1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
3 * new platform: experimental support for SBCL x86/Darwin, including
4 MacOS X 10.4.5 on Intel.
5 * new feature: Unicode character names are now known to the system
6 (through CHAR-NAME and NAME-CHAR).
7 * minor incompatible change: the contrib modules SB-POSIX and
8 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
9 this change is to make it easier to distribute
10 location-independent binaries.
11 * bug fix: as implied by AMOP, standardized classes no longer have
12 slots named by external symbols of public packages. (reported by
14 * optimization: calling functions via constant symbols -- as in
15 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
16 via the function object as in (FUNCALL #'FOO).
17 * optimization: on x86 and x86-64, the calling convention for the
18 case of dealing with an unknown number of values has been altered
19 to be friendlier to the prediction heuristics implemented,
20 particularly with reference to CALL and RET pairing. (thanks to
22 * enhancement: CONSTANTP is now able to determine constantness of
23 more complex forms, including calls to constant-foldable standardized
24 functions and some special forms beyond QUOTE.
25 * fixed bug: occasional GC crashes on Solaris/x86
27 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
28 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
29 be used for bundling the runtime and the core file into one
30 executable binary. This feature is not currently supported on all SBCL
31 platforms. (thanks to James Bielman and NIIMI Satoshi)
32 * new feature: a generational or ephemeral garbage collector is now
33 the default on the PowerPC platform (both Linux and Darwin). The
34 old Cheney (stop and copy) collector is a build-time option.
35 (thanks to Cyrus Harmon, after Raymond Toy)
36 * minor incompatible change: the method by which SBCL finds its
37 contributed modules has changed; it no longer relies on symbolic
38 links from an $SBCL_HOME/systems directory, but searches directly
39 in the subdirectories of $SBCL_HOME.
40 * enhancement: the dynamic heap size on the Linux/PPC platform is
41 markedly larger, even using the older Cheney garbage collector.
42 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
43 certain complicated string types. (reported by Gary King)
44 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
45 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
46 * fixed bug: DOCUMENTATION on structures no longer signals an
47 error if no structure type of the right name exists. (reported by
49 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
50 applicable non-standard primary or :AROUND method on
51 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
52 no longer cause unbound variable errors. (reported by Kevin Reid)
53 * fixed bug: metacircle resolution in cases where methods have slots
54 added before the slots from STANDARD-METHOD. (reported by Jean
56 * fixed bug: the Power PC assembly code for calling into lisp
57 sometimes computed its return address wrongly (depending on the
58 15th bit of the address, fixed at link time). This used to
59 manifest itself in a segmentation violation while building PCL.
60 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
61 many others over the years)
62 * fixed bug: ROOM no longer reports silly numbers for stack usage if
63 the stack pointer (treated as a signed integer) is negative.
64 (thanks to Peter van Eynde)
66 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
67 * new platform: experimental support for the Windows operating
68 system has been added. (thanks to Alastair Bridgewater)
69 * fixed several bugs in and robustified the PPC FFI (including
70 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
71 * bug fix: allow non-simple string symbol names (reported by Paul
73 * bug fix: interrupt handling on NetBSD (thanks to Richard M
75 * bug fix: saving a core corrupted callbacks on x86/x86-64
76 * bug fix: closed a loophole in metacircularity detection and
77 grounding in the PCL implementation of CLOS.
78 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
79 (reported by tomppa on #lisp)
80 * bug fix: clear the direction flag on Lisp -> C transitions, as
81 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
82 (reported by Andrej Grozin and Hendrik Maryns)
83 * optimization: major improvements to GC efficiency on GENCGC platforms
84 * optimization: faster implementation of EQUAL
85 * optimization: emit more efficient opcodes for some common
86 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
87 * optimization: several other minor code-generation improvements on x86-64
88 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
90 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
91 * minor incompatible change: (SETF CLASS-NAME) and (SETF
92 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
93 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
94 by ANSI as a generic function, and for consistency (SETF
95 GENERIC-FUNCTION-NAME) is treated likewise.
96 * fixed bug #233.b: make constraint propagation notice when a variable
97 value is changed after it is referenced but before it is used
98 * fixed bug #296: no more arbitrary behaviour from filenames with
99 odd characters as --load arguments. (reported by Adam Warner)
100 * fixed bug #390: :CHARACTER-SET pathname components now work as
101 expected. (reported by Tim Daly Jr)
102 * fixed bug #391: complicated :TYPE intersections in slot
103 definitions no longer cause an error in PCL internals.
104 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
105 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
106 returning the number of octets which would be written to the
107 file-stream. (thanks to Robert J. Macomber)
108 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
109 arguments to the internal stream functions. (thanks to David
111 * bug fix: stack exhaustion is now handled on the x86/SunOS
112 platform. (thanks to Daisuke Homma)
113 * bug fix: more accurate ROOM results on GENCGC platforms
114 * optimization: improved type inference for arithmetic-for
115 index variables in LOOP
116 * optimization: faster floating-point SQRT on x86-64
117 * fixed some bugs revealed by Paul Dietz' test suite:
118 ** DOCUMENTATION returns NIL instead of "" for method combinations
119 that don't have a docstring
121 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
122 * minor incompatible change: (SETF CLASS-NAME) and (SETF
123 GENERIC-FUNCTION-NAME) are no longer generic functions, and
124 therefore are not customizeable by user code (as seems to be at
125 least permitted and maybe required by AMOP). As a consolation,
126 however, the SBCL implementation of these functions now calls
127 REINITIALIZE-INSTANCE as specified by AMOP.
128 * bug fix: slot-definition documentation is propagated as per ANSI
129 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
130 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
131 argument, as do eslotd initialization methods. (from Pascal
132 Costanza's "Closer" project)
133 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
134 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
136 * bug fix: it is now possible to have more than one subclass of
137 STANDARD-GENERIC-FUNCTION without causing stack overflow.
138 (reported by Bruno Haible, Pascal Costanza and others)
139 * bug fix: the dependent update protocol now works for generic
140 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
142 * bug fix: condition-class instances corresponding to
143 DEFINE-CONDITION forms are now created eagerly. (reported by
144 Kalle Olavi Niemitalo on comp.lang.lisp)
145 * bug fix: floating point printing is more accurate in some
146 circumstances. (thanks to Simon Alexander)
147 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
148 merged with *DEFAULT-PATHNAME-DEFAULTS*.
149 * bug fix: callbacks on OS X now preserve stack-alignment.
150 (thanks to Cyrus Harmon)
151 * enhancement: the x86-64 disassembler is much better at
152 disassembling SSE instructions. (thanks to Lutz Euler)
153 * enhancement: CHECK-TYPE on a variable now allows type-inference
154 (thanks to Vincent Arkesteijn)
155 * optimization: improved performance of EUC-JP external format.
156 (thanks to NIIMI Satoshi)
157 * optimization: performance improvements to IO on file streams of
158 :ELEMENT-TYPE CHARACTER
159 * optimization: much faster memory allocation on x86-64
160 * optimization: faster garbage collections (latency and throughput)
162 * optimization: faster bignum subtraction and division on x86 and x86-64
163 (thanks to Lutz Euler)
164 * optimization: more accurate type-inference for loops that use a
165 floating point index variable or a negative step.
167 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
168 * bug fix: add a workaround to SBCL looping infinitely at startup on
169 Linux kernels with apparently buggy implementations of personality().
170 (thanks to Svein Ove Aas)
171 * bug fix: Unicode symbols are correctly printed in LDB backtraces
172 (thanks to David Lichteblau)
173 * bug fix: local bindings shadow global symbol macros for
174 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
175 * optimization: non-open coded uses of numeric comparison operators
176 (e.g. >) no longer cons when called with more than one parameter
177 on platforms supporting dynamic-extent allocation.
178 * enhancement: saving cores with foreign code loaded is now supported
179 on MIPS/Linux in addition to the previously supported platforms.
180 * bug fix: FIND-RESTART now tests for activity, not applicability when given
181 a restart object as identifier. (reported by Helmut Eller for CMUCL)
182 * bug fix: division by zero in sb-sprof when no samples were collected
183 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
184 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
185 (thanks to Alexander Kjeldaas)
186 * new feature: ignoring the return values of destructive functions
187 like NREVERSE signals a compile-time style-warning.
188 (thanks to Kevin Reid)
189 * new feature: the alignment of alien structure fields can be
190 explicitly specified. (thanks to Cyrus Harmon)
191 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
192 and dump core on SIGQUIT
194 ** incompatible change: threads do not inherit values of specials
195 from their parents (see manual)
196 ** bug fix: threads stacks belonging to dead threads are freed by the
197 next exiting thread, no need to gc to collect thread stacks anymore
198 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
199 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
200 ** bug fix: dynamic variable and thread start related gc lossage
201 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
203 * fixed some bugs revealed by Paul Dietz' test suite:
204 ** SUBTYPEP is slightly more accurate on heinously complicated
205 CONS types where some of the members have uncertain (in the
206 NIL, NIL sense) type relationships to each other.
208 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
209 no longer purified unless :PURIFY T is explicitly specified.
210 ** Non-purified cores are significantly smaller than before
212 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
213 * new feature: timers based on Zach Beane's excellent timer package
214 * added support for the following external formats: koi8-u,
215 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
216 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
217 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
218 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
219 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
220 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
221 * incompatible change: a threaded SBCL will no longer revert to
222 non-threaded mode on non-NPTL systems, but refuse to start entirely.
223 * bug fix: interrupts are disabled until startup is complete; no
224 more sigsegvs when receiving a signal to soon
225 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
227 * bug fix: add a workaround for the memory randomization features in
228 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
229 workaround will only be in effect on systems with the proc filesystem
231 * bug fix: printing objects of type HASH-TABLE signals a
232 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
234 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
235 have been running for over 50 days. (reported by Gilbert Baumann)
236 * bug fix: the logic for getting names of functions gets less
237 confused when confronded with alternate-metaclass
238 funcallable-instances. (reported by Cyrus Harmon)
239 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
240 consistent, even on internal alternate-metaclass objects.
241 * bug fix: flush closure information collected by physical
242 environment analysis prepass before the main pass. (bug reported
244 * bug fix: compiler pack phase does not modify a hashtable, which is
245 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
246 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
247 may be expanded. (reported by Alexander <malishev> on c.l.l)
248 * bug fix: no more style-warnings when compiling code using LAST
249 under high-SPEED low-SPACE settings. (reported by David Wallin
251 * fixed some incompatibilities between SBCL's MOP and the MOP
253 ** the METAOBJECT class is now implemented;
254 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
255 STANDARD-OBJECT, as required;
256 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
257 now compatible (as required by VALIDATE-SUPERCLASS); there
258 remains a consistency requirement over the presence or absence
259 of the FUNCTION class in the superclass list of finalized
260 classes; see the manual for more details;
261 ** the :AROUND method for COMPUTE-SLOTS on
262 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
263 requested slot ordering.
265 ** bug fix: parent thread now can be gc'ed even with a live
267 ** bug fix: binding a special with PROGV to no value is not
268 the same as that symbol not having been bound (thanks to
270 ** bug fix: binding specials is thread safe (thanks to
272 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
273 and SB-SPROF do not die with 'no handler for signal XX in
274 interrupt_handle_now(..)' anymore
275 ** bug fix: WITH-TIMEOUT works with multiple running threads
276 * fixed some bugs revealed by Paul Dietz' test suite:
277 ** ENSURE-GENERIC-FUNCTION should take a method class object for
278 the :method-class keyword argument.
