIST and
1009 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1010 CMUCL implementation by Gerd Moellmann)
1011 * fixed some bugs revealed by Paul Dietz' test suite:
1012 ** FORMAT strings with both the ~<~:;~> form of the justification
1013 directive and pretty-printing directives cause an error.
1015 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1016 * incompatible change: the internal functions
1017 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1018 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1019 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1020 instead of the old functions.
1021 * new feature: on platforms where "dladdr" is available foreign
1022 function names now appear in backtraces. (based on Helmut Eller's
1024 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1025 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1027 * documentation: ASDF manual is now distributed with SBCL.
1028 * bug fix: SBCL can now load its contributed modules with REQUIRE
1029 even if the system-provided entries have been removed from
1030 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1032 * bug fix: backtraces involving undefined functions or assembly
1033 routines are more informative. (thanks to Brian Downing)
1034 * bug fix: mutually referent alien structures now work correctly.
1035 (reported by Rick Taube)
1036 * bug fix: structures defined by WITH-ALIEN can be referred to
1037 within other definitions in the same WITH-ALIEN.
1038 * bug fix: division operators (MOD, TRUNCATE and the like) with
1039 constant zero divisors and integer dividends no longer generate
1041 * bug fix: provide default methods for INPUT-STREAM-P and
1042 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1043 * bug fix: improve the detection and resolution of MOP metacycles
1044 (where effective method computation depends on the generic function
1045 having its effective method computed). (reported by Bruno Haible)
1046 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1047 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1048 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1049 * bug fix: pathnames with relative directory components can be
1050 represented relative to default pathnames.
1051 * optimization: in taking the GCD of bignums, reduce the two bignums
1052 to approximately the same size (using Euclid's algorithm) before
1053 applying the more sophisticated binary GCD. (thanks to Juho
1055 * optimization: COUNT on bitvectors now operates word-at-a-time.
1056 * optimization: ASH with a positive, but not necessarily constant,
1057 (leftwards) shift, when in a modular context, is compiled to a
1059 * fixed some bugs revealed by Paul Dietz' test suite:
1060 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1061 if the corresponding argument is NIL.
1062 ** FORMAT directives accepting a minpad parameter treat negative
1064 ** literal commas as character parameters to FORMAT directives are
1066 ** literal spaces directly after ~<Newline> directives within a
1067 format-logical-block (~:< ~@:>) do not induce :FILL-style
1068 conditional newlines.
1069 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1070 as specified: it no longer includes conditional newlines).
1071 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1073 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1074 :PER-LINE-PREFIX argument does not evaluate to a string.
1076 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1077 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1078 the manual for details; for now, package locks can be disabled by
1079 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1080 if no major problems are found then it is likely that they will be
1081 compiled in unconditionally.
1082 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1083 unsupported operators on all platforms. To load a shared library
1084 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1085 object file, link it into a shared library outside of SBCL and
1086 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1088 * bug workaround: ROOM T and the GC were not working together
1089 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1090 reporting facility aren't preserved. That reporting has been
1091 disabled, so now until and unless someone figures out how to make it
1092 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1093 * minor incompatible change: as threatened around sbcl-0.8.0, the
1094 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1095 The symbols are also exported from SB-PCL for backwards
1096 compatibility, but more so than before SB-PCL should be treated as
1097 an implementation-internal package.
1098 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1100 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1101 It is known as of this release not to work on the Alpha, however.
1102 * fixed bug #167: errors signalled due to illegal syntax in method
1103 bodies are now more legible.
1104 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1105 designators and can hence be used with TYPEP.
1106 * fixed bug #333: CHECK-TYPE now ensures that the type error
1107 signalled, if any, has the right object to be accessed by
1108 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1109 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1110 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1112 * fixed bug #334: programmatic addition of slots using specialized
1113 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1114 and :ALLOCATION :CLASS slots.
1115 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1116 float exponents. (rereported by Peter Seibel)
1117 * fixed a bug: #\Space (and other whitespace characters) are no
1118 longer considered to be macro characters in standard syntax by
1119 GET-MACRO-CHARACTER.
1120 * fixed bug: initialization of condition class metaobjects no longer
1121 causes an instance of the condition to be created. (reported by
1123 * fixed bug: it is now possible to have slots such that
1124 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1125 description is neither :INSTANCE nor :CLASS.
1126 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1127 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1128 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1129 non-standard SLOT-DEFINITION-ALLOCATION values.
1130 * fixed bugs: various race conditions handling exiting threads.
1131 CL-PPCRE's thread torture test now passes.
1132 * fixed arguably-a-bug: GC time proportional to number of threads in
1133 system even when most of them are idle
1134 * optimization: improved performance of BIT and SBIT on bit-vectors.
1135 * .fasl file incompatibility: The fasl file version number has
1136 been incremented because of changes associated with package locks.
1138 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1139 * minor incompatible change: the system no longer provides
1140 optimization hints (conditions which are TYPEP
1141 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1142 mode; these hints are still emitted when higher SPEED optimization
1144 * new contrib module: a sampling profiler (profiling by statistical
1145 sampling, rather than by instrumenting functions) is available as
1146 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1147 * the behaviour of the standard function ED is now customizeable by
1148 third parties through a hook variable: see ED's documentation
1149 string for information on the protocol.
1150 * the compiler no longer emits efficiency notes for (FUNCALL X)
1151 when the type of X is uncertain under default optimization
1153 * fixed bug 276: mutating a binding of a specialized parameter to a
1154 method to something that is not TYPEP the specializer is now
1156 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1157 exhibit the required behaviour on the x86 platform. (thanks to
1158 Peter van Eynde, Eric Marsden and Bruno Haible)
1159 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1160 even for difficult arguments. (reported by Peter Graves)
1161 * fixed bug 141a: the backquote printer now descends quoted
1163 * fixed another bug in backquote printing: no more destructive
1164 modification of the form's list structure. (reported by Brian
1166 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1167 move between its address being taken and the call to
1168 interrupt_thread, fixing a crashing race condition.
1169 * the SB-POSIX contrib implementation has been adjusted so that it
1170 no longer exhibits ridiculously poor performance when constructing
1171 instances corresponding to C structs.
1173 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1174 * minor incompatible change: the sb-grovel contrib now treats C
1175 structures as alien (in the sense of SB-ALIEN) objects rather than
1176 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1177 has implications for memory management of client code
1178 (sb-grovel-returned objects must now be manually managed) and for
1179 type safety (alien objects now have full types).
1180 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1181 used to control emission of compiler diagnostics, rather than the
1182 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1183 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1184 quality should be considered deprecated.
1185 * install.sh now installs the user manual as well
1186 * (not quite a new documentable feature, but worth considering in
1187 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1188 beginnings of a semantically meaningful condition hierarchy is
1189 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1191 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1192 designator as the defaults argument.
1193 * fixed bug: Displaced arrays whose displaced-to array has become
1194 too small now cause ARRAY-DIMENSION to signal an error, providing
1195 for safer bounds-checking. (reported by Bruno Haible)
1196 * fixed bug: DEFCLASS slot definitions with identical :READER and
1197 :WRITER names now signal a reasonable error. (reported by Thomas
1199 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1201 * fixed bug: Invalid slot specification errors now print correctly.
1202 (thanks to Zach Beane)
1203 * fixed bug 320: Shared to local slot value transfers in class
1204 redefinitions now happen corrently from superclasses as
1205 well. (reported by Bruno Haible)
1206 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1208 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1209 improper lists correctly. (reported by Bruno Haible)
1210 * fixed bug 313: source-transform for <fun-name> was erroneously
1211 applied to a call of a value of a variable with name <fun-name>.
1212 (reported by Antonio Menezes Leitao)
1213 * fixed bug 307: The obsolete instance protocol ensures that
1214 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1215 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1216 remove dead unknown-values globs from the middle of the stack.
1217 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1218 easier to resume long computations after using *BREAK-ON-SIGNALS*
1219 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1220 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1221 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1222 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1223 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1224 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1226 * fixed a (fairly theoretical) bug in string printing: if
1227 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1228 does not have array-element-type equal to the most general string
1230 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1231 function name. (thanks to Nikodemus Siivola)
1232 * fixed bug: compile-time format string checker failed on
1233 non-closed ~{. (reported by Thomas F Burdick)
1234 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1235 don't warn on legal loop code involving more than one aggregate
1236 boolean. (thanks to Nikodemus Siivola)
1237 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1238 in keyword position to MAKE-PACKAGE was not being reported as
1240 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1241 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1243 * fixed bugs 280 and 312: the checking for multiple definitions in a
1244 file is less likely to become confused by uses of inline
1246 * fixed bug: the #S reader macro performs the keyword coercion
1247 specified for slot names. (reported by Kalle Niemitalo)
1248 * fixed bug: lambda lists may contain symbols whose names start with
1249 & but are not lambda-list-keywords; their occurrence triggers a
1251 * fixed bug 321: define-method-combination argument lambda lists do
1252 not cause invalid code to be generated when &optional or &aux
1253 variables have default values. (reported by Bruno Haible)
1254 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1255 have CLOS classes; this appears also to have fixed reported
1256 instability in stack exhaustion detection.
1257 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1258 MUFFLE-WARNING when no associated restart is present is now
1260 * optimization: rearranged the expansion of various defining macros
1261 so that each expands into only one top-level form in a
1262 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1264 * optimization: used a previously little-used slot in symbols to
1265 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1267 * fixed some bugs revealed by Paul Dietz' test suite:
1268 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1270 ** Fixed an optimization bug related to inheritance of initforms
1271 from local to shared slots.
1272 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1273 ** CAST optimizer forgot to flush argument derived type.
1274 ** print/read consistency on floats is now orders of magnitude
1275 more likely. (thanks also to Bruno Haible for a similar report
1277 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1278 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1279 for objects of type REAL. Make it so.
1280 ** SXHASH is specified (infelicitously) to respect similarity,
1281 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1282 it so. (thanks to Markus Ziegler)
1283 ** on the Alpha, the compiler succeeds in compiling functions
1284 returning a known number of arguments greater than 63.
1285 ** fixed handling of invalid NIL arguments in keyword position
1287 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1289 ** provided a readably-printable representation for RANDOM-STATE
1291 ** ensured that pathnames, where they have a namestring, always
1292 print using #P"..." syntax.
1294 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1295 * Support for the forthcoming 2.0 version of the NetBSD kernel
1296 running on x86 hardware has been added. (thanks to Perry
1297 E. Metzger most immediately, and others for their past work)
1298 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1299 current release version; SBCL's OpenBSD support had been broken
1300 since about the time of OpenBSD's switch to ELF binary format.)
1301 * [placeholder for DX summary]
1302 ** user code with &REST lists declared dynamic-extent, under high
1303 speed or space and low safety and debug optimization policy.
1304 * The manual has been converted to Texinfo format and the debugger
1305 chapter from the cmucl manual has been added.
1306 * A facility has been added to extract documentation strings from
1307 sbcl and store them as Texinfo-formatted snippets for inclusion in
1308 the manual (via Texinfo's @include directive)
1309 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1310 the test case to Dave Roberts)
1311 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1312 pointers. (reported by Sean Ross)
1313 * bug fix: PROFILE output is printed nicely even for large numerical
1314 values. (thanks to Zach Beane)
1315 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1316 greater than 32 handle EOF correctly.
1317 * bug fix: on X86 an immediate argument of the IMUL instruction is
1318 correctly printed in disassembly. (thanks to Lutz Euler)
1319 * bug fix: class slots in redefined classes preserve their old
1320 values. (thanks to Bruno Haible and Nikodemus Siivola)
1321 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1322 succeeds. (reported by Marco Baringer)
1323 * fixed some bugs revealed by Paul Dietz' test suite:
1324 ** READ-SEQUENCE now works on ECHO-STREAMs.
1325 ** RATIONALIZE works more according to its specification. (thanks
1328 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1329 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1330 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1331 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1332 should matter to you only if you rebind the printer control
1333 variables and then find you want different bindings in the
1334 debugger than in the ordinary execution of your program.)
1335 * The runtime build system has been tweaked to support building
1336 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1337 assembler and linker. (thanks to Nikodemus Siivola)
1338 * Unbound, undefined, undeclared variables now trigger full
1339 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1340 more often programmer error than deliberate exploitation of undefined
1342 * optimization: the hash algorithm for strings has changed to one
1343 that is less vulnerable to spurious collisions. (thanks to Juho
1345 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1346 needless bounds checking. (thanks to Juho Snellman)
1347 * optimization: implemented multiplication as a modular
1348 (UNSIGNED-BYTE 32) operation on the PPC backend.
1349 * fixed some bugs revealed by Paul Dietz' test suite:
1350 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1352 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1353 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1354 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1356 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1358 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1360 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1361 their output stream on EOF from read.
1362 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1363 have been read to end-of-file.
1364 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1366 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1367 description of determination of which consecutive characters
1369 ** Printing the "Space" character with escaping on now yields "#\\ ",
1370 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1371 ** Reading floating-point numbers with *READ-BASE* set to a number
1372 less than 10 works correctly.
1373 ** Reading floating-point numbers with *READ-BASE* set to a number
1374 more than 10 works correctly.
1375 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1376 the readtable currently in effect.