280 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
281 * new port: the Solaris operating system on x86 processors is now
282 mostly supported, though some rough edges in the environment
283 remain. (thanks to Daisuke Homma)
284 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
285 space for its heap. (thanks to Thiemo Seufer)
286 * optimization: SBCL on MIPS platforms now supports dynamic-extent
287 closures. (thanks to Thiemo Seufer)
288 * minor incompatible change: eof selects abort in the debugger.
289 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
290 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
292 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
293 deprecated, and will go away in a future revision of SBCL.
294 * minor incompatible change: GC-ON and GC-OFF are no longer
295 implemented with a counter, it does not matter how many times gc
296 is switched on or off
297 * bug fix: discriminating functions for generic function classes
298 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
299 make invalid assumptions about method precedence order. (reported
301 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
302 platforms. (thanks to Thiemo Seufer)
303 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
304 * bug fix: correct pathname printing with printer escaping is on.
305 (thanks to Kevin Reid)
306 * bug fix: complex VOP definitions in "user-space" no longer trigger
307 package locks. (reported by Zach Beane)
308 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
309 for funcallable-instance functions, and is no different from
311 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
312 characters in the current readtable. (reported by Nicholas Neuss)
313 * bug fix: SUBTYPEP on various CONS types returns more a more
314 accurate acknowledgment of its certainty.
315 * optimizations: REMOVE-DUPLICATES now runs in linear time on
316 lists in some cases. This partially fixes bug 384.
317 * flush all standard streams before prompting in the REPL and the
319 * bug fix: signal handling and triggering gc do not conflict
320 directly anymore, in particular a high frequency sb-sprof does
321 not prevent gc from running
322 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
323 approximation for timezone and DST information between the
324 universal time epoch and the smallest negative 32-bit time_t.
325 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
326 year 1899 so long as the encoded time is non-negative.
327 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
328 the correct answer for non-constant rotate amounts when performing
329 an inline 32-bit rotation.
331 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
332 there is only one thread in the session
333 ** bug fix: memory leak for streams created in one thread and
334 written to in another
335 ** bug fix: lockup when compiled with gcc4
336 ** bug fix: race that allows the gc to be triggered when gc is
338 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
339 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
341 ** bug fix: finished threads are gc'ed properly
342 * fixed some bugs revealed by Paul Dietz' test suite:
343 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
344 has arguments with a different set of applicable methods from
345 the orignal arguments.
346 ** The type error thrown by MAP now has a correct expected-type
348 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
349 name a compiled function.
350 ** Three MISC tests where a large bignum was improperly coerced to
351 a float that couldn't represent that bignum during type
352 derivation were fixed.
353 ** SUBTYPEP can now handle types involving the negation of a
354 list-form FUNCTION type.
355 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
356 as (COMPLEX (AND RATIO (NOT FIXNUM))).
357 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
359 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
360 * New feature: Experimental support for bivalent streams: streams
361 opened with :element-type :default now allow character and binary
362 (unsigned-byte 8) I/O
363 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
364 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
366 * bug fix: no more highly sporadic "couldn't check whether ~S is
367 readable" when reading a stream and an interrupt hits in the middle
368 of a select system call
369 * compiler better recognizes complex arrays (reported by Eduardo
371 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
372 for variables on SBCL built with linkage-tables. (reported by Luis
374 * various error reporting improvements.
375 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
376 (Thanks to Thiemo Seufer)
377 * bug fix: nested reader invokations work correctly
378 * bug fix: it is possible to have simultaneous references to foreign
379 code and foreign data with the same name.
381 ** added x86-64 support
382 ** incompatible change: the threading api now works with thread
383 objects instead of thread ids
384 ** bug fix: threads are protected from signals and interruption when
385 starting up or going down
386 ** bug fix: a race where an exiting thread could lose its stack to gc
387 ** bug fix: don't halt on infinite error in threads if possible
388 ** fixed numerous gc deadlocks introduced in the pthread merge
389 ** bug fix: fixed thread safety issues in read and print
390 ** bug fix: debugger doesn't hang on session lock if interrupted at
391 an inappropriate moment
392 ** bug fix: run-program is now thread safe(r)
393 ** bug fix: inner with-recursive-lock no longer releases the mutex
394 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
395 * fixed some bugs revealed by Paul Dietz' test suite:
396 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
397 EXPECTED-TYPE slots filled.
398 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
400 ** MAKE-CONDITION accepts classes as type-designators.
401 ** COMPILE may never return NIL.
402 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
403 range before calling Unix time functions
405 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
406 * numerous signal handling fixes to increase stability
407 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
408 * minor incompatible change: we now correctly canonize default
409 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
410 as per the MOP, rather than the historical (INITARG INITFUNCTION
411 INITFORM). (reported by Bruno Haible)
412 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
413 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
414 for more information.
415 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
416 pathname is a directory pathname.
417 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
418 * dynamic space size on PPC has been increased to 768Mb. (thanks to
420 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
421 :METACLASS argument in addition to a class name. (reported by
422 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
423 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
424 (reported by Bruno Haible; patch from Gerd Moellmann)
425 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
427 * bug fix: more cleanups to the floating point exception handling on
428 x86-64 (thanks to James Knight)
429 * bug fix: the compiler does not try to constant fold calls of
430 COERCE and BIT-* functions when they return freshly constructed
431 objects. (reported by Nikodemus Siivola and Paul Dietz)
432 * optimization: improved the MIPS versions of generic (in the
433 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
434 * optimization: direct conversion of (unsigned-byte 32) to floats on
435 the PowerPC platform.
436 * optimization: structure instances with raw slots now use less
437 memory, and probably show better memory locality. (thanks to
439 * optimization: DYNAMIC-EXTENT declarations for lists and closures
440 are treated as requests for stack allocation on the x86-64,
441 Alpha-32, PPC and SPARC platforms.
442 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
443 the wrong file descriptor; implementation of SOCKET-OPEN-P.
444 (thanks to Tony Martinez)
446 ** gcing a dead thread can no longer lead to lockups
447 ** threads block signals until they are set up properly
448 ** errno is no longer shared by threads
449 ** interrupt-thread restores the eflags register on x86
450 ** fixed some lockups due to gc/thread interaction
451 * fixed some bugs revealed by Paul Dietz' test suite:
452 ** invalid dotted lists no longer raise a read error when
454 ** signal an error if a symbol that names a declaration is used
455 as the name of a type, or vice versa
456 ** allow using the (declare (typespec var*)) abbreviation for
457 (declare (type typespec var*)) with all type specifiers
458 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
459 ** always return NIL from PPRINT-POP when OBJECT is NIL
460 ** don't signal errors when pretty-printing malformed LABELS,
461 FLET or MACROLET forms
462 ** declarations in a DOLIST body are also in scope for the
464 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
466 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
469 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
470 * fixed cross-compiler leakages that prevented building a 32-bit
471 target with a 64-bit host compiler.
472 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
473 opened with :IF-EXISTS :OVERWRITE.
474 * fixed bug 281: error for an invalid qualifier in a short-form method
475 combination method is not signalled until the faulty method is called.
476 * bug fix: iteration variable type inferrer failed to deal with open
477 intervals. (reported by Alan Shields)
478 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
479 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
480 or more runtime options were provided to the sbcl binary.
481 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
482 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
483 * optimization: there's now a fast-path for fixnum arguments in the
484 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
485 * optimization: the code generated on x86-64 is more compact thanks
486 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
487 * bug fix: floating-point exception handling now partly works on
488 x86-64. (Thanks to James Knight)
489 * improvement to the MIPS backend from Thiemo Seufer: C-style
490 64-bit long long arguments and return values to alien functions
492 * fixed some bugs revealed by Paul Dietz' test suite:
493 ** the type-error signalled from WARN has a filled-in DATUM slot.
494 ** the type-error required when a stream is not associated with
495 a file has the stream as its datum.
496 ** type-errors on single-floats on x86-64 no longer have
497 :INVALID-OBJECT as the datum
498 ** the type-errors signalled for invalid function names now have
499 a correct expected type
500 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
501 for structures defined with a :TYPE. Documentation strings for
502 typed structures are no longer immediately discarded
503 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
504 broadcast streams and synonym streams. FILE-LENGTH now also works
505 on broadcast streams.
507 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
508 * incompatible change: the --noprogrammer option, deprecated since
509 version 0.7.5, has been removed. Please use the equivalent
510 --disable-debugger option instead.
511 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
513 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
514 inoperational for a while now) has been completely removed.
515 * null lexical environments are now printed as #<NULL-LEXENV>,
516 significantly reducing the amount of clutter in typical backtraces.
517 * documentation on weak pointers, finalization, and after GC hooks
518 has been added to the manual.
519 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
520 as other specialized array types, is much faster. SUBSEQ and
521 COPY-SEQ on such arrays have also been sped up.
522 * optimization: EQL is now more efficient when at least other argument
523 is known to be of type (OR FIXNUM (NOT NUMBER)).
524 * fixed bug: the runtime is now less vulnerable to changes in the
525 size of the SBCL object on OS X, and virtual memory is reserved for
526 all spaces that need to be at a fixed address.
527 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
528 * fixed bug: finalizers and after GC hooks that cause consing are now
530 * fixed bug: compiler error messages and summaries are now printed to
531 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
532 * fixed inference of the upper bound of an iteration variable.
533 (reported by Rajat Datta).
534 * fixed bug 211e: calling local functions with duplicated constant
535 keyword argument no longer causes a bogus style warning about an
537 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
538 ftype declaration to be disregarded. (reported by Dave Roberts)
539 * fixed bug 373: caused by erronous compilation of references to alien
540 variables in the runtime on ppc/darwin.
541 * fixed bug 376: CONJUGATE type deriver.
542 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
543 (reported by Baughn on #lisp)
544 * a cross-compiler bug on non-x86 platforms has been identified and
545 fixed. (thanks to Bruno Haible)
546 * improvements to the MIPS runtime code for increased stability.
547 (thanks to Thiemo Seufer)
548 * increased the maximimum compact environment size to allow
549 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
550 * partial workaround for bug 135: don't name the function we're
551 calling for hairy cases of EVAL, so as not to accumulate one environment
552 entry per call to EVAL. (reported by Kevin Reid)
553 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
554 * optimization: type testing for non-vector arrays should be faster.
555 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
557 * bug fix: setting 31st element of a bit vector to zero did not work
559 * bug fix: redefining a class definition which failed due to a
560 previous accessor / function clash now works (but see BUGS entry
561 #380 for more problems in this area). (thanks to Zach Beane)
562 * the long form of DEFINE-METHOD-COMBINATION disables method group
563 checking when given a single method group with pattern *. (thanks
565 * on x86 compiler supports stack allocation of results of simple
566 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
567 * support for building and running on Mac OS X 10.4 "Tiger" has been added
568 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
569 * fixed some bugs related to Unicode integration:
570 ** the restarts for recovering from input and output encoding
571 errors only appear when there is in fact such an error to
573 * fixed some bugs revealed by Paul Dietz' test suite:
574 ** MISC.549 and similar: late transformation of unsafe type
575 assertions into derived types caused unexpected code
577 ** SCALE-FLOAT type deriver is less wrong.
578 ** type derivers for EXP, LOG and similar functions compute result
579 types for complex arguments better.
580 ** (MISC.563) CONJUGATE type deriver works for very restricted
582 ** out-of-line type testers for character strings are available.
583 ** EQUAL compiler transform understands specialness of objects
585 ** accessing double-floats stored on the stack now works on x86-64.
586 ** debugger internals could sometimes create invalid lispobjs,
587 resulting in GC crashes.
588 ** MISC.548: type check weakening can convert required type into
590 ** initialization forms for bindings are not in scope of free special
593 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
594 * incompatible change: thread support for non-NPTL systems has
595 been removed - locking is buggy and unreliable. A threaded
596 SBCL build will now warn at startup and refuse to create
597 new threads, unless futex support is detected in the kernel
598 * incompatible change: the top level REPL now has only an ABORT
599 restart associated with it, not TOPLEVEL and ABORT as it used to.
600 TOP and TOPLEVEL are now available as debugger commands for
601 returning to the top level.
602 * incompatible change: forms evaluated in the REPL now use the
603 global optimization policy.
604 * incompatible change: user- and system-initialization files are
605 no longer processed with LOAD, but by READ and EVAL; hence the
606 global optimization policy, startup package, readtable, etc,
608 * The .fasl file format number has been incremented because of
609 various incompatible changes.
610 * internal entry point details and argument counts no longer appear
611 in backtraces unless explicitly requested by setting
612 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
613 * built-in and standard functions no longer have names like "top
614 level local call to FOO".
615 * fixed bug 32: functions defined in non-null lexical environments
616 now have more legible printed representation
617 * fixed bug 33: functions defined in non-null lexical environemnts
618 are now more amenable to inspection by INSPECT.
619 * workaround for bug 354: XEPs no longer appear in backtraces unless
620 explicitly requested.
621 * fixed bug: receiving the signal which results from attempting to
622 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
623 SIGBUS on other *BSDs) on architectures where the C stack is also
624 the Lisp stack (x86 and x86-64) from foreign code no longer leads
625 to debugger confusion or wild instability. (reported by Cheuksan
627 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
628 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
629 (reported by Lutz Euler)
630 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
631 compiled code. (reported by Paul Dietz)
632 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
633 specialized parameter to a method to something that is not TYPEP
634 the specializer is now possible.
635 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
636 face of package deletion.
637 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
638 pathname. (reported by David Tolpin on c.l.l)
639 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
640 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
641 STRING-STREAM and FILE-STREAM, respectively.
642 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
643 than previously; in addition, its test suite is now run on build.
644 (thanks to Luke Gorrie)
645 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
646 * optimization: added a immediate representation for single-floats
648 * fixed some bugs revealed by Paul Dietz' test suite:
649 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
650 ** attempting to create a package with a colliding nickname causes
651 correctable errors to be signalled.
652 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
653 ** several x86-64 backend bugs related to sign-extension of immediate
656 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
657 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
658 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
660 * fixed loading of multiply forward-referenced layouts.
661 (thanks to Cheuksan Wang)
662 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
663 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
664 (see COMPILE-FILE documentation for details of :PRINT option)
665 * fixed bugs 19 and 317: fixed-format floating point printing is
666 more accurate. This also fixes a bug reported by Adam Warner
667 related to the ~@F format directive.
668 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
670 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
671 dispatch table if the from-char is a dispatch macro character.
672 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
673 by the wordsize no longer ignore the last word. (reported by Lutz
675 * fixed bug in type checking of dynamic-extent variables. (reported
677 * optimization: sequence traversal functions use their freedom to
678 coerce function designators to functions.
679 * optimization: code with many calls to CLOS methods specialized on
680 CLOS classes has had redundant type checks removed.
681 * optimization: added declarations to speed up operations that access
682 the internal character database (for example STRING-UPCASE)
683 * optimization: comparison operations between floats and sufficiently small
684 fixnums no longer create extra rationals
685 * fixed some bugs related to Unicode integration:
686 ** portions of multibyte characters at the end of buffers for
687 character-based file input are correctly transferred to the
688 start of the buffer at the next read.
689 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
690 passing it through to OPEN.
691 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
692 argument given to internal calls to OPEN.
693 ** the built-in comment readers (introduced by character sequences
694 ";" and "#|") are more forgiving to encoding errors; they will
695 STYLE-WARN and then attempt to resync the stream at a character
696 boundary. (thanks to Teemu Kalvas)
697 * fixed some bugs in the x86-64 port:
698 ** Negative short int return values from c-calls are sign-extended
700 ** The stack is aligned to 16-bytes for c-calls, as required by
701 the ABI. (thanks to Cheuksan Wang)
702 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
703 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
704 for enums. (thanks to Vincent Arkesteijn)
705 ** Multiple small optimizations and bugfixes for floating point
707 * fixed some bugs revealed by Paul Dietz' test suite:
708 ** Space, Tab, Linefeed, Return and Page have the invalid
709 secondary constituent character trait.
710 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
712 ** WITH-INPUT-FROM-STRING should only update the index place on
714 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
716 ** Bit-array operations (BIT-AND and similar) worked incorrectly
717 with one-dimensional arrays with fill pointers.
718 ** TYPE-OF failed on a complex with an integer realpart and a
720 ** compiler failure during type inference for the code of form
721 (IF (EQL X (THE ...)) ...) (MISC.535).
723 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
724 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
725 platform. The port supports 61-bit fixnums, large memory spaces
726 and reloading shared object files.
727 * enhancement: saving cores with foreign code loaded is now
728 supported on ppc/Darwin in addition to the previously supported
730 * enhancement: the statistical profiler now walks deeper into the
731 call stack for more meaningful call-graphs and accrued time
732 reports (x86/x86-64 only). It also now reports time spent in
734 * enhancement: it is now possible to trace most individual methods
735 of a generic function in addition to tracing the generic function
737 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
739 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
740 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
741 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
742 produces an error. (thanks to Vincent Arkesteijn)
743 * bug fix: NAMESTRING on pathnames with :WILD components in their
744 directories works correctly. (thanks to Artem V. Andreev)
745 * fixed bug 125: compiler preserves identity of closures. (reported
747 * bug fixed: functions with &REST arguments sometimes failed with
748 "Undefined function" when compiled with (DEBUG 3). (reported by
750 * bug fix: overflow during compiling of setting element of a bit
751 vector with constant index and value. (reported by Timmy Douglas)
752 * build fix: fixed the dependence on *LOAD-PATHNAME* and
753 *COMPILE-FILE-PATHNAME* being absolute pathnames.
754 * on x86 compiler partially supports stack allocation of dynamic-extent
756 * GO and RETURN-FROM do not check the extent of their exit points
757 when compiled with SAFETY 0.
758 * fixed some bugs related to Unicode integration:
759 ** encoding and decoding errors are now much more robustly
760 handled; it should now be possible to recover even from invalid
761 input or output to the terminal. (thanks to Teemu Kalvas)
762 ** provided a first cut at implementing STRING-TO-OCTETS and
763 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
764 ** altered the SB-MD5 contributed module slightly, changing the
765 interface just enough for it to be supportable for builds where
766 lisp characters are not eight bits.
767 * fixed some bugs revealed by Paul Dietz' test suite:
768 ** the FORMATTER-generated functions for ~V[ conditionals require
769 the correct number of arguments.
770 ** READ-FROM-STRING returns the mandated second value when applied
771 to displaced strings.
772 ** the #\Rubout and #\Backspace characters are treated as invalid
773 constituent characters by the tokenizer.
775 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
776 * new feature: reloading changed shared object files with
777 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
778 * new feature: references to foreign variables and functions
779 can now be compiled and loaded before the corresponding shared
780 object file is loaded, as long as the foreign definitions are
781 available at runtime.
782 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
783 supported. (thanks to Dan Debertin)
784 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
785 just keywords. (thanks to Vincent Arkesteijn)
786 * fixed bug #331: structure-class instances corresponding to
787 DEFSTRUCT forms are now created eagerly.
788 * fixed bug #345: backtraces from calls to undefined functions work
789 on x86 as well. Related bug #61 is now also partially fixed on x86
790 (backtraces from throws to unknown catch tags.)
791 * bug fix: lambda-list parsing is now stricter vrt. order and number
792 of lambda-list keywords.
793 * bug fix: as specified by AMOP, an error is signalled if a
794 class-option appears multiple times in a DEFCLASS form. (reported
796 * bug fix: RANDOM can be compiled when the compiler derives the type
797 of its numeric argument as a disjoint set of small integers.
798 (reported by Paul Dietz)
799 * bug fix: starting a core saved with shared objects loaded when
800 those objects are not available no longer causes threaded SBCL to
801 hang. (reported by Sean Ross)
802 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
804 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
805 argument, which must always be NIL. (reported by Kalle Niemitalo)
806 * bug fix: printing 1.0d+23 no longer results in an error.
807 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
808 * bug fix: structure slot setters preserve evaluation order. (thanks
810 * fixed some bugs related to Unicode integration:
811 ** RUN-PROGRAM can allow its child to take input from a Lisp
812 stream. (reported by Stefan Scholl)
813 ** ASDF-INSTALL successfully downloads .tar.gz files in all
814 locales. (reported by Ken Causey)
815 * fixed some bugs revealed by Paul Dietz' test suite:
816 ** INCF, DECF and REMF evaluate their place form as specified in
818 ** FORMATTER expands ~{ ~} iteration directives with V or #
819 parameters correctly.
820 ** FORMATTER deals with the ~@[ ~] conditional directive where the
821 consequent uses no arguments correctly.
822 ** the system has a partial understanding of the (COMPLEX RATIO)
824 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
827 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
828 * new feature: a build-time option (controlled by the :SB-UNICODE
829 keyword feature, enabled by default) for building the system with
830 support for the entire 21-bit character space defined by the
832 * new feature: the system now has rudimentary external-format
833 support; the primary user-visible change at this time is that
834 characters with the high bit set (such as lower-case-e-acute) will
835 print correctly to a terminal in a UTF-8 environment.
836 * minor incompatible change: BASE-CHAR no longer names a class;
837 however, CHARACTER continues to do so, as required by ANSI.
838 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
839 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
840 has been moved to the SB-EXT package (temporarily re-exported from
842 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
843 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
845 * fixed bug #351: better error-handlind and reporting for malformed
847 * fixed bug #350: bignum-printing is now more memory-efficient,
848 allowing printing of very large bignums, eg. (expt 2 10000000).
849 (reported by Bruno Haible)
850 * fixed bug #302: better primitive-type selection for intersection
852 * fixed bug #308: non-graphic characters now all have names, as
853 required. (reported by Bruno Haible)
854 * bug fix: redefining a class with different superclasses now correctly
855 removes it from the direct-subclasses of its previous superclasses.
856 (reported by David Morse)
857 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
858 new value now works. (reported by Bruno Haible)
859 * bug fix: correct canonicalization of multiple non-standard slot
860 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
861 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
862 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
863 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
864 now exists, an signals an error.
865 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
866 during macroexpansion for non-iteration variables as well. (reported
867 by Bruno Haible for CMUCL)
868 * bug fix: Cyclic structures and unprintable objects in compiler
869 messages no longer cause errors. (reported by Bruno Haible)
870 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
871 and READ-FROM-STRING all now return a primary value of NIL if
872 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
873 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
874 (reported by Bruno Haible for CMUCL)
875 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
876 do not designate a condition. (reported by Bruno Haible for
878 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
879 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
880 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
881 permitting resolution in favour of any of the conflicting symbols.