1378 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1379 * minor incompatible change: parsing of namestrings on a physical
1380 (Unix) host has changed; numbers after the final #\. in a
1381 namestring are no longer interpreted as a version field. This is
1382 intented to be largely invisible to the user, except that the
1383 meaning of the namestring "*.*.*" has changed: it now refers to a
1384 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1385 should usually be replaced by
1386 (make-pathname :name :wild :type :wild :version :wild)
1387 with the added benefit that this is more likely to be portable.
1388 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1389 option to OPEN now signals an error if the file being opened
1390 exists; this may have an impact on existing code.
1391 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1392 signals received, on the PowerPC platforms (both Linux and
1393 Darwin). (thanks to Pierre Mai for pointing out the location of
1395 * several fixes on OS X: The system now builds and runs cleanly on
1396 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1397 (thanks to Brian Mastenbrook)
1398 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1399 second-resolution: integer multiples of 1/3600 between -24 and 24.
1400 (thanks to Vincent Arkesteijn)
1401 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1402 type when called with 1 argument; PEEK-CHAR checked type of
1403 PEEK-TYPE only after having read first character from a
1404 stream. (reported by Peter Graves)
1405 * bug fix: the garbage collector now has much better locality
1406 behaviour, and in particular no longer treats all memory as being
1407 exhausted when many small objects point to each other in a deeply
1409 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1410 recognized as being TYPEP their class.
1411 * bug fix: the PUSHNEW documentation string has been corrected.
1412 (thanks to Vincent Arkesteijn)
1413 * bug fix: defaulting of the value for the last of an atypically
1414 large number of multiple values being bound was not being
1415 performed correctly on the Alpha or PPC platforms
1416 * optimization: implemented multiplication as a modular
1417 (UNSIGNED-BYTE 32) operation on the x86 backend.
1418 * optimization: SEARCH on simple-base-strings can now be open-coded.
1419 (see also contrib/compiler-extras.lisp for inspiration for
1420 teaching the compiler about the Boyer-Moore algorithm).
1421 * value, returned by MAX (and MIN) called with several EQUALP, but
1422 not EQL, arguments now does not depend on compiler settings.
1423 * fixed some bugs revealed by Paul Dietz' test suite:
1424 ** in stack analysis liveness information is propagated from
1425 non-local entry points.
1426 ** pathwise CAST removing failed when the CAST node did not start
1428 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1429 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1431 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1432 host is already defined.
1433 ** RENAME-FILE works on streams instead of signalling an internal
1435 ** PEEK-CHAR uses the current readtable when determining whether
1436 or not a character is whitespace.
1437 ** MERGE-PATHNAMES handles the case when the pathname does not
1438 specify a name while the default-pathname specifies a version.
1439 ** Pathnames now stand a better chance of respecting print/read
1441 ** Attempting to use standardized file system operators with a
1442 pathname with invalid :DIRECTORY components signals a
1444 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1446 ** DIRECTORY on logical pathnames is more correct.
1447 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1448 signal a TYPE-ERROR if their argument is not a stream.
1449 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1450 designator argument does not designate a stream.
1451 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1452 examining the synonym.
1453 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1455 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1456 element-type larger than ([UN]SIGNED-BYTE 32).
1458 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1459 * When built with the :SB-FUTEX feature, threaded builds now take
1460 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1461 for faster/more reliable mutex and condition variable support.
1462 * Incompatible change (but one you probably shouldn't have been using
1463 anyway): the interface and code for arbitrating between multiple
1464 threads in the same user session has been redesigned.
1465 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1466 error when called without an explicit environment argument.
1467 (thanks to Nikodemus Siivola)
1468 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1469 confused about their position. (thanks to Adam Warner and Gerd
1471 * bug fix: Pretty printing backquoted forms with unquotations in the
1472 argument list position of various code constructs such as LAMBDA
1473 now works correctly. (reported by Paul Dietz)
1474 * bug fix: Pretty printing unquotations no longer loses all
1475 stream position information.
1476 * optimization: performance of string output streams is now less
1477 poor for multiple small sequence writes.
1478 * optimization: performance of CSUBTYPEP in the presence of complex
1479 expressions involving CONS and NOT many times has been improved.
1480 (reported by Paul Dietz)
1481 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1483 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1484 streams for the REPL, for the debugger, and so on. (thanks to
1486 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1487 * fixed some bugs revealed by Paul Dietz' test suite:
1488 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1489 is ignored if it is too small, rather than propagating through
1490 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1491 ** extremely complex negations of CONS types were not being
1492 sufficiently canonicalized, leading to inconsistencies in
1494 ** VALUES tranformer lost derived type.
1496 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1497 * fixed a bootstrapping bug: the build process no longer assumes
1498 that the various BOOLE-related constants have the same value in
1499 host and target lisps. (noted by Paul Dietz' test suite on an
1500 SBCL binary built from CLISP)
1501 * The system can now be dynamically linked on the MIPS platform,
1502 which enables dynamic loading of foreign code from Lisp. (thanks
1503 to Ralf Baechle for discussions on the MIPS ABI)
1504 * The system now records debugging information for its own source
1505 files in a filesystem-position-independent manner, relative to
1506 the "SYS" logical host.
1507 * fixed a compiler bug: MV-LET convertion did not check references
1508 to the "max args" entry point. (reported by Brian Downing)
1509 * tweaked disassembly notes to be less confident about proclaiming
1510 some instruction as an LRA. (thanks to Brian Downing)
1511 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1512 now each have their own history, command character, and other
1513 characteristics. (thanks to David Lichteblau)
1514 * fixed some bugs revealed by Paul Dietz' test suite:
1515 ** compiler failure in compiling LOGAND expressions including a
1517 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1518 shift greater than 32.
1519 ** FLUSH-DEST did not mark blocks for type check regeneration.
1520 ** HANDLER-CASE failed to accept declarations in handler clauses
1521 in some circumstances.
1523 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1524 * New code in contrib/sb-introspect (still probably not entirely
1525 stable yet) provides some support for smart Lisp development
1526 environments like SLIME.
1527 * The conditions signalled for errors occurring when loading .fasl
1528 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1529 in a way which should help ASDF recover gracefully.
1530 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1531 stuff has been cleaned up. If you code contrib/ stuff, this might
1532 affect you, and you can look at contrib/README, contrib/STANDARDS,
1533 and/or the 0.8.4.27 diff to check.
1534 * In full calls the compiler now does not generate checks for declared
1535 argument types for all arguments.
1536 * various threading fixes
1537 ** and some experimental patches which didn't make it into
1538 the main tree for this release, but which are shipped in
1539 contrib/experimental-thread.patch as a possible fix for some
1540 failures (deadlock, spinning...) in GC-intensive multithreaded
1542 * fixed PPC build problem (source code incompatibility of different
1543 library versions): added offsetof() hackery which attempts to divine
1544 where glibc maintainers put uc_mcontext today
1545 * fixed bug 282: compiler does not trust type assertions while passing
1546 arguments to a full call.
1547 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1548 &OPTIONAL VALUES type parameter.
1549 * fix bug 214: algorithm for noting rejected templates is now more
1550 similar to that of template seletion. (also reported by rydis on
1552 * fixed bug 141b: printing backquoted information readably and prettily
1553 inserts a space where necessary.
1554 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1555 (CHARACTER 10) are now reported as errors, rather than propagated
1556 as unknown types. (reported by piso on #lisp)
1557 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1558 with values NIL and :ERROR. (thanks to Milan Zamazal)
1559 * fixed bug 191c: CLOS now does proper keyword argument checking as
1560 described in CLHS 7.6.5 and 7.6.5.1.
1561 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1562 longer raise an error; further, using a list as a for-as-arithmetic
1563 counter now raises a meaningful error.
1564 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1565 now understood by sequence creation functions such as MAKE-SEQUENCE
1567 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1568 a STRING-INPUT-STREAM.
1569 * compiler enhancement: SIGNUM is now better able to derive the type
1571 * type declarations inside WITH-SLOTS are checked. (reported by
1573 * fixed some bugs revealed by Paul Dietz' test suite:
1574 ** incorrect optimization of TRUNCATE for a positive first
1575 argument and negative second.
1576 ** compiler failure in let-convertion during flushing dead code.
1577 ** compiler failure while deriving type of TRUNCATE on an
1578 interval, containing 0.
1579 ** ASH of a negative bignum by a negative bignum count now returns
1581 ** intersection of CONS types now canonicalizes properly, fixing
1582 inconsistencies in SUBTYPEP.
1584 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1585 * incompatible change: The --disable-debugger command line
1586 option now clobbers the debugger at a more fundamental
1587 level, by redefining #'INVOKE-DEBUGGER instead of by
1588 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1589 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1590 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1591 enter the debugger REPL and then suffer recursive errors
1592 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1593 while under the new system, BREAK in --disable-debugger mode
1594 terminates the system just as an unhandled error would.
1595 * fixed compiler performance when processing loops with a step >1;
1596 * bug fix: DOCUMENTATION now retrieves generic function
1597 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1598 support has been systematized, and now supports the methods
1599 specified by ANSI, along with a default method and a method for
1600 slot documentation. (reported by Nathan Froyd)
1601 * bug fix: effective methods associated with a generic function are
1602 no longer cached over a change of that generic function's method
1603 combination. (reported by Andreas Fuchs)
1604 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1605 names a non-existent directory. (thanks to Andreas Fuchs)
1606 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1607 the PPC platform, lead to stack corruption; this has been fixed.
1608 (reported by Rainer Joswig)
1609 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1610 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1611 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1612 arithmetic in general, is now much more reliable on the PPC
1614 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1615 platform now returns the right answer.
1616 * optimization: restored some effective method precomputation in
1617 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1618 precomputation is now tunable.
1619 * optimization: compiler-internal data structure use has been
1620 reviewed, and changes have been made that should improve the
1621 performance of the compiler by about 20%.
1622 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1623 simple-base-strings and simple-bit-vectors is improved.
1624 * optimization: the optimization of 32-bit logical and arithmetic
1625 functions introduced in version 0.8.3 on the x86 has been
1626 implemented on the mips, ppc and sparc platforms; an
1627 implementation of the same facility, but for 64-bit arithmetic,
1628 has been added for the alpha.
1629 * microoptimization: the compiler is better able to make use of the
1630 x86 LEA instruction for multiplication by constants.
1631 * bug fix: in some situations compiler did not report usage of
1632 generic arithmetic in (SPEED 3) policy.
1633 * bug 145b fix: compiler used wrong type specifier while converting
1634 MEMBER-types to numeric.
1635 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1637 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1638 index argument now works properly on the Alpha platform.
1639 * bug fix: floating point exception treatment on the Alpha platform
1641 * bug fix: FILE-POSITION works much better on string input and
1642 output streams. (thanks to Nikodemus Siivola)
1643 * bug fix: many threading/garbage collection symptoms sorted.
1644 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1645 might be pseudo-atomic.
1646 * internal change: Stopping for GC is now done with signals not
1647 ptrace. GC is now done in whichever thread wanted it, instead of
1649 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1651 * bug fix: result form in DO is not contained in the implicit
1653 * incompatible change: ICR structure is changed; the value part of
1654 CONTINUATION is now called LVAR; corresponding functions are
1655 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1656 * added type deriver for ISQRT (thanks to Robert E. Brown).
1657 * bug fix: better support for loading from the command line when an
1658 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1660 * fixed some bugs revealed by Paul Dietz' test suite:
1661 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1662 ** ROUND and FROUND now give the right answer when given very
1663 small float arguments.
1664 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1666 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1667 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1668 ** LCM should return a non-negative integer.
1669 ** PARSE-INTEGER returned the index of a terminator instead of the
1670 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1671 ** PARSE-INTEGER returned an incorrect index being applied to a
1673 ** LCM with two arguments of 0 returns 0 rather than signalling
1675 ** unsigned addition of a 32-bit constant with the high bit set no
1676 longer causes an internal compiler error on the x86.
1677 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1678 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1679 with negative last argument.
1680 ** byte specifiers with zero size and position no longer cause
1681 an error during type derivation.
1682 ** bignum multiplication on the Alpha platform now returns the
1684 * porting: The system now builds on SuSE AMD64, although it still
1685 generates a 32-bit binary.
1686 * .fasl file incompatibility: The fasl file version number has
1687 been incremented (because of the changes to internal compiler
1688 data structures referred to above).
1690 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1691 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1692 more accurately, on the Darwin kernel running on PowerPC hardware.
1693 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1694 * Compiler code deletion notes now signal a condition of type
1695 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1696 an associated MUFFLE-WARNING restart.
1697 * The compiler now performs limited argument count validation of
1698 constant format strings in FORMAT, and where appropriate in ERROR,
1699 CERROR and WARN. (thanks to Gerd Moellmann)
1700 * New ASDF-INSTALL contrib can be used for automatic download and
1701 installation of third-party Lisp code from CCLAN or other sites
1703 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1704 which forces another thread to execute a function supplied by the
1706 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1707 now accept and act upon their :ELEMENT-TYPE keyword argument.