882 (reported by Bruno Haible for CMUCL)
883 * bug fix: EQUAL compiler optimizations is less aggressive on
884 strings which can potentially compare true despite having distinct
885 specialized array element types.
886 * bug fix: unit enumerations can be defined without dividing by
887 zero. (thanks to Vincent Arkesteijn)
888 * FORMAT compile-time argument count checking has been enhanced.
889 (report from Bruno Haible for CMUCL)
890 * a partial workaround for the bug 262: the compiler does not try to
891 inline-expand a local function doing RETURN-FROM from a deleted
892 BLOCK. (thanks to Peter Denno for the bug report and to David
893 Wragg for the simple test case)
894 * fixed some bugs revealed by Paul Dietz' test suite:
895 ** RENAME-PACKAGE allows all package designators as new package
897 ** constraint propagation and lambda variable substitution are
898 more cautious in dealing with partially deleted code.
899 ** compiler sometimes lost reoptimization passes.
900 ** CERROR, given a condition as condition designator, passes its
901 remaining arguments to the continue format control without
903 ** Case-altering FORMAT directives work correctly on non-ASCII
905 ** The REAL type specifier handles bounds outside the floating
906 point ranges without signalling FLOATING-POINT-OVERFLOW.
907 ** Functions with IR1-transformations can create intercomponent
908 references to global functions.
909 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
911 ** FORMAT ~R treats a nil value for its first parameter correctly.
913 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
914 * enhancement: saving cores with foreign code loaded is now
915 supported on x86/NetBSD and sparc/Linux in addition to the previously
917 * bug fix: on some platforms repeated installations caused multiple
918 copies of HTML documentation to be installed -- should not happen
919 any more. (reported by Stefan Scholl)
920 * bug fix: parsing self-recursive alien record types multiple times
921 no longer causes infinite recursion. (reported by Thomas F. Burdick,
922 original patch by Helmut Eller for CMUCL)
923 * bug fix: stack-exhaustion detection works now on NetBSD as well.
924 (thanks to Richard Kreuter)
925 * bug fix: defining classes whose accessors are methods on existing
926 generic functions in other (locked) packages no longer signals
927 bogus package lock violations. (reported by François-René Rideau)
928 * bug fix: special variables as DEFMETHOD parameters no longer have
929 associated bogus type declarations. (reported by David Wragg and
931 * bug fix: read-write consistency on streams of element-type
932 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
933 * bug fix: redefiniton of the only method of a generic function with
934 no DEFGENERIC no longer emits a full WARNING. In addition,
935 redefinition of generic functions with no DEFGENERIC to an
936 incompatible lambda list now signals an error. (thanks to Zach
938 * bug fix: DEFGENERIC now works even when there's a function of the
939 same name in an enclosing lexical environment. (thanks to Zach
941 * fixed compiler failure, caused by instrumenting code during
942 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
943 * optimization: added loop analysis and improved register allocation
944 * fixed some bugs revealed by Paul Dietz' test suite:
945 ** POSITION on displaced vectors with non-zero displacement
946 returns the right answer.
947 ** (SIMPLE-STRING) is a valid type specifier for sequence
949 ** *PRINT-LEVEL* handling for slotless structures is pedantically
951 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
952 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
954 ** The justification version of the FORMAT ~< directive treats
955 non-zero minpad parameter correctly.
957 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
958 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
959 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
960 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
961 the supported interface.
962 * new feature: Single-stepping of code compiled with DEBUG 2 or
963 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
964 * new feature: saving cores with foreign code loaded is now
965 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
966 Timothy Moore's work for CMUCL)
967 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
968 parameters to * instead of NIL if no initform is supplied.
969 (reported by Johan Bockgaard)
970 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
971 correctly when FUNCALL appears as the car of the form. Note:
972 despite this FUNCALL forms are not currently subject to
973 compiler-macro expansion. (port of Raymond Toy's fix for the
974 same from CMUCL, reported by Johan Bockgaard)
975 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
976 (thanks for Teemu Kalvas)
977 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
978 case instead of terminating the iteration (thanks for Julian
979 Squires, Sean Champ and Raymond Toy)
980 * bug fix: incorrect expansion of defgeneric that caused a style
981 warning. (thanks for Zach Beane)
982 * bug fix: slot accessor effective method computation works properly
983 for classes with multiple non-standard applicable methods on
984 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
985 * on x86 compiler supports stack allocation of results of LIST and
986 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
987 CMUCL implementation by Gerd Moellmann)
988 * fixed some bugs revealed by Paul Dietz' test suite:
989 ** FORMAT strings with both the ~<~:;~> form of the justification
990 directive and pretty-printing directives cause an error.
992 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
993 * incompatible change: the internal functions
994 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
995 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
996 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
997 instead of the old functions.
998 * new feature: on platforms where "dladdr" is available foreign
999 function names now appear in backtraces. (based on Helmut Eller's
1001 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1002 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1004 * documentation: ASDF manual is now distributed with SBCL.
1005 * bug fix: SBCL can now load its contributed modules with REQUIRE
1006 even if the system-provided entries have been removed from
1007 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1009 * bug fix: backtraces involving undefined functions or assembly
1010 routines are more informative. (thanks to Brian Downing)
1011 * bug fix: mutually referent alien structures now work correctly.
1012 (reported by Rick Taube)
1013 * bug fix: structures defined by WITH-ALIEN can be referred to
1014 within other definitions in the same WITH-ALIEN.
1015 * bug fix: division operators (MOD, TRUNCATE and the like) with
1016 constant zero divisors and integer dividends no longer generate
1018 * bug fix: provide default methods for INPUT-STREAM-P and
1019 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1020 * bug fix: improve the detection and resolution of MOP metacycles
1021 (where effective method computation depends on the generic function
1022 having its effective method computed). (reported by Bruno Haible)
1023 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1024 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1025 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1026 * bug fix: pathnames with relative directory components can be
1027 represented relative to default pathnames.
1028 * optimization: in taking the GCD of bignums, reduce the two bignums
1029 to approximately the same size (using Euclid's algorithm) before
1030 applying the more sophisticated binary GCD. (thanks to Juho
1032 * optimization: COUNT on bitvectors now operates word-at-a-time.
1033 * optimization: ASH with a positive, but not necessarily constant,
1034 (leftwards) shift, when in a modular context, is compiled to a
1036 * fixed some bugs revealed by Paul Dietz' test suite:
1037 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1038 if the corresponding argument is NIL.
1039 ** FORMAT directives accepting a minpad parameter treat negative
1041 ** literal commas as character parameters to FORMAT directives are
1043 ** literal spaces directly after ~<Newline> directives within a
1044 format-logical-block (~:< ~@:>) do not induce :FILL-style
1045 conditional newlines.
1046 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1047 as specified: it no longer includes conditional newlines).
1048 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1050 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1051 :PER-LINE-PREFIX argument does not evaluate to a string.
1053 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1054 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1055 the manual for details; for now, package locks can be disabled by
1056 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1057 if no major problems are found then it is likely that they will be
1058 compiled in unconditionally.
1059 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1060 unsupported operators on all platforms. To load a shared library
1061 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1062 object file, link it into a shared library outside of SBCL and
1063 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1065 * bug workaround: ROOM T and the GC were not working together
1066 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1067 reporting facility aren't preserved. That reporting has been
1068 disabled, so now until and unless someone figures out how to make it
1069 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1070 * minor incompatible change: as threatened around sbcl-0.8.0, the
1071 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1072 The symbols are also exported from SB-PCL for backwards
1073 compatibility, but more so than before SB-PCL should be treated as
1074 an implementation-internal package.
1075 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1077 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1078 It is known as of this release not to work on the Alpha, however.
1079 * fixed bug #167: errors signalled due to illegal syntax in method
1080 bodies are now more legible.
1081 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1082 designators and can hence be used with TYPEP.
1083 * fixed bug #333: CHECK-TYPE now ensures that the type error
1084 signalled, if any, has the right object to be accessed by
1085 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1086 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1087 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1089 * fixed bug #334: programmatic addition of slots using specialized
1090 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1091 and :ALLOCATION :CLASS slots.
1092 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1093 float exponents. (rereported by Peter Seibel)
1094 * fixed a bug: #\Space (and other whitespace characters) are no
1095 longer considered to be macro characters in standard syntax by
1096 GET-MACRO-CHARACTER.
1097 * fixed bug: initialization of condition class metaobjects no longer
1098 causes an instance of the condition to be created. (reported by
1100 * fixed bug: it is now possible to have slots such that
1101 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1102 description is neither :INSTANCE nor :CLASS.
1103 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1104 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1105 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1106 non-standard SLOT-DEFINITION-ALLOCATION values.
1107 * fixed bugs: various race conditions handling exiting threads.
1108 CL-PPCRE's thread torture test now passes.
1109 * fixed arguably-a-bug: GC time proportional to number of threads in
1110 system even when most of them are idle
1111 * optimization: improved performance of BIT and SBIT on bit-vectors.
1112 * .fasl file incompatibility: The fasl file version number has
1113 been incremented because of changes associated with package locks.
1115 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1116 * minor incompatible change: the system no longer provides
1117 optimization hints (conditions which are TYPEP
1118 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1119 mode; these hints are still emitted when higher SPEED optimization
1121 * new contrib module: a sampling profiler (profiling by statistical
1122 sampling, rather than by instrumenting functions) is available as
1123 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1124 * the behaviour of the standard function ED is now customizeable by
1125 third parties through a hook variable: see ED's documentation
1126 string for information on the protocol.
1127 * the compiler no longer emits efficiency notes for (FUNCALL X)
1128 when the type of X is uncertain under default optimization
1130 * fixed bug 276: mutating a binding of a specialized parameter to a
1131 method to something that is not TYPEP the specializer is now
1133 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1134 exhibit the required behaviour on the x86 platform. (thanks to
1135 Peter van Eynde, Eric Marsden and Bruno Haible)
1136 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1137 even for difficult arguments. (reported by Peter Graves)
1138 * fixed bug 141a: the backquote printer now descends quoted
1140 * fixed another bug in backquote printing: no more destructive
1141 modification of the form's list structure. (reported by Brian
1143 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1144 move between its address being taken and the call to
1145 interrupt_thread, fixing a crashing race condition.
1146 * the SB-POSIX contrib implementation has been adjusted so that it
1147 no longer exhibits ridiculously poor performance when constructing
1148 instances corresponding to C structs.
1150 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1151 * minor incompatible change: the sb-grovel contrib now treats C
1152 structures as alien (in the sense of SB-ALIEN) objects rather than
1153 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1154 has implications for memory management of client code
1155 (sb-grovel-returned objects must now be manually managed) and for
1156 type safety (alien objects now have full types).
1157 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1158 used to control emission of compiler diagnostics, rather than the
1159 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1160 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1161 quality should be considered deprecated.
1162 * install.sh now installs the user manual as well
1163 * (not quite a new documentable feature, but worth considering in
1164 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1165 beginnings of a semantically meaningful condition hierarchy is
1166 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1168 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1169 designator as the defaults argument.
1170 * fixed bug: Displaced arrays whose displaced-to array has become
1171 too small now cause ARRAY-DIMENSION to signal an error, providing
1172 for safer bounds-checking. (reported by Bruno Haible)
1173 * fixed bug: DEFCLASS slot definitions with identical :READER and
1174 :WRITER names now signal a reasonable error. (reported by Thomas
1176 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1178 * fixed bug: Invalid slot specification errors now print correctly.
1179 (thanks to Zach Beane)
1180 * fixed bug 320: Shared to local slot value transfers in class
1181 redefinitions now happen corrently from superclasses as
1182 well. (reported by Bruno Haible)
1183 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1185 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1186 improper lists correctly. (reported by Bruno Haible)
1187 * fixed bug 313: source-transform for <fun-name> was erroneously
1188 applied to a call of a value of a variable with name <fun-name>.
1189 (reported by Antonio Menezes Leitao)
1190 * fixed bug 307: The obsolete instance protocol ensures that
1191 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1192 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1193 remove dead unknown-values globs from the middle of the stack.
1194 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1195 easier to resume long computations after using *BREAK-ON-SIGNALS*
1196 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1197 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1198 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1199 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1200 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1201 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1203 * fixed a (fairly theoretical) bug in string printing: if
1204 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1205 does not have array-element-type equal to the most general string
1207 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1208 function name. (thanks to Nikodemus Siivola)
1209 * fixed bug: compile-time format string checker failed on
1210 non-closed ~{. (reported by Thomas F Burdick)
1211 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1212 don't warn on legal loop code involving more than one aggregate
1213 boolean. (thanks to Nikodemus Siivola)
1214 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1215 in keyword position to MAKE-PACKAGE was not being reported as
1217 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1218 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1220 * fixed bugs 280 and 312: the checking for multiple definitions in a
1221 file is less likely to become confused by uses of inline
1223 * fixed bug: the #S reader macro performs the keyword coercion
1224 specified for slot names. (reported by Kalle Niemitalo)
1225 * fixed bug: lambda lists may contain symbols whose names start with
1226 & but are not lambda-list-keywords; their occurrence triggers a
1228 * fixed bug 321: define-method-combination argument lambda lists do
1229 not cause invalid code to be generated when &optional or &aux
1230 variables have default values. (reported by Bruno Haible)
1231 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1232 have CLOS classes; this appears also to have fixed reported
1233 instability in stack exhaustion detection.
1234 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1235 MUFFLE-WARNING when no associated restart is present is now
1237 * optimization: rearranged the expansion of various defining macros
1238 so that each expands into only one top-level form in a
1239 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1241 * optimization: used a previously little-used slot in symbols to
1242 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1244 * fixed some bugs revealed by Paul Dietz' test suite:
1245 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1247 ** Fixed an optimization bug related to inheritance of initforms
1248 from local to shared slots.
1249 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1250 ** CAST optimizer forgot to flush argument derived type.
1251 ** print/read consistency on floats is now orders of magnitude
1252 more likely. (thanks also to Bruno Haible for a similar report
1254 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1255 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1256 for objects of type REAL. Make it so.
1257 ** SXHASH is specified (infelicitously) to respect similarity,
1258 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1259 it so. (thanks to Markus Ziegler)
1260 ** on the Alpha, the compiler succeeds in compiling functions
1261 returning a known number of arguments greater than 63.
1262 ** fixed handling of invalid NIL arguments in keyword position
1264 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1266 ** provided a readably-printable representation for RANDOM-STATE
1268 ** ensured that pathnames, where they have a namestring, always
1269 print using #P"..." syntax.
1271 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1272 * Support for the forthcoming 2.0 version of the NetBSD kernel
1273 running on x86 hardware has been added. (thanks to Perry
1274 E. Metzger most immediately, and others for their past work)
1275 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1276 current release version; SBCL's OpenBSD support had been broken
1277 since about the time of OpenBSD's switch to ELF binary format.)
1278 * [placeholder for DX summary]
1279 ** user code with &REST lists declared dynamic-extent, under high
1280 speed or space and low safety and debug optimization policy.
1281 * The manual has been converted to Texinfo format and the debugger
1282 chapter from the cmucl manual has been added.
1283 * A facility has been added to extract documentation strings from
1284 sbcl and store them as Texinfo-formatted snippets for inclusion in
1285 the manual (via Texinfo's @include directive)
1286 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1287 the test case to Dave Roberts)
1288 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1289 pointers. (reported by Sean Ross)
1290 * bug fix: PROFILE output is printed nicely even for large numerical
1291 values. (thanks to Zach Beane)
1292 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1293 greater than 32 handle EOF correctly.
1294 * bug fix: on X86 an immediate argument of the IMUL instruction is
1295 correctly printed in disassembly. (thanks to Lutz Euler)
1296 * bug fix: class slots in redefined classes preserve their old
1297 values. (thanks to Bruno Haible and Nikodemus Siivola)
1298 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1299 succeeds. (reported by Marco Baringer)
1300 * fixed some bugs revealed by Paul Dietz' test suite:
1301 ** READ-SEQUENCE now works on ECHO-STREAMs.
1302 ** RATIONALIZE works more according to its specification. (thanks
1305 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1306 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1307 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1308 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1309 should matter to you only if you rebind the printer control
1310 variables and then find you want different bindings in the
1311 debugger than in the ordinary execution of your program.)
1312 * The runtime build system has been tweaked to support building
1313 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1314 assembler and linker. (thanks to Nikodemus Siivola)
1315 * Unbound, undefined, undeclared variables now trigger full
1316 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1317 more often programmer error than deliberate exploitation of undefined
1319 * optimization: the hash algorithm for strings has changed to one
1320 that is less vulnerable to spurious collisions. (thanks to Juho
1322 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1323 needless bounds checking. (thanks to Juho Snellman)
1324 * optimization: implemented multiplication as a modular
1325 (UNSIGNED-BYTE 32) operation on the PPC backend.
1326 * fixed some bugs revealed by Paul Dietz' test suite:
1327 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1329 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1330 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1331 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1333 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1335 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1337 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1338 their output stream on EOF from read.
1339 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1340 have been read to end-of-file.
1341 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1343 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1344 description of determination of which consecutive characters
1346 ** Printing the "Space" character with escaping on now yields "#\\ ",
1347 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1348 ** Reading floating-point numbers with *READ-BASE* set to a number
1349 less than 10 works correctly.
1350 ** Reading floating-point numbers with *READ-BASE* set to a number
1351 more than 10 works correctly.
1352 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1353 the readtable currently in effect.
1355 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1356 * minor incompatible change: parsing of namestrings on a physical
1357 (Unix) host has changed; numbers after the final #\. in a
1358 namestring are no longer interpreted as a version field. This is
1359 intented to be largely invisible to the user, except that the
1360 meaning of the namestring "*.*.*" has changed: it now refers to a
1361 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1362 should usually be replaced by
1363 (make-pathname :name :wild :type :wild :version :wild)
1364 with the added benefit that this is more likely to be portable.
1365 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1366 option to OPEN now signals an error if the file being opened
1367 exists; this may have an impact on existing code.
1368 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1369 signals received, on the PowerPC platforms (both Linux and
1370 Darwin). (thanks to Pierre Mai for pointing out the location of
1372 * several fixes on OS X: The system now builds and runs cleanly on
1373 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1374 (thanks to Brian Mastenbrook)
1375 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1376 second-resolution: integer multiples of 1/3600 between -24 and 24.
1377 (thanks to Vincent Arkesteijn)
1378 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1379 type when called with 1 argument; PEEK-CHAR checked type of
1380 PEEK-TYPE only after having read first character from a
1381 stream. (reported by Peter Graves)
1382 * bug fix: the garbage collector now has much better locality
1383 behaviour, and in particular no longer treats all memory as being
1384 exhausted when many small objects point to each other in a deeply
1386 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1387 recognized as being TYPEP their class.
1388 * bug fix: the PUSHNEW documentation string has been corrected.
1389 (thanks to Vincent Arkesteijn)
1390 * bug fix: defaulting of the value for the last of an atypically
1391 large number of multiple values being bound was not being
1392 performed correctly on the Alpha or PPC platforms
1393 * optimization: implemented multiplication as a modular
1394 (UNSIGNED-BYTE 32) operation on the x86 backend.
1395 * optimization: SEARCH on simple-base-strings can now be open-coded.
1396 (see also contrib/compiler-extras.lisp for inspiration for
1397 teaching the compiler about the Boyer-Moore algorithm).
1398 * value, returned by MAX (and MIN) called with several EQUALP, but
1399 not EQL, arguments now does not depend on compiler settings.
1400 * fixed some bugs revealed by Paul Dietz' test suite:
1401 ** in stack analysis liveness information is propagated from
1402 non-local entry points.
1403 ** pathwise CAST removing failed when the CAST node did not start
1405 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1406 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1408 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1409 host is already defined.
1410 ** RENAME-FILE works on streams instead of signalling an internal
1412 ** PEEK-CHAR uses the current readtable when determining whether
1413 or not a character is whitespace.
1414 ** MERGE-PATHNAMES handles the case when the pathname does not
1415 specify a name while the default-pathname specifies a version.
1416 ** Pathnames now stand a better chance of respecting print/read
1418 ** Attempting to use standardized file system operators with a
1419 pathname with invalid :DIRECTORY components signals a
1421 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1423 ** DIRECTORY on logical pathnames is more correct.
1424 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1425 signal a TYPE-ERROR if their argument is not a stream.
1426 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1427 designator argument does not designate a stream.
1428 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1429 examining the synonym.
1430 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1432 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1433 element-type larger than ([UN]SIGNED-BYTE 32).
1435 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1436 * When built with the :SB-FUTEX feature, threaded builds now take
1437 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1438 for faster/more reliable mutex and condition variable support.
1439 * Incompatible change (but one you probably shouldn't have been using
1440 anyway): the interface and code for arbitrating between multiple
1441 threads in the same user session has been redesigned.
1442 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1443 error when called without an explicit environment argument.
1444 (thanks to Nikodemus Siivola)
1445 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1446 confused about their position. (thanks to Adam Warner and Gerd
1448 * bug fix: Pretty printing backquoted forms with unquotations in the
1449 argument list position of various code constructs such as LAMBDA
1450 now works correctly. (reported by Paul Dietz)
1451 * bug fix: Pretty printing unquotations no longer loses all
1452 stream position information.
1453 * optimization: performance of string output streams is now less
1454 poor for multiple small sequence writes.
1455 * optimization: performance of CSUBTYPEP in the presence of complex
1456 expressions involving CONS and NOT many times has been improved.
1457 (reported by Paul Dietz)
1458 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1460 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1461 streams for the REPL, for the debugger, and so on. (thanks to
1463 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1464 * fixed some bugs revealed by Paul Dietz' test suite:
1465 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1466 is ignored if it is too small, rather than propagating through
1467 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1468 ** extremely complex negations of CONS types were not being
1469 sufficiently canonicalized, leading to inconsistencies in
1471 ** VALUES tranformer lost derived type.
1473 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1474 * fixed a bootstrapping bug: the build process no longer assumes
1475 that the various BOOLE-related constants have the same value in
1476 host and target lisps. (noted by Paul Dietz' test suite on an
1477 SBCL binary built from CLISP)
1478 * The system can now be dynamically linked on the MIPS platform,
1479 which enables dynamic loading of foreign code from Lisp. (thanks
1480 to Ralf Baechle for discussions on the MIPS ABI)
1481 * The system now records debugging information for its own source
1482 files in a filesystem-position-independent manner, relative to
1483 the "SYS" logical host.