1708 (reported by Martin Atzmueller, Edi Weitz)
1709 * bug fix: FILE-POSITION now accepts position designators up to
1710 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1711 is the greater. (thanks to Patrik Nordebo)
1712 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1713 Kalle Olavi Niemitalo)
1714 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1715 printable. (reported by Eric Marsden)
1716 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1717 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1718 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1719 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1720 (thanks to Stig Sandoe)
1721 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1722 UNREAD-CHAR now correctly includes the unread character in the
1723 target sequence. (thanks to Gerd Moellmann)
1724 * bug fix in threaded builds: the system can now be suspended and
1725 resumed by shell job control with minimal disruption.
1726 * bug fixes in times and timezones >2038AD
1727 * better handling of "where is GNU make?" problem in build scripts
1728 (thanks to Nikodemus Siivola)
1729 * new optimization: inside a named function any reference to a
1730 function with the same name is considered to be a self-reference;
1731 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1732 optimization quality.
1733 * new optimization on x86: logical functions and + now have
1734 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1735 used when the result is truncated to 32 bits.
1736 * VALUES declaration is partially enabled.
1737 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1738 * fixed some bugs revealed by Paul Dietz' test suite:
1739 ** The system now obeys the constraint imposed by
1740 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1741 types form a lattice under type intersection.
1742 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1743 ** ASSOC now ignores NIL elements in an alist.
1744 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1745 and (1+ MOST-POSITIVE-FIXNUM) answers.
1746 ** The addition of a method with invalid qualifiers to a generic
1747 function does not cause an error to be signalled immediately;
1748 a warning is signalled, and the error is generated only on
1749 calling the generic function.
1750 * changed .fasl file version number, in part to add type codes for
1751 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1752 obscure ANSI requirements
1754 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1755 * fixed bug 148: failure to inline-expand a local function left
1756 garbage, confusing the compiler.
1757 * fixed bugs 3cd: structure slot readers perform type check if the
1758 slot can have an invalid value (i.e. it is either not initialized
1759 or can be written with a less specific slot writer).
1760 * bug fix: the compiler now traps array references to elements off
1761 the end of an array; previously, the bounds checking in some
1762 circumstances could go off-by-one.
1763 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1765 * type declarations for array element types now obey the description
1766 on the CLHS page "Declaration TYPE", as per discussions on
1767 sbcl-help around 2003-05-08. This means that a declaration
1768 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1769 declaration, all references to BAR will be asserted or assumed
1770 (with THE, so dependent on compiler policy) to involve objects of
1771 type FOO. Note that no such declaration is implied in
1772 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1773 * declared types of functions from the "Conditions"
1774 chapter. (reported by Paul Dietz)
1775 * bug fix: CERROR accepts a function as its first argument.
1776 * bug fix: NTH an NTHCDR accept a bignum as index
1777 arguments. (reported by Adam Warner)
1778 * optimization: character compare routines now optimize comparing
1779 against a constant character. (reported by Gilbert Baumann)
1780 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1781 argument now works properly on the MIPS platform.
1782 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1783 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1784 declared to have a functional type, does not cause an error
1786 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1787 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1788 * fixed bug: if last continuation of a deleted block has a
1789 destination, this destination should be deleted too. (reported by
1791 * fixed a bug in the bootstrap process: the host compiler's values
1792 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1793 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1794 test case from Patrik Nordebo)
1795 * improved the ability of the disassembler on the PPC platform to
1796 provide helpful disassembly notes.
1797 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1798 the class in more cases than previously.
1799 * bug fix: FILE-POSITION now understands :START and :END for
1800 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1801 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1802 8) data. (thanks to David Lichteblau)
1803 * bug fix: it is possible to add a method to a generic function
1804 without lambda list.
1805 * bug fix: reader failed to signal END-OF-FILE inside an
1806 object representation. (reported by Nikodemus Siivola)
1807 * fixed some bugs revealed by Paul Dietz' test suite:
1808 ** LAST and [N]BUTLAST should accept a bignum.
1809 ** condition slot accessors are methods.
1810 ** (VECTOR NIL) is a subtype of STRING.
1812 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1813 * minor incompatible change: some nonsensical specialized lambda
1814 lists (used in DEFMETHOD) which were previously ignored now signal
1816 * minor incompatible change: the system is now aware of the types of
1817 variables in the COMMON-LISP package, and will signal errors for
1818 most violations of these type constraints (where previously they
1819 were silently accepted).
1820 * minor incompatible change: COMPILE-FILE now uses the freedom
1821 afforded (ANSI 3.2.2.3) to use derived function types for
1822 functions defined in the same file. This also permits the system
1823 to warn on static type mismatches and function
1824 redefinition. (Currently it does not work with high DEBUG level.)
1825 * minor incompatible change: VALUES declaration is disabled.
1826 * When issuing notes, the compiler now signals a condition of type
1827 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1828 restart for use in user handlers. It is expected that the
1829 COMPILER-NOTE condition will eventually become a condition
1830 supertype to a hierarchy of note types, which will then be
1831 handleable in a similar fashion. However, at the moment, no such
1832 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1833 but it's an implementation detail, not a classification for the
1835 * Changes in type checking closed the following bugs:
1836 ** type checking of unused values (192b, 194d, 203);
1837 ** template selection based on unsafe type assertions (192c, 236);
1838 ** type checking in branches (194bc).
1839 * A short form of VALUES type specifier has ANSI meaning (it has
1840 increased the number of situations when SBCL cannot perform type
1842 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1843 keywords or constants is permissible.
1844 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1845 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1846 classes. (thanks to Antonio Martinez)
1847 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1848 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1849 argument to a pretty-print stream. Instead, it leaves any such
1850 operations to DESCRIBE-OBJECT methods.
1851 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1853 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1854 lambda lists. (thanks to David Lichteblau)
1855 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1856 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1857 (thanks to Gerd Moellmann)
1858 * fixed bug 246: increased compilation speed of long
1859 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1861 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1863 * DEFCONSTANT now throws a condition of type
1864 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1865 constant to a non-EQL value; CONTINUE and ABORT restarts
1866 respectively change and preserve the value.
1867 * fixed bug 63: The code walker, part of the implementation of CLOS,
1868 is now better at handling symbol macros.
1869 * bug fix: There is no longer an internal implementation type named
1870 CL:LENGTH. (reported by Raymond Toy)
1871 * bug fix: In macro-like defining macros/special operators the
1872 implicit block does not enclose the lambda list.
1873 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1874 enclosed in parentheses) are not suitable as type specifiers, and
1875 their use properly signals an error now.
1876 * bug fix: An argument count mismatch for a type specifier in code
1877 being compiled no longer causes an unhandled error at compile
1878 time, but signals a compile-time warning.
1879 * fixed simple vector readable printing
1880 * bug fix: DESCRIBE takes more care over whether the class
1881 precedence list slot of a class is bound before accessing it.
1882 (reported by Markus Krummenacker)
1883 * bug fix: FORMATTER can successfully compile pretty-printer format
1884 strings which use variants of the ~* directive inside.
1885 * bug fix: SEARCH now applies its TEST predicate to the elements of
1886 the arguments in the correct order. (thanks to Wolfhard Buss)
1887 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1888 in type inference. (thanks to Robert E. Brown)
1889 * bug fix: Reading in symbols with an explicit package name of ""
1890 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1891 * fixed some bugs revealed by Paul Dietz' test suite:
1892 ** NIL is now allowed as a structure slot name.
1893 ** Arbitrary numbers, not just REALs, are allowed in certain
1894 circumstances in LOOP for-as-arithmetic clauses.
1895 ** Multiple class redefinitions before slot access no longer
1896 causes a type error.
1897 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1898 association between the name and a class.
1899 ** Generic functions with non-standard method-combination and over
1900 five methods all of which return constants no longer return NIL
1901 after the first few invocations. (thanks to Gerd Moellmann)
1902 ** CALL-NEXT-METHOD with no arguments now passes the original
1903 values of the arguments, even in the presence of assignment.
1904 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1906 ** DISASSEMBLE works with closures and funcallable instances.
1907 ** ADD-METHOD now returns the generic function, not the new method.
1908 ** FIND-METHOD signals an error if the lengths of the specializers
1909 is incompatible with the generic function, even if the ERRORP
1911 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1912 which its argument is a member.
1913 ** DEFCLASS only redefines the class named by its class-name
1914 argument if that name is the proper name of the class;
1915 otherwise, it creates a new class.
1916 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1917 of the UNBOUND-SLOT condition to the name of the slot.
1918 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1919 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1920 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1921 SLOT-MAKUNBOUND in the specified fashion.
1923 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1924 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1925 cross-compilation host. As a consequence, we can now bootstrap our
1926 way up to SBCL starting with a bare gcc toolchain and human-readable
1927 source code (first the source to CLISP, then the source to SBCL).
1928 * A contributed module containing a partial implementation of the
1929 simple-streams interface has been included. (thanks to Rudi
1931 * A contributed module implementing the RFC1321 Message Digest
1932 Algorithm, known as MD5, has been included.
1933 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1934 no longer has any effect, as the code controlled by this feature
1935 has been deleted. (As far as we know, no-one has ever built using
1936 this feature, and its semantics were confused in any case).
1937 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1938 work on conditions (as required by the ANSI specification),
1939 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1940 expected behaviour on conditions. Users should note, however,
1941 that such behaviour is not required by the ANSI specification,
1942 and so use of this behaviour may render their code unportable.
1943 * fixed some bugs revealed by Paul Dietz' test suite:
1944 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1946 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1947 ** SXHASH on condition objects no longer returns NIL.
1948 ** :ALLOCATION :CLASS slots are better treated; their values are
1949 updated on class redefinition, and initforms inherited from
1950 superclasses are applied.
1951 ** REMOVE-METHOD returns its generic function argument even when
1952 no method was removed.
1953 ** SHARED-INITIALIZE now initializes the values of the requested
1954 slots, including those with :ALLOCATION :CLASS.
1955 ** ALLOCATE-INSTANCE now works on structure classes defined via
1956 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1958 ** SLOT-EXISTS-P now works on conditions, as well as structures
1960 ** MAKE-LOAD-FORM now has the required methods on
1961 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1962 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1963 keyword, and now implements the SLOT-NAMES argument.
1964 ** methods with &OPTIONAL arguments no longer allow too many
1965 arguments to be passed in the call without error.
1966 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1967 option is consistent with the required arguments of the generic
1968 function lambda list.
1969 * bug fix: REQUIRE accepts a string designator. (Thanks to
1971 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1972 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1973 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1974 * bug fix: The compiler now checks for duplicated variables in macro
1976 * bug fix: SETQ on globals returns the correct value.
1977 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1978 not a COMPILER-ERROR (followed by some other strange error on
1979 choosing the CONTINUE restart).
1980 * bug fix: make.sh and friends are now more consistent in the way that
1981 they look for GNU "make".
1983 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1984 * experimental native threads support (on x86 Linux >=2.4 only).
1985 This is not compiled in by default: you need to add :SB-THREAD to
1986 the target features. See the "Beyond ANSI" chapter of the manual
1988 * fix for longstanding nonANSIism: The old distinction between
1989 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1990 The return value from CL:FIND-CLASS is now a CLOS class, and
1991 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1992 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1993 * An interface to the MetaObject Protocol, as described in Kiczales,
1994 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1995 MIT Press, 1991, is available from the SB-MOP package.
1996 * incompatible change: the SB-PCL package should now be considered
1997 a private implementation detail, and no longer a semi-private MOP
1999 * minor incompatible change: due to rearrangement for threads, the
2000 control stack and binding stack are now allocated at arbitrary
2001 addresses instead of being hardcoded per-port. Users affected by
2002 this probably have to be doing advanced things with shared
2003 libraries, and will know who they are.
2004 * minor incompatible change: Previously, all --eval forms used were
2005 processed with READ before any of them were processed with EVAL.
2006 Now each --eval form is processed with both READ and EVAL before
2007 the next --eval form is processed. (Thus package operations like
2008 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2009 work as the user might reasonably expect.)
2010 * minor incompatible change: *STANDARD-INPUT* is now only an
2011 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2013 * minor incompatible change: Y-OR-N-P is now character-oriented,
2014 not line oriented. Also, YES-OR-NO-P now works without errors.
2015 (thanks to Antonio Martinez)
2016 * sb-aclrepl module improvements: an integrated inspector, added
2017 repl features, and a bug fix to :trace command.
2018 * Known functions, which cannot be open coded by the backend, are
2019 considered to be able to check types of their arguments. (fixing
2020 a bug report by Nathan J. Froyd)
2021 * fixed a bug in computing method discriminating functions: It is
2022 now possible to define methods specialized on classes which have
2023 forward-referenced superclasses. (thanks to Gerd Moellmann)
2024 * fixed evaluation order in optional entries (reported by Gilbert
2026 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2027 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2028 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2029 required-by-AMOP NAME argument, as well as CLASS and
2030 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2031 * fixed bug 20: DEFMETHOD can define methods using names that are
2032 not the proper names of classes to designate class specializers.
2033 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2034 Unix file descriptors, instead of blowing up. (thanks to Antonio
2036 * Garbage collection refactoring: user-visible change is that a
2037 call to the GC function during WITHOUT-GCING will not do garbage
2038 collection until the end of the WITHOUT-GCING. If you were doing
2039 this you were probably losing anyway.
2040 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2041 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2042 TYPEP the latter but not the former.
2043 * The compiler issues a full WARNING for calls to undefined functions
2044 with names from the CL package.
2045 * MAP-INTO for a vector destination is open coded. (reported by
2046 Brian Downing on c.l.l)
2047 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2048 documentation string.