1484 * fixed a compiler bug: MV-LET convertion did not check references
1485 to the "max args" entry point. (reported by Brian Downing)
1486 * tweaked disassembly notes to be less confident about proclaiming
1487 some instruction as an LRA. (thanks to Brian Downing)
1488 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1489 now each have their own history, command character, and other
1490 characteristics. (thanks to David Lichteblau)
1491 * fixed some bugs revealed by Paul Dietz' test suite:
1492 ** compiler failure in compiling LOGAND expressions including a
1494 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1495 shift greater than 32.
1496 ** FLUSH-DEST did not mark blocks for type check regeneration.
1497 ** HANDLER-CASE failed to accept declarations in handler clauses
1498 in some circumstances.
1500 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1501 * New code in contrib/sb-introspect (still probably not entirely
1502 stable yet) provides some support for smart Lisp development
1503 environments like SLIME.
1504 * The conditions signalled for errors occurring when loading .fasl
1505 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1506 in a way which should help ASDF recover gracefully.
1507 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1508 stuff has been cleaned up. If you code contrib/ stuff, this might
1509 affect you, and you can look at contrib/README, contrib/STANDARDS,
1510 and/or the 0.8.4.27 diff to check.
1511 * In full calls the compiler now does not generate checks for declared
1512 argument types for all arguments.
1513 * various threading fixes
1514 ** and some experimental patches which didn't make it into
1515 the main tree for this release, but which are shipped in
1516 contrib/experimental-thread.patch as a possible fix for some
1517 failures (deadlock, spinning...) in GC-intensive multithreaded
1519 * fixed PPC build problem (source code incompatibility of different
1520 library versions): added offsetof() hackery which attempts to divine
1521 where glibc maintainers put uc_mcontext today
1522 * fixed bug 282: compiler does not trust type assertions while passing
1523 arguments to a full call.
1524 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1525 &OPTIONAL VALUES type parameter.
1526 * fix bug 214: algorithm for noting rejected templates is now more
1527 similar to that of template seletion. (also reported by rydis on
1529 * fixed bug 141b: printing backquoted information readably and prettily
1530 inserts a space where necessary.
1531 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1532 (CHARACTER 10) are now reported as errors, rather than propagated
1533 as unknown types. (reported by piso on #lisp)
1534 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1535 with values NIL and :ERROR. (thanks to Milan Zamazal)
1536 * fixed bug 191c: CLOS now does proper keyword argument checking as
1537 described in CLHS 7.6.5 and 7.6.5.1.
1538 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1539 longer raise an error; further, using a list as a for-as-arithmetic
1540 counter now raises a meaningful error.
1541 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1542 now understood by sequence creation functions such as MAKE-SEQUENCE
1544 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1545 a STRING-INPUT-STREAM.
1546 * compiler enhancement: SIGNUM is now better able to derive the type
1548 * type declarations inside WITH-SLOTS are checked. (reported by
1550 * fixed some bugs revealed by Paul Dietz' test suite:
1551 ** incorrect optimization of TRUNCATE for a positive first
1552 argument and negative second.
1553 ** compiler failure in let-convertion during flushing dead code.
1554 ** compiler failure while deriving type of TRUNCATE on an
1555 interval, containing 0.
1556 ** ASH of a negative bignum by a negative bignum count now returns
1558 ** intersection of CONS types now canonicalizes properly, fixing
1559 inconsistencies in SUBTYPEP.
1561 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1562 * incompatible change: The --disable-debugger command line
1563 option now clobbers the debugger at a more fundamental
1564 level, by redefining #'INVOKE-DEBUGGER instead of by
1565 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1566 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1567 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1568 enter the debugger REPL and then suffer recursive errors
1569 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1570 while under the new system, BREAK in --disable-debugger mode
1571 terminates the system just as an unhandled error would.
1572 * fixed compiler performance when processing loops with a step >1;
1573 * bug fix: DOCUMENTATION now retrieves generic function
1574 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1575 support has been systematized, and now supports the methods
1576 specified by ANSI, along with a default method and a method for
1577 slot documentation. (reported by Nathan Froyd)
1578 * bug fix: effective methods associated with a generic function are
1579 no longer cached over a change of that generic function's method
1580 combination. (reported by Andreas Fuchs)
1581 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1582 names a non-existent directory. (thanks to Andreas Fuchs)
1583 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1584 the PPC platform, lead to stack corruption; this has been fixed.
1585 (reported by Rainer Joswig)
1586 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1587 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1588 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1589 arithmetic in general, is now much more reliable on the PPC
1591 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1592 platform now returns the right answer.
1593 * optimization: restored some effective method precomputation in
1594 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1595 precomputation is now tunable.
1596 * optimization: compiler-internal data structure use has been
1597 reviewed, and changes have been made that should improve the
1598 performance of the compiler by about 20%.
1599 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1600 simple-base-strings and simple-bit-vectors is improved.
1601 * optimization: the optimization of 32-bit logical and arithmetic
1602 functions introduced in version 0.8.3 on the x86 has been
1603 implemented on the mips, ppc and sparc platforms; an
1604 implementation of the same facility, but for 64-bit arithmetic,
1605 has been added for the alpha.
1606 * microoptimization: the compiler is better able to make use of the
1607 x86 LEA instruction for multiplication by constants.
1608 * bug fix: in some situations compiler did not report usage of
1609 generic arithmetic in (SPEED 3) policy.
1610 * bug 145b fix: compiler used wrong type specifier while converting
1611 MEMBER-types to numeric.
1612 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1614 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1615 index argument now works properly on the Alpha platform.
1616 * bug fix: floating point exception treatment on the Alpha platform
1618 * bug fix: FILE-POSITION works much better on string input and
1619 output streams. (thanks to Nikodemus Siivola)
1620 * bug fix: many threading/garbage collection symptoms sorted.
1621 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1622 might be pseudo-atomic.
1623 * internal change: Stopping for GC is now done with signals not
1624 ptrace. GC is now done in whichever thread wanted it, instead of
1626 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1628 * bug fix: result form in DO is not contained in the implicit
1630 * incompatible change: ICR structure is changed; the value part of
1631 CONTINUATION is now called LVAR; corresponding functions are
1632 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1633 * added type deriver for ISQRT (thanks to Robert E. Brown).
1634 * bug fix: better support for loading from the command line when an
1635 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1637 * fixed some bugs revealed by Paul Dietz' test suite:
1638 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1639 ** ROUND and FROUND now give the right answer when given very
1640 small float arguments.
1641 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1643 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1644 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1645 ** LCM should return a non-negative integer.
1646 ** PARSE-INTEGER returned the index of a terminator instead of the
1647 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1648 ** PARSE-INTEGER returned an incorrect index being applied to a
1650 ** LCM with two arguments of 0 returns 0 rather than signalling
1652 ** unsigned addition of a 32-bit constant with the high bit set no
1653 longer causes an internal compiler error on the x86.
1654 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1655 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1656 with negative last argument.
1657 ** byte specifiers with zero size and position no longer cause
1658 an error during type derivation.
1659 ** bignum multiplication on the Alpha platform now returns the
1661 * porting: The system now builds on SuSE AMD64, although it still
1662 generates a 32-bit binary.
1663 * .fasl file incompatibility: The fasl file version number has
1664 been incremented (because of the changes to internal compiler
1665 data structures referred to above).
1667 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1668 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1669 more accurately, on the Darwin kernel running on PowerPC hardware.
1670 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1671 * Compiler code deletion notes now signal a condition of type
1672 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1673 an associated MUFFLE-WARNING restart.
1674 * The compiler now performs limited argument count validation of
1675 constant format strings in FORMAT, and where appropriate in ERROR,
1676 CERROR and WARN. (thanks to Gerd Moellmann)
1677 * New ASDF-INSTALL contrib can be used for automatic download and
1678 installation of third-party Lisp code from CCLAN or other sites
1680 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1681 which forces another thread to execute a function supplied by the
1683 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1684 now accept and act upon their :ELEMENT-TYPE keyword argument.
1685 (reported by Martin Atzmueller, Edi Weitz)
1686 * bug fix: FILE-POSITION now accepts position designators up to
1687 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1688 is the greater. (thanks to Patrik Nordebo)
1689 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1690 Kalle Olavi Niemitalo)
1691 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1692 printable. (reported by Eric Marsden)
1693 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1694 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1695 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1696 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1697 (thanks to Stig Sandoe)
1698 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1699 UNREAD-CHAR now correctly includes the unread character in the
1700 target sequence. (thanks to Gerd Moellmann)
1701 * bug fix in threaded builds: the system can now be suspended and
1702 resumed by shell job control with minimal disruption.
1703 * bug fixes in times and timezones >2038AD
1704 * better handling of "where is GNU make?" problem in build scripts
1705 (thanks to Nikodemus Siivola)
1706 * new optimization: inside a named function any reference to a
1707 function with the same name is considered to be a self-reference;
1708 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1709 optimization quality.
1710 * new optimization on x86: logical functions and + now have
1711 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1712 used when the result is truncated to 32 bits.
1713 * VALUES declaration is partially enabled.
1714 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1715 * fixed some bugs revealed by Paul Dietz' test suite:
1716 ** The system now obeys the constraint imposed by
1717 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1718 types form a lattice under type intersection.
1719 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1720 ** ASSOC now ignores NIL elements in an alist.
1721 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1722 and (1+ MOST-POSITIVE-FIXNUM) answers.
1723 ** The addition of a method with invalid qualifiers to a generic
1724 function does not cause an error to be signalled immediately;
1725 a warning is signalled, and the error is generated only on
1726 calling the generic function.
1727 * changed .fasl file version number, in part to add type codes for
1728 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1729 obscure ANSI requirements
1731 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1732 * fixed bug 148: failure to inline-expand a local function left
1733 garbage, confusing the compiler.
1734 * fixed bugs 3cd: structure slot readers perform type check if the
1735 slot can have an invalid value (i.e. it is either not initialized
1736 or can be written with a less specific slot writer).
1737 * bug fix: the compiler now traps array references to elements off
1738 the end of an array; previously, the bounds checking in some
1739 circumstances could go off-by-one.
1740 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1742 * type declarations for array element types now obey the description
1743 on the CLHS page "Declaration TYPE", as per discussions on
1744 sbcl-help around 2003-05-08. This means that a declaration
1745 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1746 declaration, all references to BAR will be asserted or assumed
1747 (with THE, so dependent on compiler policy) to involve objects of
1748 type FOO. Note that no such declaration is implied in
1749 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1750 * declared types of functions from the "Conditions"
1751 chapter. (reported by Paul Dietz)
1752 * bug fix: CERROR accepts a function as its first argument.
1753 * bug fix: NTH an NTHCDR accept a bignum as index
1754 arguments. (reported by Adam Warner)
1755 * optimization: character compare routines now optimize comparing
1756 against a constant character. (reported by Gilbert Baumann)
1757 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1758 argument now works properly on the MIPS platform.
1759 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1760 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1761 declared to have a functional type, does not cause an error
1763 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1764 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1765 * fixed bug: if last continuation of a deleted block has a
1766 destination, this destination should be deleted too. (reported by
1768 * fixed a bug in the bootstrap process: the host compiler's values
1769 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1770 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1771 test case from Patrik Nordebo)
1772 * improved the ability of the disassembler on the PPC platform to
1773 provide helpful disassembly notes.
1774 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1775 the class in more cases than previously.
1776 * bug fix: FILE-POSITION now understands :START and :END for
1777 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1778 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1779 8) data. (thanks to David Lichteblau)
1780 * bug fix: it is possible to add a method to a generic function
1781 without lambda list.
1782 * bug fix: reader failed to signal END-OF-FILE inside an
1783 object representation. (reported by Nikodemus Siivola)
1784 * fixed some bugs revealed by Paul Dietz' test suite:
1785 ** LAST and [N]BUTLAST should accept a bignum.
1786 ** condition slot accessors are methods.
1787 ** (VECTOR NIL) is a subtype of STRING.
1789 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1790 * minor incompatible change: some nonsensical specialized lambda
1791 lists (used in DEFMETHOD) which were previously ignored now signal
1793 * minor incompatible change: the system is now aware of the types of
1794 variables in the COMMON-LISP package, and will signal errors for
1795 most violations of these type constraints (where previously they
1796 were silently accepted).
1797 * minor incompatible change: COMPILE-FILE now uses the freedom
1798 afforded (ANSI 3.2.2.3) to use derived function types for
1799 functions defined in the same file. This also permits the system
1800 to warn on static type mismatches and function
1801 redefinition. (Currently it does not work with high DEBUG level.)
1802 * minor incompatible change: VALUES declaration is disabled.
1803 * When issuing notes, the compiler now signals a condition of type
1804 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1805 restart for use in user handlers. It is expected that the
1806 COMPILER-NOTE condition will eventually become a condition
1807 supertype to a hierarchy of note types, which will then be
1808 handleable in a similar fashion. However, at the moment, no such
1809 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1810 but it's an implementation detail, not a classification for the
1812 * Changes in type checking closed the following bugs:
1813 ** type checking of unused values (192b, 194d, 203);
1814 ** template selection based on unsafe type assertions (192c, 236);
1815 ** type checking in branches (194bc).
1816 * A short form of VALUES type specifier has ANSI meaning (it has
1817 increased the number of situations when SBCL cannot perform type
1819 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1820 keywords or constants is permissible.
1821 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1822 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1823 classes. (thanks to Antonio Martinez)
1824 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1825 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1826 argument to a pretty-print stream. Instead, it leaves any such
1827 operations to DESCRIBE-OBJECT methods.
1828 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1830 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1831 lambda lists. (thanks to David Lichteblau)
1832 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1833 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1834 (thanks to Gerd Moellmann)
1835 * fixed bug 246: increased compilation speed of long
1836 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1838 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1840 * DEFCONSTANT now throws a condition of type
1841 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1842 constant to a non-EQL value; CONTINUE and ABORT restarts
1843 respectively change and preserve the value.
1844 * fixed bug 63: The code walker, part of the implementation of CLOS,
1845 is now better at handling symbol macros.
1846 * bug fix: There is no longer an internal implementation type named
1847 CL:LENGTH. (reported by Raymond Toy)
1848 * bug fix: In macro-like defining macros/special operators the
1849 implicit block does not enclose the lambda list.
1850 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1851 enclosed in parentheses) are not suitable as type specifiers, and
1852 their use properly signals an error now.
1853 * bug fix: An argument count mismatch for a type specifier in code
1854 being compiled no longer causes an unhandled error at compile
1855 time, but signals a compile-time warning.
1856 * fixed simple vector readable printing
1857 * bug fix: DESCRIBE takes more care over whether the class
1858 precedence list slot of a class is bound before accessing it.
1859 (reported by Markus Krummenacker)
1860 * bug fix: FORMATTER can successfully compile pretty-printer format
1861 strings which use variants of the ~* directive inside.
1862 * bug fix: SEARCH now applies its TEST predicate to the elements of
1863 the arguments in the correct order. (thanks to Wolfhard Buss)
1864 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1865 in type inference. (thanks to Robert E. Brown)
1866 * bug fix: Reading in symbols with an explicit package name of ""
1867 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1868 * fixed some bugs revealed by Paul Dietz' test suite:
1869 ** NIL is now allowed as a structure slot name.
1870 ** Arbitrary numbers, not just REALs, are allowed in certain
1871 circumstances in LOOP for-as-arithmetic clauses.
1872 ** Multiple class redefinitions before slot access no longer
1873 causes a type error.
1874 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1875 association between the name and a class.
1876 ** Generic functions with non-standard method-combination and over
1877 five methods all of which return constants no longer return NIL
1878 after the first few invocations. (thanks to Gerd Moellmann)
1879 ** CALL-NEXT-METHOD with no arguments now passes the original
1880 values of the arguments, even in the presence of assignment.
1881 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1883 ** DISASSEMBLE works with closures and funcallable instances.
1884 ** ADD-METHOD now returns the generic function, not the new method.
1885 ** FIND-METHOD signals an error if the lengths of the specializers
1886 is incompatible with the generic function, even if the ERRORP
1888 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1889 which its argument is a member.
1890 ** DEFCLASS only redefines the class named by its class-name
1891 argument if that name is the proper name of the class;
1892 otherwise, it creates a new class.
1893 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1894 of the UNBOUND-SLOT condition to the name of the slot.
1895 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1896 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1897 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1898 SLOT-MAKUNBOUND in the specified fashion.
1900 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1901 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1902 cross-compilation host. As a consequence, we can now bootstrap our
1903 way up to SBCL starting with a bare gcc toolchain and human-readable
1904 source code (first the source to CLISP, then the source to SBCL).
1905 * A contributed module containing a partial implementation of the
1906 simple-streams interface has been included. (thanks to Rudi
1908 * A contributed module implementing the RFC1321 Message Digest
1909 Algorithm, known as MD5, has been included.
1910 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1911 no longer has any effect, as the code controlled by this feature
1912 has been deleted. (As far as we know, no-one has ever built using
1913 this feature, and its semantics were confused in any case).
1914 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1915 work on conditions (as required by the ANSI specification),
1916 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1917 expected behaviour on conditions. Users should note, however,
1918 that such behaviour is not required by the ANSI specification,
1919 and so use of this behaviour may render their code unportable.
1920 * fixed some bugs revealed by Paul Dietz' test suite:
1921 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1923 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1924 ** SXHASH on condition objects no longer returns NIL.
1925 ** :ALLOCATION :CLASS slots are better treated; their values are
1926 updated on class redefinition, and initforms inherited from
1927 superclasses are applied.
1928 ** REMOVE-METHOD returns its generic function argument even when
1929 no method was removed.
1930 ** SHARED-INITIALIZE now initializes the values of the requested
1931 slots, including those with :ALLOCATION :CLASS.
1932 ** ALLOCATE-INSTANCE now works on structure classes defined via
1933 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1935 ** SLOT-EXISTS-P now works on conditions, as well as structures
1937 ** MAKE-LOAD-FORM now has the required methods on
1938 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1939 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1940 keyword, and now implements the SLOT-NAMES argument.
1941 ** methods with &OPTIONAL arguments no longer allow too many
1942 arguments to be passed in the call without error.
1943 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1944 option is consistent with the required arguments of the generic
1945 function lambda list.
1946 * bug fix: REQUIRE accepts a string designator. (Thanks to
1948 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1949 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1950 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1951 * bug fix: The compiler now checks for duplicated variables in macro
1953 * bug fix: SETQ on globals returns the correct value.
1954 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1955 not a COMPILER-ERROR (followed by some other strange error on
1956 choosing the CONTINUE restart).
1957 * bug fix: make.sh and friends are now more consistent in the way that
1958 they look for GNU "make".
1960 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1961 * experimental native threads support (on x86 Linux >=2.4 only).
1962 This is not compiled in by default: you need to add :SB-THREAD to
1963 the target features. See the "Beyond ANSI" chapter of the manual
1965 * fix for longstanding nonANSIism: The old distinction between
1966 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1967 The return value from CL:FIND-CLASS is now a CLOS class, and
1968 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1969 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1970 * An interface to the MetaObject Protocol, as described in Kiczales,
1971 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1972 MIT Press, 1991, is available from the SB-MOP package.
1973 * incompatible change: the SB-PCL package should now be considered
1974 a private implementation detail, and no longer a semi-private MOP
1976 * minor incompatible change: due to rearrangement for threads, the
1977 control stack and binding stack are now allocated at arbitrary
1978 addresses instead of being hardcoded per-port. Users affected by
1979 this probably have to be doing advanced things with shared
1980 libraries, and will know who they are.
1981 * minor incompatible change: Previously, all --eval forms used were
1982 processed with READ before any of them were processed with EVAL.
1983 Now each --eval form is processed with both READ and EVAL before
1984 the next --eval form is processed. (Thus package operations like
1985 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1986 work as the user might reasonably expect.)
1987 * minor incompatible change: *STANDARD-INPUT* is now only an
1988 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1990 * minor incompatible change: Y-OR-N-P is now character-oriented,
1991 not line oriented. Also, YES-OR-NO-P now works without errors.
1992 (thanks to Antonio Martinez)
1993 * sb-aclrepl module improvements: an integrated inspector, added
1994 repl features, and a bug fix to :trace command.
1995 * Known functions, which cannot be open coded by the backend, are
1996 considered to be able to check types of their arguments. (fixing
1997 a bug report by Nathan J. Froyd)
1998 * fixed a bug in computing method discriminating functions: It is
1999 now possible to define methods specialized on classes which have
2000 forward-referenced superclasses. (thanks to Gerd Moellmann)
2001 * fixed evaluation order in optional entries (reported by Gilbert
2003 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2004 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2005 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2006 required-by-AMOP NAME argument, as well as CLASS and
2007 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2008 * fixed bug 20: DEFMETHOD can define methods using names that are
2009 not the proper names of classes to designate class specializers.
2010 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2011 Unix file descriptors, instead of blowing up. (thanks to Antonio
2013 * Garbage collection refactoring: user-visible change is that a
2014 call to the GC function during WITHOUT-GCING will not do garbage
2015 collection until the end of the WITHOUT-GCING. If you were doing
2016 this you were probably losing anyway.
2017 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2018 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2019 TYPEP the latter but not the former.
2020 * The compiler issues a full WARNING for calls to undefined functions
2021 with names from the CL package.
2022 * MAP-INTO for a vector destination is open coded. (reported by
2023 Brian Downing on c.l.l)
2024 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2025 documentation string.
2026 * fixed some bugs revealed by Paul Dietz' test suite:
2027 ** COPY-ALIST now signals an error if its argument is a dotted
2029 ** Condition slots are now accessed more correctly in the presence
2030 of multiple initargs for a given slot.
2031 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2032 exclude restarts of the same name associated with a different
2034 ** DEFCLASS of forward-referenced classes with another
2035 forward-referenced class in the superclasses list no longer
2037 ** Condition slots are now initialized once each, not multiple
2038 times. (thanks to Gerd Moellmann)
2039 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2041 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2042 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2043 arguments contain duplicated elements.
2044 ** RESTART-CASE understands local macros.
2045 ** RESTART-CASE associates exactly its own restarts with a condition.