2049 * fixed some bugs revealed by Paul Dietz' test suite:
2050 ** COPY-ALIST now signals an error if its argument is a dotted
2052 ** Condition slots are now accessed more correctly in the presence
2053 of multiple initargs for a given slot.
2054 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2055 exclude restarts of the same name associated with a different
2057 ** DEFCLASS of forward-referenced classes with another
2058 forward-referenced class in the superclasses list no longer
2060 ** Condition slots are now initialized once each, not multiple
2061 times. (thanks to Gerd Moellmann)
2062 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2064 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2065 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2066 arguments contain duplicated elements.
2067 ** RESTART-CASE understands local macros.
2068 ** RESTART-CASE associates exactly its own restarts with a condition.
2069 ** ENDP in safe mode checks its argument to be of type LIST.
2070 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2071 in question is unbound.
2072 ** Optimized MAKE-INSTANCE functions no longer cause internal
2073 assertion failures in the presence of duplicate initargs.
2074 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2075 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2076 propagated even in the ctor optimized implementation of
2078 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2080 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2081 * a better implementation of SXHASH on (simple) bit vectors,
2082 measured both in execution speed and in distribution of results
2083 over the positive fixnums, has been installed. Likewise, a better
2084 implementation of EQUAL for simple bit vectors is now available.
2085 * fixed CEILING optimization for a divisor of form 2^k.
2086 * fixed bug 240 (emitting extra style warnings "using the lexical
2087 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2088 by Antonio Martinez)
2089 * fixed SXHASH, giving different results for NIL depending on type
2090 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2091 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2092 variable. (found by Rolf Wester)
2093 * SBCL does not ignore type declarations for special
2094 variables. (reported by rif on c.l.l 2003-03-05)
2095 * some bug fixes in contrib/sb-aclrepl/
2096 * fixed some bugs revealed by Paul Dietz' test suite:
2097 ** a bug in the CONS type specifier, whereby the CAR and CDR
2098 types got intertwined, has been fixed;
2099 ** the type system is now able to reason about the interaction
2100 between INTEGER and RATIO types more completely;
2101 ** APPEND, [N]REVERSE and NRECONC check that those their
2102 arguments, which must be proper lists, are really so;
2103 ** An array specialized to be unable to hold elements has been
2104 implemented, as required -- yes, really -- by ANSI;
2105 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2106 on malformed property lists;
2108 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2109 * incompatible packaging change: in line with Unix convention,
2110 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2111 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2112 * REQUIRE and PROVIDE are now optionally capable of doing something
2113 useful. See the documentation string for REQUIRE.
2114 * infrastructure for a managed SBCL contrib system: contributed
2115 modules in this release include:
2116 ** the ASDF system definition facility;
2117 ** an interface to the BSD Sockets API;
2118 ** an ACL-like convenience interface to the repl;
2119 (thanks to Kevin Rosenberg)
2120 ** an implementation of ROTATE-BYTE, with efficient implementation
2122 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2123 gives a closer approximation to the right answer than previously.
2124 (thanks to Raymond Toy)
2125 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2126 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2127 optional environment arguments, as required by ANSI.
2128 * fixed bugs in other functions taking environment objects, allowing
2129 calls with an explicit NIL environment argument to be compiled
2131 * fixed bug 228: primary return values from
2132 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2133 COMPILE or FUNCTION.
2134 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2135 :TYPE will no longer signal errors on innocuous objects.
2136 * fixed bug 231b: SETQ is better at respecting type declarations in
2137 the lexical environment.
2138 * fixed a bug in DEFCLASS: classes named by symbols with no or
2139 unprintable packages can now be defined.
2140 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2141 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2142 * fixed compiler failure related to checking types of functions.
2143 (reported by Robert E. Brown)
2144 * the compiler is now much more consistent in its error-checking
2145 treatment of bounding index arguments to sequence functions: in
2146 (SAFETY 3) code, errors will be signalled in almost all cases if
2147 invalid sequence bounding indices are passed to functions defined
2148 by ANSI to operate on sequences.
2149 * fixed a bug in the build procedure: documentation of SBCL-specific
2150 packages is now preserved and available in the final Lisp image.
2151 * lifted FDEFINITION lookup out of loops in the implementation of
2152 many list operations. (thanks to Robert E. Brown)
2153 * fixed a bug in the reader: the #n# reader macro now works for
2154 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2155 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2156 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2157 E. Brown and Tony Martinez respectively)
2158 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2159 SLOT-BOUNDP: in particular, it is now able to optimize them much
2160 better, and is now not vulnerable to having packages renamed.
2161 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2162 always called when a slot is not present in an instance. (thanks
2164 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2165 longer signal an error on unreading a character following EOF on
2166 the previous constituent stream. (thanks to Tony Martinez)
2167 * fixed some bugs revealed by Paul Dietz' test suite:
2168 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2169 not just nonnegative fixnums;
2170 ** the logical bit-array operators such as BIT-AND now accept an
2171 explicit NIL for their "opt-arg" argument (to indicate a
2172 freshly-consed result bit-array);
2173 ** ELT now signals an error on an invalid sequence index in safe
2175 ** the type system is now cleverer about negations of numeric
2176 types, and consequently understands the BIGNUM and RATIO types
2178 ** the type system is now cleverer about the interaction between
2179 INTEGER and RATIO types: while bugs still remain, many more
2180 cases are accurately computed;
2181 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2182 if it is in the last clause;
2183 ** CONSTANTLY now correctly returns a side-effect-free function in
2185 ** DECLARE is no longer treated as a special-operator; in
2186 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2187 * incremented fasl file version number due to the change in the
2188 DEFSTRUCT-SLOT-DESCRIPTION structure.
2190 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2191 * minor incompatible change: code processed by the "interpreter" or
2192 EVAL now has a compilation optimization policy of (DEBUG 2)
2193 (changed from (DEBUG 1)) to improve debuggability of interactive
2194 development, and to allow the use of the debug RETURN command in
2196 * an experimental implementation of the RETURN command for the
2197 debugger has been included. (thanks to Frederik Kuivinen)
2198 * fixed bug 62: constraints were not propagated into a loop.
2199 * fixed bug in embedded calls of SORT (reported and investigated by
2201 * fixed some bugs revealed by Paul Dietz' test suite:
2202 ** printing and reading of arrays with some dimensions having
2203 length 0 (thanks to Gerd Moellmann);
2204 ** BOA constructor with &AUX argument without a default value does
2205 not cause a type error;
2206 ** CONSTANTP now returns true for all self-evaluating objects.
2208 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2209 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2210 accessors that are related by inheritance, as specified in the
2211 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2212 Valtteri Vuorikoski)
2213 * The compiler is now able to inline functions that were defined in
2214 a complex lexical environment (e.g. inside a MACROLET).
2215 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2217 * Support for the upcoming FreeBSD-5.0 release has been included.
2218 (thanks to Dag-Erling Smorgrav)
2219 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2220 effect when it is not in a toplevel context.
2221 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2222 stand a better chance of being correct. (thanks to Gerd
2224 * fixed bug in COERCE, which now signals an error on coercing a
2225 rational to a bounded real type which excludes the expected
2227 * The compiler is now able to derive types more accurately from the
2228 COERCE and COMPILE functions.
2229 * fixed bug 223: functional binding is considered to be constant
2230 only for symbols in the CL package.
2231 * fixed bug 231: SETQ did not check the type of a variable being set
2232 (reported by Robert E. Brown)
2233 * A new optimization for MAKE-INSTANCE has been included, fixing
2234 various bugs (including relating to :ALLOCATION :CLASS slots and
2235 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2237 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2238 ** As required by ANSI, LOOP now disallows anonymous collection
2239 clauses such as COLLECT I in conjunction with aggregate boolean
2240 clauses such as THEREIS (= I 1);
2241 ** LOOP now signals an error when any variable is reused in the
2242 same loop (including the potentially useful construct analogous
2243 to WITH A = 1 WITH A = (1+ A);
2244 ** IT is only a special loop symbol within the first clause of a
2245 conditional loop clause;
2246 ** LOOP with a typed iteration variable over a hashtable now
2247 signals a type error iff it should.
2248 * fixed some other bugs revealed by Paul Dietz' test suite:
2249 ** FILE-STREAM now names the class previously known as FD-STREAM;
2250 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2251 argument) no longer signals an error;
2252 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2253 of the slot symbol, rather than using the current package
2254 ((:CONC-NAME "") continues to intern the slot's name in the
2256 * incremented fasl file version number, because of the incompatible
2257 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2258 of the new implementation of DEFINE-COMPILER-MACRO.
2260 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2261 * Support for building SBCL for MIPS platforms running in
2262 little-endian mode has now been checked in, and basic
2263 functionality on said platforms verified.
2264 * minor incompatible change: PCL now records the pathname of a file
2265 in which methods and the like are defined, rather than its
2267 * minor incompatible change: TRUENAME now considers the truename of
2268 a file naming a directory to be the pathname with :DIRECTORY
2269 component indicating that directory.
2270 * minor incompatible change: a NAMED clause in the extended form of
2271 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2272 reason for the previous behaviour is unclear.
2273 * more systematization and improvement of CLOS and MOP conformance
2274 in PCL (thanks to Gerd Moellman and Pierre Mai):
2275 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2277 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2278 primary methods with no specializers;
2279 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2281 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2282 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2283 FORWARD-REFERENCED-CLASSes; error reporting on
2284 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2286 ** SXHASH on CLOS instances now uses a slot internal to the
2287 instance to return different numbers on distinct instances,
2288 while preserving the same return value through invocations of
2290 ** DEFMETHOD signals errors when methods with longer incongruent
2291 lambda lists are added to generic functions;
2292 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2293 CLASS, as specified in AMOP;
2294 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2295 on the order returned by the primary method for classes of
2296 class STANDARD-CLASS;
2297 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2298 * fixed some bugs shown by Paul Dietz' test suite:
2299 ** DOLIST puts its body in TAGBODY;
2300 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2302 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2303 value producing form;
2304 ** if more variables are given to PROGV than values, extra
2305 variables are bound and made to have no value;
2306 ** NSUBSTITUTE on list arguments gets the right answer with
2308 ** ELT signals an error of type TYPE-ERROR when the index argument
2309 is not a valid sequence index;
2310 ** LOOP signals (at macroexpansion time) an error of type
2311 PROGRAM-ERROR when duplicate variable names are found;
2312 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2313 ** FUNCALL of special-operators now cause an error of type
2315 ** PSETQ now works as required in the presence of side-effecting
2316 symbol-macro places;
2317 ** NCONC accepts any object as its last argument;
2318 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2320 ** loop-for-as-package does not require a package to be explicitely
2322 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2324 * fixed bug 166: compiler preserves "there is a way to go"
2325 invariant when deleting code.
2326 * fixed bug 172: macro lambda lists with required arguments after
2327 &REST arguments now cause an error to be signalled. (thanks to
2329 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2330 bug. (thanks to Matthew Danish)
2331 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2333 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2334 arguments are lexically rebound. (thanks to Gerd Moellmann and
2336 * fixed bug 194: error messages are now more informative when there
2337 is no primary method applicable in a call to a generic
2338 function. (thanks to Gerd Moellmann)
2339 * fixed bug in command line argument checking (thanks to Julian
2341 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2342 (thanks to Matthew Danish)
2343 * incremented fasl file version number, because of the
2344 SXHASH-related changes in the layout of CLOS data structures
2346 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2347 * minor incompatible change: The runtime (the Unix executable named
2348 "sbcl") is now much pickier about the .core files it will load.
2349 Essentially it now requires .core files to descend from the same
2350 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2351 as the runtime does. (The intent is to prevent the crashes which
2352 can occur, and which can even be reported as mysterious failures,
2353 when people patch the sources or change the build parameters
2354 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2355 sbcl and .core files.)
2356 * fixed bug: VALUES-LIST is no longer optimized away.
2357 * fixed bug 142: The FFI conversion of C string values to Lisp
2358 string values no longer conses excessively. (thanks to Nathan
2359 Froyd porting Raymond Toy's fix to CMU CL)
2360 * began to systematize and improve MOP conformance in PCL (thanks to
2361 Nathan Froyd, Gerd Moellman and Pierre Mai):
2362 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2364 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2365 ** FINALIZE-INHERITANCE is now called on class finalization;
2366 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2367 argument precedence order.
2368 * fixed bug 202: The compiler no longer fails on functions whose
2369 derived types contradict their declared type.
2370 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2371 so it can be non-toplevel.
2372 * The fasl file version number has changed (because of the new
2373 implementation of DEFMACRO).
2374 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2375 safe code, that any length requirement by their type-specifier
2376 argument is valid. The exceptions to this are described in bug
2378 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2379 functions now check the types of their inputs as required by ANSI.
2380 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2381 introduced symbol is DECLAREd to be SPECIAL.
2382 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2383 * fixed bug: :COUNT argument to sequence functions may be negative
2384 * fixed bug: body of DO-SYMBOLS may contain declarations
2385 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2386 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2387 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2388 symbol macro only once
2389 * fixed printing of call frame when argument list is unavailable
2390 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2391 * compiler no longer signals WARNING on unknown keyword
2394 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2395 * A beta-quality port to the mips architecture running Linux,
2396 based on the old CMUCL backend, has been made. It has been tested
2397 on a big-endian kernel, and works sufficiently well to be able to
2398 rebuild itself; it has not been tested in little-endian mode.
2399 * fixed an inconsistency between gencgc.c and purify.c which made
2400 dumping/loading .core files unreliable
2401 * fixed bug 120a: The compiler now deals correctly with IFs where
2402 the consequent is the same as the alternative, instead of
2403 misderiving the return type. (thanks to Alexey Dejneka)
2404 * fixed bug 113: Logical pathnames are now dumpable (the logical
2405 host is resolved at load-time, throwing an error if it is not
2407 * fixed bug 174: FORMAT's error message is slightly clearer when a
2408 non-printing character is used in a format directive.
2409 * fixed several bugs in compiler checking of type declarations, i.e.
2410 violations of the Python "declarations are assertions" principle
2411 (thanks to Alexey Dejneka)
2412 * fixed several bugs in PCL's error checking (thanks to Gerd
2414 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2416 * fixed bug in compilation of functions as first class values
2417 (thanks to Antonio Martinez-Shotton)
2418 * The compiler's handling TYPE-ERRORs which it can prove will
2419 inevitably happen at runtime has been cleaned up and corrected
2420 in several ways. (thanks to Alexey Dejneka)
2421 * improved argument type checking for various basic arithmetic
2422 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2423 lavished on them in the past that they can be compiled in many
2424 ways in different special cases
2425 * fixed bug 181: compiler checks validity of user supplied type
2427 * cleaned up code flushing in optimization: Function calls which
2428 should signal errors for safety purposes (e.g. which ANSI says
2429 should signal errors when their arguments are of incorrect type)
2430 are no longer optimized away.
2431 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2432 * incremented fasl file version number, because changes in the
2433 implementation of sequence functions like COERCE caused
2434 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2435 in old inline expansions) to become undefined. (Actually these
2436 changes were later undone, so we might very well be binary
2437 compatible with 0.7.7 after all, but leaving the version number
2438 incremented seemed like the simplest and most conservative
2441 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2442 * An alpha-quality port to the parisc architecture running Linux,
2443 based on the old CMUCL backend, has been made. This, even more so
2444 than the other backends, should be considered still a work in
2445 progress; known problems include that the Linux kernel in 64-bit
2446 mode does not propagate the correct sigcontext structure to
2447 userspace, and consequently SBCL on a parisc64 kernel will not
2449 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2450 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2451 suppressed.) Also now that the compiler is looking at declarations
2452 in the environment, it checks optimization declarations as well,
2453 and suppresses inlining when (> DEBUG SPEED).
2454 * More fixes have been made to treatment of floating point exception
2455 treatment and other Unix signals. In particular, floating point
2456 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2457 * The detection and handling of control stack exhaustion (infinite
2458 or very deeply nested recursion) has changed. Stack exhaustion
2459 detection is now done by write-protecting pages at the OS level
2460 and applies at all optimization settings; when found, a
2461 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2462 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2463 be caught using IGNORE-ERRORS.
2464 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2465 correctly when transforming recursive LABELS functions to LETs.
2466 (thanks to Alexey Dejneka)
2467 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2468 that are names of constants or global variables.
2469 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2470 alien routines with docstrings.
2471 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2472 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2474 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2475 (thanks to Eric Marsden)
2476 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2477 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2478 * Bug fix: The SPARC backend can now compile functions involving
2479 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2480 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2481 to a functions expecting a PATHNAME-DESIGNATOR.
2482 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2483 lambda lists. (thanks to Alexey Dejneka)
2484 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2485 functions so that EXPORTing the name of the class doesn't cause
2486 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2487 to Antonio Martinez for reporting this)
2488 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2489 constant-folded without failing an assertion. (thanks to Einar
2490 Floystad Dorum for reporting this)
2491 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2492 types are now dealt with more correctly by the compiler.
2493 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2494 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2495 behaviour analogous to RENAME-FILE. This puts its behaviour more
2496 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2498 * The fasl file version number has changed again. (because of the
2499 bug fix involving the names of PCL MAKE-INSTANCE functions)
2501 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2502 * bug fix: Floating point exceptions are treated much more
2503 consistently on the x86/Linux and PPC/Linux platforms.
2504 * Array initialization with :INITIAL-ELEMENT is now much faster for
2505 cases when the compiler cannot open code the array creation, but
2506 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2507 array accesses have also seen a speed increase.
2508 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2509 specified. (This was at the root of some bad interactions between
2510 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2512 * bug fix: Internal error arguments for undefined functions are now
2513 computed correctly on the PPC/Linux platform.
2514 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2515 Raymond Toy's patch for CMU CL)
2516 * Support for the Solaris 9 operating environment has been included
2517 (thanks to Daniel Merritt)
2518 * A very ugly but hopefully complete draft of the missing FFI chapter
2519 of the manual has been created by reformatting the corresponding
2520 CMU CL manual chapter into (currently very ugly and incoherent)
2521 DocBook and bringing it up to date for SBCL behavior. Thus, the
2522 manual is now essentially complete, at least by my extreme
2523 once-and-only-once standards, whereby it's acceptable to refer to
2524 the doc strings of SB-EXT functions as the primary documentation.
2525 * The fasl file version number has changed again, due to cleanup of
2526 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2527 is no longer a static symbol.)
2529 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2530 * SBCL now builds with OpenMCL (version 0.12) as the
2531 cross-compilation host; also, more progress has been made toward
2532 bootstrapping under CLISP.
2533 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2535 * bug 158 fixed: The compiler can now deal with integer loop
2536 increments different from 1; fixing this turned out also to fix
2538 * bug 169 fixed: no more bogus warnings about using lexical bindings
2539 despite the presence of perfectly good SPECIAL declarations (thanks
2540 to David Lichteblau)
2541 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2542 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2543 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2545 * bug 180 fixed: Method combination specifications no longer ignore
2546 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2547 * bug fix: Structure type predicate functions now check their argument
2548 count as they should.
2549 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2550 correctly. (thanks to Pierre Mai)
2551 * minor incompatible change: The --noprogrammer option is deprecated
2552 in favor of the new --disable-debugger option, which is very similar.
2553 (The major difference is that it takes effect at a slightly different
2554 time at startup, causing handling of errors in --sysinit and
2555 --userinit files will be affected differently.) The
2556 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2557 been added to allow this functionality to be controlled from ordinary
2558 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2559 maintainers, who might want to run non-interactive scripts to
2560 build SBCL cores which will later be used interactively.)
2561 * minor incompatible change: The LOAD function no longer, when given
2562 a wild pathname to load, loads all files matching that pathname.
2563 Instead, an error of type FILE-ERROR is signalled.
2565 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2566 * bug 147 fixed: The compiler preserves its block link/count
2567 invariants more correctly now so that it doesn't crash. (thanks
2569 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2571 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2572 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2573 no-value-for-this-character as NIL (as specified by ANSI).
2574 * HOST-NAMESTRING on physical pathnames now returns a string that is
2575 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2576 (thanks to Christophe Rhodes)
2577 * The Alpha port handles icache flushing more correctly. (thanks to
2579 * More progress has been made toward bootstrapping under CLISP. (thanks
2580 to Christophe Rhodes)
2581 * The fasl file format has changed again, because dynamic loading
2582 on OpenBSD (which has non-ELF object files) motivated some cleanups
2583 in the way that foreign symbols are transformed and passed around.
2584 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2585 is no longer treated as whitespace by the reader, but instead as
2586 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2587 |A
\7fB|, instead of A as it used to.
2589 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2590 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2591 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2592 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2593 this as of 0.7.1.45, but was left out of the previous news section
2594 (thanks to Dan Barlow)
2595 * SBCL now runs on the Solaris operating system on SPARC architectures
2596 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2597 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2598 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2599 * SPARC backend cleanups, allowing builds of cores optimized for V8
2600 and V9 SPARCS, and also emission of code targeted to a particular
2601 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2603 * SBCL is closer to bootstrapping under CLISP, thanks to various
2604 fixes by Christophe Rhodes.
2605 * The fasl file format has changed again, to allow the compiler's
2606 INFO database to support symbol macros.
2607 * The user manual (in doc/) is formatted into HTML more nicely.
2608 (thanks to coreythomas)
2609 * The system is smarter about SUBTYPEP relationships, especially
2610 those involving NOT types (including types such as ATOM which are
2611 represented internally using NOT types). Thus SUBTYPEP is less
2612 likely to return (VALUES NIL NIL) in general, and in particular
2613 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2615 * The fasl file format has changed again, because the internal
2616 representation of types now includes a new slot to support the new
2617 SUBTYPEP-of-NOT-types logic.
2618 * (not a change in the main branch of SBCL, but a related prototype
2619 which can hopefully be merged into the main branch of SBCL in the
2620 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2621 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2622 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2623 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2624 dispatch macro character combinations. (thanks to Alexey Dejneka)
2625 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2626 that superficially look like logical namestrings correctly.
2627 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2629 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2630 default supplied-p) for &optional and &key arguments. (thanks to
2633 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2634 * incompatible change: The compiler is now less aggressive about
2635 tail call optimization, doing it only when (> SPACE DEBUG) or
2636 (> SPEED DEBUG). (This is an incompatible change because there are
2637 programs which relied on the old CMU-CL-style behavior to optimize
2638 away their unbounded recursion which will now die of stack overflow.)
2639 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2640 for non-GENCGC systems has been increased to 20M (since that
2641 seems much closer to the likely performance optimum for modern
2642 systems than the old 4M value was)
2643 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2644 and *DEBUG-PRINT-LEVEL*
2645 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2646 of CMU CL's support for SPARC, and various endianness and other
2647 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2648 * new syntactic sugar for the Unix command line: --load foo.bar is now
2649 an alternate notation for --eval '(load "foo.bar")'.
2651 ** The system now detects stack overflow and handles it gracefully,
2652 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2653 optimization settings. (This is a good thing in general, and
2654 its introduction in this version should be particularly timely
2655 for anyone whose code fails because of suppression of tail
2657 ** The system now hunts for the C variable "environ" in a more
2658 devious way, to avoid segfaults when the C library version differs
2659 between compile time and run time. (thanks to Christophe Rhodes)
2660 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2661 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2662 ** The compiler no longer issues bogus style warnings for undefined
2663 classes in the same source file as the DEFCLASSes which defined
2664 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2666 ** fixes in CONDITION class precedence list for undefined function
2667 errors (thanks to Alexei Dejneka)
2668 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2669 correctly. (thanks to Dan Barlow)
2670 ** portability fixes aiming at bootstrapping under CLISP (thanks
2671 to Dave McDonald and Christophe Rhodes)
2672 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2673 ** fixes in type translation and and type inference (thanks to
2675 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2676 ** various fixes in the new ports (thanks to Dan Barlow)
2677 * several changes related to debugging:
2678 ** suppression of tail recursion, as noted above
2679 ** stack overflow detection, as noted above
2680 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2681 is now the default. (For some time encapsulation has been more
2682 reliable than the breakpoint-based :ENCAPSULATE NIL
2683 implementation, at least on X86 systems; and I just noticed that
2684 encapsulation also seems closer to the spirit of the ANSI
2687 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2689 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2690 up properly again. (There was a packaging bug in 0.7.0 which
2691 left their definitions in SB-SYS::LOAD-FOREIGN and
2692 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2693 vital for most things which interface to C-level interfaces,
2694 like extensions working with sockets or databases or
2695 Perl-compatible regexes or whatever, and the need to fix
2696 this bug is the main reason that 0.7.1 was released so
2698 ** DEFGENERIC is now choosier about the methods it redefines, so that
2699 reLOADing a previously-LOADed file containing DEFGENERICs does
2700 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2701 cycle now works as it should. (thanks to Alexey Dejneka)
2702 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2703 fixed by Christophe Rhodes. (He actually submitted this patch
2704 months ago, and I delayed until after 0.7.0.)
2705 ** Bug 111 (internal compiler confusion about runtime checks on
2706 FUNCTION types) was fixed by Alexey Dejneka.
2707 * Some internal cleanups (getting rid of variables which aren't
2708 needed now that the byte interpreter is gone) caused the fasl
2709 file format number to change again.
2711 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2712 * major incompatible change: The default fasl file extension, i.e. the
2713 default extension for files produced by COMPILE-FILE, has changed
2714 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2716 ** There are many changes in the implementation of the compiler.
2717 SBCL is now essentially a compiler-only implementation of ANSI
2718 Common Lisp. EVAL still "interprets" a few special cases, but
2719 almost all the interesting cases are handled by creating
2720 a LAMBDA expression, calling COMPILE on it, then calling
2721 FUNCALL on the result.
2722 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2723 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2724 Since the code is newer, there might still be some new bugs
2725 (though not as many as before Martin Atzmueller's fixes:-). But
2726 the new code is substantially simpler and clearer, and hopefully
2727 any remaining bugs will be simpler, less fundamental, and more
2728 fixable then the bugs in the old code.
2729 ** The revised compiler is still a little unsteady on its feet.
2731 *** The debugging information it produces (particularly the names
2732 of FUNCTION objects) is sometimes much less useful than what
2733 the old compiler produced.
2734 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2735 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2736 within a MACROLET) has been temporarily weakened.