2046 ** ENDP in safe mode checks its argument to be of type LIST.
2047 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2048 in question is unbound.
2049 ** Optimized MAKE-INSTANCE functions no longer cause internal
2050 assertion failures in the presence of duplicate initargs.
2051 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2052 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2053 propagated even in the ctor optimized implementation of
2055 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2057 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2058 * a better implementation of SXHASH on (simple) bit vectors,
2059 measured both in execution speed and in distribution of results
2060 over the positive fixnums, has been installed. Likewise, a better
2061 implementation of EQUAL for simple bit vectors is now available.
2062 * fixed CEILING optimization for a divisor of form 2^k.
2063 * fixed bug 240 (emitting extra style warnings "using the lexical
2064 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2065 by Antonio Martinez)
2066 * fixed SXHASH, giving different results for NIL depending on type
2067 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2068 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2069 variable. (found by Rolf Wester)
2070 * SBCL does not ignore type declarations for special
2071 variables. (reported by rif on c.l.l 2003-03-05)
2072 * some bug fixes in contrib/sb-aclrepl/
2073 * fixed some bugs revealed by Paul Dietz' test suite:
2074 ** a bug in the CONS type specifier, whereby the CAR and CDR
2075 types got intertwined, has been fixed;
2076 ** the type system is now able to reason about the interaction
2077 between INTEGER and RATIO types more completely;
2078 ** APPEND, [N]REVERSE and NRECONC check that those their
2079 arguments, which must be proper lists, are really so;
2080 ** An array specialized to be unable to hold elements has been
2081 implemented, as required -- yes, really -- by ANSI;
2082 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2083 on malformed property lists;
2085 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2086 * incompatible packaging change: in line with Unix convention,
2087 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2088 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2089 * REQUIRE and PROVIDE are now optionally capable of doing something
2090 useful. See the documentation string for REQUIRE.
2091 * infrastructure for a managed SBCL contrib system: contributed
2092 modules in this release include:
2093 ** the ASDF system definition facility;
2094 ** an interface to the BSD Sockets API;
2095 ** an ACL-like convenience interface to the repl;
2096 (thanks to Kevin Rosenberg)
2097 ** an implementation of ROTATE-BYTE, with efficient implementation
2099 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2100 gives a closer approximation to the right answer than previously.
2101 (thanks to Raymond Toy)
2102 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2103 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2104 optional environment arguments, as required by ANSI.
2105 * fixed bugs in other functions taking environment objects, allowing
2106 calls with an explicit NIL environment argument to be compiled
2108 * fixed bug 228: primary return values from
2109 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2110 COMPILE or FUNCTION.
2111 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2112 :TYPE will no longer signal errors on innocuous objects.
2113 * fixed bug 231b: SETQ is better at respecting type declarations in
2114 the lexical environment.
2115 * fixed a bug in DEFCLASS: classes named by symbols with no or
2116 unprintable packages can now be defined.
2117 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2118 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2119 * fixed compiler failure related to checking types of functions.
2120 (reported by Robert E. Brown)
2121 * the compiler is now much more consistent in its error-checking
2122 treatment of bounding index arguments to sequence functions: in
2123 (SAFETY 3) code, errors will be signalled in almost all cases if
2124 invalid sequence bounding indices are passed to functions defined
2125 by ANSI to operate on sequences.
2126 * fixed a bug in the build procedure: documentation of SBCL-specific
2127 packages is now preserved and available in the final Lisp image.
2128 * lifted FDEFINITION lookup out of loops in the implementation of
2129 many list operations. (thanks to Robert E. Brown)
2130 * fixed a bug in the reader: the #n# reader macro now works for
2131 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2132 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2133 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2134 E. Brown and Tony Martinez respectively)
2135 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2136 SLOT-BOUNDP: in particular, it is now able to optimize them much
2137 better, and is now not vulnerable to having packages renamed.
2138 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2139 always called when a slot is not present in an instance. (thanks
2141 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2142 longer signal an error on unreading a character following EOF on
2143 the previous constituent stream. (thanks to Tony Martinez)
2144 * fixed some bugs revealed by Paul Dietz' test suite:
2145 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2146 not just nonnegative fixnums;
2147 ** the logical bit-array operators such as BIT-AND now accept an
2148 explicit NIL for their "opt-arg" argument (to indicate a
2149 freshly-consed result bit-array);
2150 ** ELT now signals an error on an invalid sequence index in safe
2152 ** the type system is now cleverer about negations of numeric
2153 types, and consequently understands the BIGNUM and RATIO types
2155 ** the type system is now cleverer about the interaction between
2156 INTEGER and RATIO types: while bugs still remain, many more
2157 cases are accurately computed;
2158 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2159 if it is in the last clause;
2160 ** CONSTANTLY now correctly returns a side-effect-free function in
2162 ** DECLARE is no longer treated as a special-operator; in
2163 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2164 * incremented fasl file version number due to the change in the
2165 DEFSTRUCT-SLOT-DESCRIPTION structure.
2167 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2168 * minor incompatible change: code processed by the "interpreter" or
2169 EVAL now has a compilation optimization policy of (DEBUG 2)
2170 (changed from (DEBUG 1)) to improve debuggability of interactive
2171 development, and to allow the use of the debug RETURN command in
2173 * an experimental implementation of the RETURN command for the
2174 debugger has been included. (thanks to Frederik Kuivinen)
2175 * fixed bug 62: constraints were not propagated into a loop.
2176 * fixed bug in embedded calls of SORT (reported and investigated by
2178 * fixed some bugs revealed by Paul Dietz' test suite:
2179 ** printing and reading of arrays with some dimensions having
2180 length 0 (thanks to Gerd Moellmann);
2181 ** BOA constructor with &AUX argument without a default value does
2182 not cause a type error;
2183 ** CONSTANTP now returns true for all self-evaluating objects.
2185 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2186 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2187 accessors that are related by inheritance, as specified in the
2188 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2189 Valtteri Vuorikoski)
2190 * The compiler is now able to inline functions that were defined in
2191 a complex lexical environment (e.g. inside a MACROLET).
2192 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2194 * Support for the upcoming FreeBSD-5.0 release has been included.
2195 (thanks to Dag-Erling Smorgrav)
2196 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2197 effect when it is not in a toplevel context.
2198 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2199 stand a better chance of being correct. (thanks to Gerd
2201 * fixed bug in COERCE, which now signals an error on coercing a
2202 rational to a bounded real type which excludes the expected
2204 * The compiler is now able to derive types more accurately from the
2205 COERCE and COMPILE functions.
2206 * fixed bug 223: functional binding is considered to be constant
2207 only for symbols in the CL package.
2208 * fixed bug 231: SETQ did not check the type of a variable being set
2209 (reported by Robert E. Brown)
2210 * A new optimization for MAKE-INSTANCE has been included, fixing
2211 various bugs (including relating to :ALLOCATION :CLASS slots and
2212 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2214 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2215 ** As required by ANSI, LOOP now disallows anonymous collection
2216 clauses such as COLLECT I in conjunction with aggregate boolean
2217 clauses such as THEREIS (= I 1);
2218 ** LOOP now signals an error when any variable is reused in the
2219 same loop (including the potentially useful construct analogous
2220 to WITH A = 1 WITH A = (1+ A);
2221 ** IT is only a special loop symbol within the first clause of a
2222 conditional loop clause;
2223 ** LOOP with a typed iteration variable over a hashtable now
2224 signals a type error iff it should.
2225 * fixed some other bugs revealed by Paul Dietz' test suite:
2226 ** FILE-STREAM now names the class previously known as FD-STREAM;
2227 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2228 argument) no longer signals an error;
2229 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2230 of the slot symbol, rather than using the current package
2231 ((:CONC-NAME "") continues to intern the slot's name in the
2233 * incremented fasl file version number, because of the incompatible
2234 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2235 of the new implementation of DEFINE-COMPILER-MACRO.
2237 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2238 * Support for building SBCL for MIPS platforms running in
2239 little-endian mode has now been checked in, and basic
2240 functionality on said platforms verified.
2241 * minor incompatible change: PCL now records the pathname of a file
2242 in which methods and the like are defined, rather than its
2244 * minor incompatible change: TRUENAME now considers the truename of
2245 a file naming a directory to be the pathname with :DIRECTORY
2246 component indicating that directory.
2247 * minor incompatible change: a NAMED clause in the extended form of
2248 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2249 reason for the previous behaviour is unclear.
2250 * more systematization and improvement of CLOS and MOP conformance
2251 in PCL (thanks to Gerd Moellman and Pierre Mai):
2252 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2254 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2255 primary methods with no specializers;
2256 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2258 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2259 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2260 FORWARD-REFERENCED-CLASSes; error reporting on
2261 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2263 ** SXHASH on CLOS instances now uses a slot internal to the
2264 instance to return different numbers on distinct instances,
2265 while preserving the same return value through invocations of
2267 ** DEFMETHOD signals errors when methods with longer incongruent
2268 lambda lists are added to generic functions;
2269 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2270 CLASS, as specified in AMOP;
2271 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2272 on the order returned by the primary method for classes of
2273 class STANDARD-CLASS;
2274 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2275 * fixed some bugs shown by Paul Dietz' test suite:
2276 ** DOLIST puts its body in TAGBODY;
2277 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2279 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2280 value producing form;
2281 ** if more variables are given to PROGV than values, extra
2282 variables are bound and made to have no value;
2283 ** NSUBSTITUTE on list arguments gets the right answer with
2285 ** ELT signals an error of type TYPE-ERROR when the index argument
2286 is not a valid sequence index;
2287 ** LOOP signals (at macroexpansion time) an error of type
2288 PROGRAM-ERROR when duplicate variable names are found;
2289 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2290 ** FUNCALL of special-operators now cause an error of type
2292 ** PSETQ now works as required in the presence of side-effecting
2293 symbol-macro places;
2294 ** NCONC accepts any object as its last argument;
2295 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2297 ** loop-for-as-package does not require a package to be explicitely
2299 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2301 * fixed bug 166: compiler preserves "there is a way to go"
2302 invariant when deleting code.
2303 * fixed bug 172: macro lambda lists with required arguments after
2304 &REST arguments now cause an error to be signalled. (thanks to
2306 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2307 bug. (thanks to Matthew Danish)
2308 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2310 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2311 arguments are lexically rebound. (thanks to Gerd Moellmann and
2313 * fixed bug 194: error messages are now more informative when there
2314 is no primary method applicable in a call to a generic
2315 function. (thanks to Gerd Moellmann)
2316 * fixed bug in command line argument checking (thanks to Julian
2318 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2319 (thanks to Matthew Danish)
2320 * incremented fasl file version number, because of the
2321 SXHASH-related changes in the layout of CLOS data structures
2323 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2324 * minor incompatible change: The runtime (the Unix executable named
2325 "sbcl") is now much pickier about the .core files it will load.
2326 Essentially it now requires .core files to descend from the same
2327 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2328 as the runtime does. (The intent is to prevent the crashes which
2329 can occur, and which can even be reported as mysterious failures,
2330 when people patch the sources or change the build parameters
2331 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2332 sbcl and .core files.)
2333 * fixed bug: VALUES-LIST is no longer optimized away.
2334 * fixed bug 142: The FFI conversion of C string values to Lisp
2335 string values no longer conses excessively. (thanks to Nathan
2336 Froyd porting