2737 ** There are new compiler optimizations for various functions:
2738 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2739 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2740 *** the math functions TRUNCATE, FLOOR, and CEILING
2741 *** the function-of-all-trades COERCE
2742 Mostly these should be transparent, but there's one
2743 potentially-annoying problem (bug 117): when the compiler
2744 inline-expands a function and does type analysis on the result,
2745 it can create control paths which have type mismatches, and
2746 when it can't prove that those control paths aren't taken,
2747 it will issue WARNINGs about the type mismatches. This is
2748 a particular problem in practice for the new sequence functions.
2749 It's not clear how this should be fixed, and for now, a
2750 workaround is given in the entry for 117 in the BUGS file.
2751 ** (Because of the interaction between the two previous items --
2752 occasional inlining problems and new inline expansions -- some
2753 of the new sequence function optimizations won't really kick in
2754 completely until debugging information, and then inlining, are
2755 straightened out in some future version.)
2756 * minor incompatible changes:
2757 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2758 DIRECTORY no longer implicitly promotes NIL slots of its
2759 pathname argument to :WILD. In particular, when you ask for the
2760 contents of a directory (which you used to be able to do without
2761 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2762 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2763 ** changes in behavior that ANSI explicitly defines to be
2764 implementation dependent:
2765 *** The new compiler-only implementation still conforms with ANSI,
2766 but acts a little different than before. Besides the obvious
2767 changes in performance tradeoffs (that the cost per form passed
2768 to EVAL has gone up, and the cost per form executed by EVAL
2769 has gone down), the behavior of the system changes a little
2770 because there are no longer any interpreted function objects.
2771 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2772 e.g. doing COMPILE on the output of interactive DEFUN is
2774 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2776 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2777 has changed from (:CL) to NIL.
2778 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2779 e.g. #\Newline and #\Tab, don't have names specified in the
2780 ANSI Common Lisp standard, is now based on their ASCII symbolic
2781 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2782 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2783 are no longer used for output.
2784 ** changes in internal implementation constants:
2785 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2786 4 million. (If your application spends a lot of time GCing and
2787 you have a lot of RAM, you might want to experiment with
2788 increasing it even more.)
2789 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2790 However, almost all old code should still continue to work without
2791 immediate update, as SB-C-CALL is now a (deprecated) nickname
2793 ** Old operator names in the style DEF-FOO are now deprecated in
2794 favor of new corresponding names DEFINE-FOO, for consistency with
2795 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2796 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2797 internal symbols, but a few supported extensions like
2798 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2799 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2800 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2801 etc. as you get deeper into recursive calls to the debugger
2802 command loop, instead of the old "5]", "5]]", "5]]]"
2803 sequence. (I was motivated to do this when squabbles between
2804 ILISP and SBCL left me very deeply nested in the debugger. In the
2805 short term, this change will probably provoke more ILISP/SBCL
2806 squabbles, but hopefully it will be an improvement in the long run.)
2807 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2808 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2809 function, since declaiming return types involving aliens is
2810 (1) annoyingly messy to do by hand and (2) vital to efficient
2811 compilation of code which calls such functions.
2812 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2813 longer reexported by the SB-EXT package. They're solely useful
2814 for alien code, so it seems more logical that you should get
2815 them from the SB-ALIEN package, not in SB-EXT.
2816 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2817 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2818 features. Instead, the code that they used to control is always
2819 built into the system.
2820 * many other bug fixes
2821 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2822 advantage of the new EVAL-WHEN stuff and to clean them up in
2823 general, and they are now more ANSI-compliant in a number of
2824 ways. Martin Atzmueller is responsible for a lot of this.
2825 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2827 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2828 *** correct ERROR type for various file operations
2829 *** some fixes for Lisp streams
2830 *** DEFMETHOD syntax checking
2831 *** changing old weird representation of debug information as
2832 strings (which, among their other deficiencies, don't transform
2833 correctly when you rename packages, and don't change their
2834 print representation when you change things like *PACKAGE*
2835 and *PRINT-LENGTH*) to symbols and lists of symbols
2836 He also made several improvements and fixed several bugs in DESCRIBE.
2837 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2839 *** misbehavior of WRITE-STRING/WRITE-LINE
2840 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2841 and several other LOOP problems as well
2842 *** DIRECTORY when similar filenames are present
2843 *** DEFGENERIC with :METHOD options
2844 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2845 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2846 *** argument ordering in FIND with :TEST option
2847 *** mishandled package designator argument in APROPOS-LIST
2848 *** various problems in the backquote readmacro
2849 *** a bug in APROPOS
2850 *** probably some others that I'm not describing very well here,
2851 since the CVS log documents them by reference to sbcl-devel
2852 messages, and the SourceForge archives aren't working well.:-(
2853 ** Dan Barlow improved the Alpha port (and is making progress on the
2854 PPC port, for those of you who think different).
2855 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2856 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2857 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2858 continues to work on the SPARC port (for those of you in a position
2859 to look down upon our little PC-compatible boxes from a great height).
2860 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2861 and uses that copy, rather than the current dynamic value, when
2862 it's trying to decide whether to truncate output. Thus e.g.
2863 (let ((*print-lines* 50))
2864 (pprint-logical-block (stream nil)
2866 (let ((*print-lines* 8))
2867 (print (aref possiblybigthings i) stream)))))
2868 should now truncate the logical block only at 50 lines, instead of
2869 often truncating it at 8 lines, as it did before.
2870 * The doc/cmucl/ directory, containing old CMU CL documentation
2871 from the time of the fork, is no longer part of the base system.
2872 SourceForge has shut down its anonymous FTP service, and with it
2873 my original plan for distributing the old CMU CL documentation
2874 there. For now, if you need these files you can download an old
2875 SBCL source release and extract them from it.
2876 * The fasl file version number changed again, for dozens of reasons,
2877 some of which are apparent above.
2879 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2880 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2881 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2882 patch, so that DISASSEMBLE output is much nicer.
2883 * The code in the SB-PROFILE package now seems reasonably stable.
2884 I still haven't decided what the final interface should look like
2885 (I'd like PROFILE to interact cleanly with TRACE, since both
2886 facilities use function encapsulation) but if you have a need
2887 for profiling now, you can probably use it successfully with
2888 the current CMU-CL-style interface.
2889 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2890 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2891 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2892 link now returns the dangling link itself, and for similar
2893 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2894 link itself. (In these cases the old code signalled an error and
2895 looped endlessly, respectively.) Thus, DIRECTORY now works even
2896 in the presence of dangling and cyclic symbolic links.
2897 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2898 is now a supported extension again, since the consensus on
2899 sbcl-devel was that it can be useful for ordinary development
2900 work, not just for debugging SBCL itself.
2901 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2902 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2903 that currently-defined functions might be redefined later with
2904 different return types.
2905 * Hash tables can be printed readably, as inspired by CMU CL code
2906 of Eric Marsden and SBCL code of Martin Atzmueller.
2907 * better error handling in CLOS method combination, thanks to
2908 Martin Atzmueller porting Pierre Mai's CMU CL patches
2909 * more overflow fixes for >16Mbyte I/O buffers
2910 * A bug in READ has been fixed, so that now a single Ctrl-D
2911 character suffices to cause end-of-file on character streams.
2912 In particular, now you only need one Ctrl-D at the command
2913 line (not two) to exit SBCL.
2914 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2916 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2917 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2918 does the right thing.
2919 * The implementation of some type tests, especially for CONDITION
2920 types, is now tidier and maybe faster, due to CMU CL code
2921 originally by Douglas Crosher, ported by Martin Atzmueller.
2922 * Some math functions have been fixed, and there are new
2923 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2924 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2925 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2926 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2927 likely to be built into sbcl-0.7.0.)
2928 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2929 CL, ported to SBCL by Martin Atzmueller)
2930 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2931 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2932 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2933 * There's a new slam.sh hack to shorten the edit/compile/debug
2934 cycle for low-level changes to SBCL itself, and a new
2935 :SB-AFTER-XC-CORE target feature to control the generation of
2936 the after-xc.core file needed by slam.sh.
2937 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2938 COMPILE-FILE is no longer supported, so that now every function
2939 gets an entry point, so that block compilation looks a little
2940 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2941 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2942 returns the number of bytes consed since the system started,
2943 rather than the number consed since the first time the function
2944 was called. (The new definition parallels ANSI functions like
2945 CL:GET-INTERNAL-RUN-TIME.)
2946 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2947 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2948 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2949 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2950 * Fasl file version numbers are now independent of the target CPU,
2951 since historically most system changes which required version
2952 number changes have affected all CPUs equally. Similarly,
2953 the byte fasl file version is now equal to the ordinary
2956 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2957 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2958 is no longer recognized. I apologize for this, because it was
2959 listed in SB-EXT as a supported extension, but I found that
2960 its existing behavior was poorly specified, as well as incorrectly
2961 specified, and it looked like too much of a mess to straighten it
2962 out. I have enough on my hands trying to get ANSI stuff to work..
2963 * many patches ported from CMU CL by Martin Atzmueller, with
2964 half a dozen bug fixes in pretty-printing and the debugger, and
2965 half a dozen others elsewhere
2966 * fixed bug 13: Floating point infinities are now supported again.
2967 They might still be a little bit flaky, but thanks to bug reports
2968 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2969 as flaky as they were.
2970 * The --noprogrammer command line option is now supported. (Its
2971 behavior is slightly different in detail from what the old man
2972 page claimed it would do, but it's still appropriate under the
2973 same circumstances that the man page talks about.)
2974 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2975 are now supported, and enabled by default. Thus, the compiler can
2976 handle many floating point and complex operations much less
2977 inefficiently. (Thus e.g. you can implement a complex FFT
2979 * The compiler now detects type mismatches between DECLAIM FTYPE
2980 and DEFUN better, and implements CHECK-TYPE more correctly, and
2981 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2982 thanks to patches from Martin Atzmueller.
2983 * various fixes to make the cross-compiler more portable to
2984 ANSI-conforming-but-different cross-compilation hosts (notably
2985 Lispworks for Windows, following bug reports from Arthur Lemmens)
2986 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2987 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2988 fixed thanks to Pierre Mai's CMU CL patches.
2989 * fixes to keep the system from overflowing internal counters when
2990 it tries to use i/o buffers larger than 16M bytes
2991 * fixed bug 45a: Various internal functions required to support
2992 complex special functions have been merged from CMU CL sources.
2993 (When I was first setting up SBCL, I misunderstood a compile-time
2994 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2995 * improved support for type intersection and union, fixing bug 12
2996 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2997 more obscure bugs as well
2998 * some steps toward byte-compiling non-performance-critical
2999 parts of the system, courtesy of patches from Martin Atzmueller
3000 * Christophe Rhodes has made some debian packages of sbcl at
3001 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3002 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3003 stable, but are nonetheless usable. When he's ready, I'd be happy
3004 to add them to the SourceForge "File Releases" section. (And if
3005 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3006 * new fasl file format version number (because of changes in
3007 internal representation of (OR ..) types to accommodate the new
3008 support for (AND ..) types, among other things)
3010 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3011 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3013 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3015 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3016 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3017 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3018 * DESCRIBE now works on structure objects again.
3019 * Most function call argument type mismatches are now handled as
3020 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3021 know whether the function will be redefined before the call is
3022 executed. (The compiler could flag local calls with full WARNINGs,
3023 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3024 it doesn't keep track of enough information to know whether calls
3025 are local in this sense.)
3026 * Compiler output is now more verbose, with messages truncated
3027 later than before. (There should be some supported way for users
3028 to override the default verbosity, but I haven't decided how to
3029 provide it yet, so this behavior is still controlled by the internal
3030 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3031 src/compiler/ir1util.lisp.)
3032 * Fasl file format version numbers have increased again, because
3033 support for the Gray streams extension changes the layout of the
3034 system's STREAM objects.
3035 * The Gray subclassable streams extension now works, thanks to a
3036 patch from Martin Atzmueller.
3037 * The full LOAD-FOREIGN extension (not just the primitive
3038 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3039 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3040 but like most other programs, it defaults to copying the Unix
3041 environment from the original process instead of starting the
3042 new process in an empty environment.
3043 * Extensions which manipulate the Unix environment now support
3044 an :ENVIRONMENT keyword option which doesn't smash case or
3045 do other bad things. The CMU-CL-style :ENV option is retained
3046 for porting convenience.
3047 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3048 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3050 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3052 * A patch from Martin Atzmueller seems to have solved the SIGINT
3053 problem, and as far as we know, signal-handling now works cleanly.
3054 (If you find any new bugs, please report them!)
3055 * The system no longer defaults Lisp source file names to types
3056 ".l", ".cl", or ".lsp", but only to ".lisp".
3057 * The compiler no longer uses special default file extensions for
3058 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3059 seems to expect a single default extension for all compiled code,
3060 and there's no compelling reason to try to stretch the standard
3061 to allow two different extensions.) Instead, byte-compiled files
3062 default to the same extension as native-compiled files.
3063 * Fasl file format version numbers have increased again, because
3064 a rearrangement of internal implementation packages made some
3065 dumped symbols in old fasl files unreadable in new cores.
3066 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3067 many fewer weird special cases.
3068 * Bug #17 (differing COMPILE-FILE behavior between logical and
3069 physical pathnames) has been fixed, and some related misbehavior too,
3070 thanks to a patch from Martin Atzmueller.
3071 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3072 by Tim Moore, ported to SBCL by Martin Atzmueller.
3073 * Martin Atzmueller fixed several filesystem-related problems,
3074 including bug #36, in part by porting CMU CL patches, which were
3075 written in part by Paul Werkowski.
3076 * More compiler warnings in src/runtime/ are gone, thanks to
3077 more patches from Martin Atzmueller.
3078 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3081 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3083 * DESCRIBE now works on CONDITION objects.
3084 * The debugger now handles errors which arise when trying to print
3085 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3087 * The build system now uses an additional file, customize-target-features.lisp,
3088 to allow local modifications to the target *FEATURES* list. (The point of
3089 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3090 debugging features enabled, without having to worry about propagating it
3091 into everyone's system when I do a "cvs update".) When no
3092 customize-target-features.lisp file exists, the target *FEATURES* list
3093 should be constructed the same way as before.
3094 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3095 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3096 warn about behavior which is undefined under the ANSI spec. Note: This
3097 is specified by ANSI, but it's not very popular with programmers.
3098 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3099 macro in the SBCL sources for an example of a workaround which you
3100 might use to make portable ANSI-standard code which does what you want.
3101 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3102 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3103 when it's used as a non-toplevel form. (This is required in order
3104 to implement the DEFCONSTANT-EQX macro.)
3105 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3106 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3107 funny things, due to bugs in the implementation of EVAL-WHEN
3108 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3109 non-toplevel DEFCONSTANTs are uncommon.)
3110 * The core file version number and fasl file version number have been
3111 incremented, because the old noncompliant DEFCONSTANT behavior involved
3112 calling functions which no longer exist, and because I also took the
3113 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3114 * fixed bug 1 (error handling before read-eval-print loop starts), and
3115 redid debugger restarts and related debugger commands somewhat while
3117 ** The QUIT debugger command is gone, since it did something
3118 rather different than the SB-EXT:QUIT command, and since it never
3119 worked properly outside the main toplevel read/eval/print loop.
3120 Invoking the new TOPLEVEL restart provides the same functionality.
3121 ** The GO debugger command is also gone, since you can just invoke
3122 the CONTINUE restart directly instead.
3123 ** The TOP debugger command is also gone, since it's redundant with the
3124 FRAME 0 command, and since it interfered with abbreviations for the
3126 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3128 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3129 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3130 some porting work by Martin Atzmueller)
3131 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3132 with a boatload of patches which he ported to SBCL. Now that those
3134 ** The system tries to make sure that its low-priority messages
3135 are prefixed by semicolons, to help people who like to use
3136 syntax highlighting in their ILISP buffer. (This patch
3137 was originally due to Raymond Toy.)
3138 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3139 patches originally written by Raymond Toy.
3140 ** The compiler understands coercion between single-value and
3141 multiple-VALUES type expressions better, getting rid of some very
3142 weird behavior, thanks to patches originally by Robert MacLachlan
3143 and Douglas Crosher.
3144 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3145 lambda lists, thanks to a patch originally by Pierre Mai.
3146 ** The system no longer bogusly warns about "abbreviated type
3148 ** The compiler gets less confused by inlining and RETURN-FROM,
3149 thanks to some patches originally by Tim Moore.
3150 ** The system no longer hangs when dumping circular lists to fasl
3151 files, thanks to a patch originally from Douglas Crosher.
3152 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3153 undefined function error.
3154 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3155 in structure slot initforms) for now, documented workaround instead:-|
3156 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3157 * fixed bug 5: added stubs for various Gray stream functions called
3158 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3159 installed, at least appropriate type errors are generated
3160 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3161 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3162 consistently in DEFMETHOD forms.
3163 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3164 it doesn't seem to affect SBCL after all
3165 * The C runtime system now builds with better optimization and many
3166 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3168 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3170 * The system is now under CVS at SourceForge (instead of the
3171 CVS repository on my home machine).
3172 * The new signal handling code has been tweaked to treat register
3173 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3174 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3175 downstream, e.g. in debug-int.lisp, has implicit dependencies
3176 on the unsignedness of integer representation of machine words,
3177 and that caused the system to bomb out with infinite regress
3178 when trying to recover from type errors involving signed values,
3179 e.g. (BUTLAST '(1 2 3) -1).)
3180 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3181 (This was one of the bugs Peter Van Eynde reported back in July.)
3182 * The system now uses code inspired by Colin Walters' O(N)
3183 implementation of MAP (from the cmucl-imp@cons.org mailing
3184 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3185 inline the MAP operation, and there is more than one
3186 sequence argument to the MAP call (so that it can't just
3187 do ETYPECASE once and for all based on the type of the
3188 single sequence argument). (The old non-inline implementation
3189 of the general M-argument sequence-of-length-N case required
3190 O(M*N*N) time when any of the sequence arguments were LISTs.)
3191 * The QUIT :UNIX-CODE keyword argument has been renamed to
3192 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3193 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3195 * Raymond Wiker's patches to port dynamic loading from Linux to
3196 FreeBSD have been added.
3197 * The BUGS file is now more nearly up to date, thanks in large part
3198 to Martin Atzmueller's review of it.
3199 * The debugger now flushes standard output streams before it begins
3200 its output ("debugger invoked" and so forth).
3201 * The core version number and fasl file version number have both
3202 been incremented, because of incompatible changes in the layout
3204 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3205 used to not be called for a saved Lisp image.)
3206 * Martin Atzmueller's version of a patch to fix a compiler crash,
3207 as posted on sbcl-devel 13 September 2000, has been installed.
3208 * Instead of installing Martin Atzmueller's patch for the
3209 compiler transform for SUBSEQ, I deleted the compiler transform,
3210 and transforms for some similar consing operations.
3211 * A bug in signal handling which kept TRACE from working on OpenBSD
3213 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3214 away by constant folding
3215 * The system now defines its address space constants in one place
3216 (in the Lisp sources), and propagates them automatically elsewhere
3217 (through GENESIS and the sbcl.h file). Therefore, patching the
3218 address map is less unnecessarily tedious and error-prone. The
3219 Lisp names of address space constants have also been systematized.
3220 * CVS tags like dollar-Header-dollar have been removed from
3221 the sources, because they have never saved me trouble and
3222 they've been source of trouble working with patches and other
3223 diff-related operations.
3224 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3225 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3227 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3229 * The system has been ported to OpenBSD.
3230 * The system now compiles with a simple "sh make.sh" on the systems
3231 that it's supported on. I.e., now you no longer need to tweak
3232 text in the target-features.lisp-expr and symlinks in src/runtime/
3233 by hand, the make.sh takes care of it for you.
3234 * The system is no longer so grossly inefficient when compiling code
3235 involving vectors implemented as general (not simple) vectors (VECTOR T),
3236 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3237 which dares to use the various sequence functions on non-simple
3238 vectors, takes less of a performance hit.
3239 * There is now a primitive type predicate VECTOR-T-P
3240 to test for the (VECTOR T) type, so that e.g.
3241 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3242 can now be compiled with some semblance of efficiency. (The old code
3243 turned the type declaration into a full call to %TYPEP at runtime!)
3244 * AREF on (VECTOR T) is still not fast, since it's still compiled
3245 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3246 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3247 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3248 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3249 without making full calls to SUBTYPEP at runtime.
3250 (Some analogous efficiency issues for non-simple vectors specialized to
3251 element types other than T, or for non-simple multidimensional arrays,
3252 have not been addressed. They could almost certainly be handled the
3253 same way if anyone is motivated to do so.)
3254 * The changes in array handling break binary compatibility, so
3255 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3256 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3257 instead of (VALUES T T).
3258 * By following the instructions that Dan Barlow posted to sbcl-devel
3259 on 2 July 2000, I was able to enable primitive dynamic object
3260 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3261 functionality is not implemented (since it calls ld to resolve
3262 library references automatically, requiring RUN-PROGRAM for its
3263 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3264 try to resolve library references) is now supported.
3265 * The system now flushes the standard output streams when it terminates,
3266 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3267 them at several other probably-convenient times, e.g. in each pass of
3268 the toplevel read-eval-print loop, and after evaluating a form given
3269 as an "--eval" command-line option. (These changes were motivated by a
3270 discussion of stream flushing issues on cmucl-imp in August 2000.)
3271 * The source transform for TYPEP of array types no longer assumes
3272 that an array whose element type is a not-yet-defined type
3273 is implemented as an array of T, but instead punts, so that the
3274 type will be interpreted at runtime.
3275 * There is now some support for cross-compiling in make.sh: each of
3276 the phases of make.sh has its own script. (This should be transparent
3277 to people doing ordinary, non-cross-compile builds.)
3278 * Since my laptop doesn't have hundreds of megabytes of memory like
3279 my desktop machine, I became more motivated to do some items on
3280 my to-do list in order to reduce the size of the system a little:
3281 ** Arrange for various needed-only-at-cold-init things to be
3282 uninterned after cold init. To support this, those things have
3283 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3284 symbols with such names are now uninterned after cold init).
3285 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3286 fasl files for cold load.
3287 ** Remove the old compiler structure pooling code (which used to
3288 be conditional on the target feature :SB-ALLOC) completely.
3289 ** Redo the representation of some data in cold init to be more compact.
3290 (I also looked into supporting byte compiled code at bootstrap time,
3291 which would probably reduce the size of the system a lot, but that
3292 looked too complicated, so I punted for now.)
3293 * The maximum signal nesting depth in the src/runtime/ support code has
3294 been reduced from 4096 to 256. (I don't know any reason for the very
3295 large old value. If the new smaller value turns out to break something,
3296 I'll probably just bump it back up.)
3297 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3299 * Many, many bugs reported by Peter Van Eynde have been added to
3300 the BUGS list; some have even been fixed.
3301 * While enabling dynamic object file loading, I tried to make the
3302 code easier to understand, renaming various functions and variables
3303 with less ambiguous names, and changing some function calling
3304 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3305 * While trying to figure out how to do the OpenBSD port, I tried to
3306 clean up some of the code in src/runtime/. In particular, I dropped
3307 support for non-POSIX signal handling, added various comments,
3308 tweaked the code to reduce the number of compilation warnings, and
3309 renamed some files to increase consistency.
3310 * To support the new automatic configuration functionality in make.sh,
3311 the source file target-features.lisp-expr has been replaced with the
3312 source file base-target-features.lisp-expr and the machine-generated
3313 file local-target-features.lisp-expr.
3314 * fixed a stupid quoting error in make.sh so that using CMU CL
3315 "lisp -batch" as cross-compilation host works again
3317 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3319 * DESCRIBE no longer tries to call itself recursively to describe
3320 bound/fbound values, so that it no longer fails on symbols which are
3321 bound to themselves (like keywords, T, and NIL).
3322 * DESCRIBE now works on generic functions.
3323 * The printer now prints less-screwed-up representations of closures
3324 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3325 * A private symbol is used instead of the :EMPTY keyword previously
3326 used to mark empty slots in hash tables. Thus
3327 (DEFVAR *HT* (MAKE-HASH-TABLE))
3328 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3329 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3330 now does what ANSI says that it should. (You can still get
3331 similar noncompliant behavior if bang on the hash table
3332 implementation with all the symbols you get back from
3333 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3334 This breaks binary compatibility, since tests for equality to
3335 :EMPTY are wired into things like the macroexpansion of
3336 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3338 * There's now a minimal placeholder implementation for CL:STEP,
3339 as required by ANSI.
3340 * An obscure bug in the interaction of the normal compiler, the byte
3341 compiler, inlining, and structure predicates has been patched
3342 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3343 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3345 * Missing ordinary arguments in a macro call are now detected even
3346 when the macro lambda list contains &KEY or &REST.
3347 * The debugger no longer complains about encountering the top of the
3348 stack when you type "FRAME 0" to explicitly instruct it to go to
3349 the top of the stack. And it now prints the frame you request even
3350 if it's the current frame (instead of saying "You are here.").
3351 * As specified by ANSI, the system now always prints keywords
3352 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3354 * The default initial SIZE of HASH-TABLEs is now smaller.
3355 * Type information from CLOS class dispatch is now propagated
3356 into DEFMETHOD bodies, so that e.g.
3357 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3359 is now basically equivalent to
3360 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3361 (DECLARE (TYPE SINGLE-FLOAT X))
3363 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3364 operation, without having to do run-time type dispatch.
3365 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3366 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3367 * Since I don't seem to be making big file reorganizations very often
3368 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3369 is overflowing my ability to conveniently back them up), I've finally
3370 checked the system into CVS. (The CVS repository is on my home system,
3371 not at SourceForge -- putting it on SourceForge might come later.)
3372 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3373 high-level GC-NOTIFY-FOO functions send their output. (There's
3374 still very little control of where low-level verbose GC functions
3375 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3376 less than it used to -- the GC-NOTIFY-FOO functions are now under
3377 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3378 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3379 in only one place in the source code, and propagates it automatically
3380 everywhere that it's needed. Thus e.g. when I bump the version from
3381 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3382 * The C source files now include boilerplate legalese and documentation
3383 at the head of each file (just as the Lisp source files already did).
3384 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3385 to his page will be replaced with a link to his new CLiki service.
3387 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3389 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3390 * The build process now looks for GNU make under the default name "gmake",
3391 instead of "make" as it used to. If GNU make is not available as "gmake"
3392 on your system, you can change this default behavior by setting the
3393 GNUMAKE environment variable.
3394 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3395 can build without error under CMU CL.
3397 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3399 * There is now a partial SBCL user manual (with some new text and some
3400 text cribbed from the CMU CL manual).
3401 * The beginnings of a profiler have been added (starting with the
3402 CMU CL profiler and simplifying and cleaning up). Eventually the
3403 main interface should be through the TRACE macro, but for now,
3404 it's still accessed through vaguely CMU-CL-style functions and macros
3405 exported from the package SB-PROFILE.
3406 * Some problems left over from porting CMU CL to the new
3407 cross-compilation bootstrap process have been cleaned up:
3408 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3409 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3410 part when getting rid of the compiler *BACKEND* stuff.)
3411 ** The value of *NULL-TYPE* was screwed up, because it was
3412 being initialized before the type system knew the final
3413 definition of the 'NULL type. This screwed up several key
3414 optimizations in the compiler, causing inefficiency in all sorts
3415 of places. (I found it because I wanted to understand why
3416 GET-INTERNAL-RUN-TIME was consing.)
3417 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3418 PROCLAIM FTYPE information. Unfortunately this broke binary
3419 compatibility again, since now the forms output by DEFGENERIC
3420 to refer to functions which didn't exist in 0.6.3.
3421 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3422 can use the new (as of 0.6.3) transform for SOME into MAP into
3424 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3425 values to (INTEGER 0 1000000), so that the time code will no longer
3426 occasionally get blown up by Linux returning 1000000 microseconds
3427 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3428 its output conform to the ANSI spec. (Alas, this makes its output
3429 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3430 * A full call to MAP NIL with a single sequence argument no longer conses.
3431 * fixes to problems pointed out by Martin Atzmueller:
3432 * The manual page no longer talks about multiprocessing as though
3433 it were currently supported.
3434 * The ILISP support patches have been removed from the distribution,
3435 because as of version 5.10.1, ILISP now supports SBCL without us
3436 having to maintain patches.
3437 * added a modified version of Raymond Toy's recent CMU CL patch for
3438 EQUALP comparison of HASH-TABLE
3440 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3442 * The system still can't cross-compile itself with
3443 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3444 entails), but at least it can compile more of itself that way
3445 than it used to be able to, and various buglets which were uncovered
3446 by trying to cross-compile itself that way have now been fixed.
3447 * This release breaks binary compatibility again. This time
3448 at least I've incremented the FASL file format version to 2, so that the
3449 problem can be detected reliably instead of just causing weird errors.
3450 * various new style warnings:
3451 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3452 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3453 ** using the lexical binding of a variable named in the *FOO* style
3454 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3455 as specified by ANSI.
3456 * *RANDOM-STATE* is no longer automatically initialized from
3457 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3458 default behavior of the system is to repeat its behavior every time
3459 it's run. If you'd like to change this behavior, you can always
3460 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3461 old convention there was no comparably easy way to get the system to
3462 repeat its behavior every time it was run.
3463 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3464 FTYPE declarations has been removed, in favor of their ANSI
3465 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3466 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3467 the types of their sequence arguments can be determined at compile time.
3468 This is done through a new open code expansion for MAP which eliminates
3469 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3470 argument types can be determined at compile time.
3471 * The optimizer now transforms COERCE into an identity operation when it
3472 can prove that the coerced object is already of the correct type. (This
3473 can be a win for machine generated code, including the output of other
3474 optimization transforms, such as the MAP transform above.)
3475 * Credit information has been moved from source file headers into CREDITS.
3476 * Source file headers have been made more standard.
3477 * The CASE macro now compiles without complaining even when it has
3480 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3482 * (Note that the way that the PCL macroexpansions were rewritten
3483 to accommodate the change in DEFGENERIC below breaks binary
3484 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3485 not run under sbcl-0.6.2. Once we get out of alpha releases,
3486 i.e. hit release 1.0.0, we'll probably try to maintain binary
3487 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3488 and sbcl-1.4.4. Until then, however, it might be fairly common
3489 for maintenance releases to break binary compatibility.)
3490 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3492 * The reporting of unhandled signals has been changed to print some
3493 explanatory text as well as the report form. (Previously only
3494 the report form was printed.)
3495 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3496 it defines, so that the compiler no longer issues undefined function
3497 warnings for compiled-but-not-yet-loaded generic functions.
3498 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3499 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3500 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3501 section of the ANSI spec are supported.
3502 * The "" nickname for the "KEYWORD" package has been removed.
3503 The reader still handles symbol tokens which begin with a package marker
3504 as keywords, but it doesn't expose its mechanism for doing so in the
3505 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3506 * The system now issues STYLE-WARNINGs for contradictory TYPE
3507 proclamations. (Warnings for contradictory FTYPE proclamations would
3508 be nice too, but those can't be done usefully unless the type system
3509 is made smarter about FUNCTION types.)
3510 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3511 systematized, so that "*target-*.lisp is supposed to exist only on the
3512 target and imply that there's a related file which exists on the
3513 host, and *host-*.lisp is supposed to exist only on the host and imply
3514 that there's a related file which exists on the target. This involves a
3515 lot of renaming. Hopefully the acute confusion caused by the renaming
3516 will be justified by the reduction in chronic confusion..
3517 ** runtime-type.lisp -> early-target-type.lisp
3518 ** target-type.lisp -> late-target-type.lisp
3519 ** early-host-format.lisp -> early-format.lisp
3520 ** late-host-format.lisp -> late-format.lisp
3521 ** host-error.lisp -> misc-error.lisp
3522 ** early-error.lisp -> early-target-error.lisp
3523 ** late-error.lisp -> late-target-error.lisp
3524 ** host-defboot.lisp -> early-defboot.lisp
3525 ** code/misc.lisp -> code/target-misc.lisp
3526 ** code/host-misc.lisp -> code/misc.lisp
3527 ** code/numbers.lisp -> code/target-numbers.lisp
3528 ** code/early-numbers.lisp -> numbers.lisp
3529 ** early-host-type.lisp -> early-type.lisp
3530 ** late-host-type.lisp -> late-type.lisp
3531 ** host-typep.lisp -> typep.lisp
3532 ** load.lisp -> target-load.lisp
3533 ** host-load.lisp -> load.lisp
3534 ** host-disassem.lisp -> disassem.lisp
3535 ** host-insts.lisp -> insts.lisp
3536 ** byte-comp.lisp -> target-byte-comp.lisp
3537 ** host-byte-comp.lisp -> byte-comp.lisp
3538 ** host-signal.lisp -> signal.lisp
3539 ** host-defstruct.lisp -> defstruct.lisp
3540 ** late-target-type.lisp -> deftypes-for-target.lisp
3541 Furthermore, several other previously target-only files foo.lisp (e.g.
3542 hash-table.lisp and random.lisp) have been split into a target-and-host
3543 foo.lisp file and a target-only target-foo.lisp file, with their key type
3544 definitions in the target-and-host part, so that the cross-compiler will
3545 know more about target types.
3546 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3547 gone away. In their place are various *BACKEND-FOO* variables
3548 corresponding to the slots of the old structure.
3549 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3550 into the target SBCL has been fixed.
3551 * The chill.lisp system for loading cold code into a running SBCL
3553 * Support for the CMU CL "scavenger hook" extension has been removed.
3554 (It was undocumented and unused in the CMU CL sources that SBCL was
3555 derived from, and stale in sbcl-0.6.1.)
3556 * Various errors in the cross-compiler type system were detected
3557 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3558 (enabling various consistency checks). Many of them were fixed,
3559 but some hard problems remain, so the compiler is back to
3560 running without *TYPE-SYSTEM-INITIALIZED* for now.
3561 * As part of the cross-compiler type system cleanup, I implemented
3562 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3563 * I have started adding UNCROSS calls throughout the type system
3564 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3565 on cross-compiler input files will be able to go away, and various
3567 * CONSTANTP now returns true for quoted forms (as explicitly required
3570 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3572 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3573 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3574 (MAKE-LIST -1). In the longer term, it ought to become true
3575 that declarations are assertions even at SAFETY 1. For now, it's not
3576 quite true even at SAFETY 3, but it's at least more nearly true..
3577 (Note that this change seems to increases the size of the system by
3578 O(5%) and to decrease the speed of the compiler by 20% or more.)
3579 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3580 for the problem of printing dozens of lines of distracting information
3581 about low-level system machinery as part of the top stack frame
3582 on entry to the debugger when an undefined function was called.
3583 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3584 is not reset to COMMON-LISP-USER.
3585 * Compilation of stuff related to dyncount.lisp has been made conditional
3586 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3587 smaller. The various dyncount-related symbols have been moved into
3588 a new "SB-DYNCOUNT" package.
3589 * tty-inspect.lisp has been renamed to inspect.lisp.
3590 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3591 feature has gone away. (When we eventually port to other flavors of
3592 libc and/or Unix, we'll try to make the differences between flavors
3593 invisible at the user level.)
3594 * Various other *FEATURES* tags, and/or their associated conditionals,
3595 have been removed if obsolescent, or given better documentation, or
3596 sometimes given more-mnemonic names.
3598 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3600 * tidied up "make.sh" script
3601 * tidied up system directory structure
3602 * better "clean.sh" behavior
3603 * added doc/FOR-CMUCL-DEVELOPERS
3604 * many many small tweaks to output format, e.g. removing possibly-confusing
3605 trailing #\. character in DESCRIBE-INSTANCE
3606 * (EQUALP #\A 'A) no longer signals an error.
3607 * new hashing code, including EQUALP hashing
3608 * tidied up Lisp initialization and toplevel
3609 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3610 * command line argument processing
3611 * added POSIX-GETENV function to deal with Unix-ish environment variables
3612 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3613 terminating SBCL on EOF
3614 * non-verbose GC by default
3615 * There is no more "sbcl" shell script; the sbcl file is now the C
3616 runtime executable (just like CMU CL).
3617 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3619 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3620 symbol and package information as Lisp native symbol and package objects
3621 instead of strings naming symbols and strings naming packages. This way,
3622 whenever packages are renamed (as in warm init), debug information is
3623 transformed along with everything else.
3624 * tweaked the optimization policy declarations which control the building
3625 of SBCL itself. Now, among other things, the system no longer saves
3626 source location debugging information. (This helps two problems at once
3627 by reducing SBCL size and by keeping SBCL from trying to look for its
3628 sources -- which may not exist -- when reporting errors.)
3629 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3630 debugging and testing purposes
3631 * cleaned up printing, making the printer call PRINT-OBJECT for
3632 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3633 methods, giving nearly-ANSI behavior
3634 * converted almost all special variables to use *FOO* naming convention
3635 * deleted PARSE-TIME functionality, since it can be done portably
3636 * moved some files out of cold init into warm init
3637 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3639 * regularized formatting of source files
3640 * added an install.sh script
3641 * fixed ridiculous memory usage of cross-compiler by making
3642 compiler/alloc.lisp not try to do pooling unless it can hook
3643 itself into the GC of the cross-compilation host. Now the system
3644 builds nicely on my old laptop.
3645 * added :SB-ALLOC in target-features.lisp-expr
3646 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3647 was not implemented)
3648 * re-did condition handling and note reporting in the compiler. Notes
3649 are no longer handled by signalling conditions. Style warnings
3650 and warnings are handled more correctly and reported in such a way
3651 that it's easy to find one or the other in your output (so that you
3652 can e.g. figure out which of many problems caused COMPILE-FILE to
3654 * changed the severity of several compiler warnings from full WARNING
3655 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3656 compiler note reporting so that it doesn't use the condition system
3657 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3658 COMPILE-FILE command)
3659 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3660 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3661 no longer supported, since their implementation was deeply intertwingled
3662 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3663 * removed bogus "support" for compiler macros named (SETF FOO), and
3664 removed the compiler macro for SETF INFO (but only after making a fool
3665 of myself on the cmucl-imp mailing list by posting a bogus patch for
3666 DEFINE-COMPILER-MACRO..)
3667 * Compiled files containing forms which have side effects on the Lisp
3668 reader (such as DEFPACKAGE forms) are now handled more correctly.
3669 (Compiler queuing of top level lambdas has been suppressed by setting
3670 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3671 * deleted various currently-unused source files, e.g. gengc.lisp. They
3672 may be added back at some point e.g. when porting to other architectures,
3673 but until they are it's distracting to distribute them and to try to
3675 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3676 there were so many of them they're just distractions, and UNCROSS is
3677 known to be able to handle the current sources
3678 * moved PROFILE functionality into TRACE, so that it will be clear
3679 how the wrapping and unwrapping of functions when you profile them
3680 interacts with the wrapping and unwrapping of functions when you
3681 trace them. (Actually, the functionality isn't there yet, but at least
3682 the interface specification is there. Hopefully, the functionality will
3683 arrive with some future maintenance release.)
3684 * removed host-oops.lisp
3685 * changed signature of QUIT function to allow UNIX-CODE argument
3686 * fixed READ-SEQUENCE bug
3687 * tweaked verbose GC output so that it looks more like the progress
3688 output that ANSI specifies for functions like LOAD
3689 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3690 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob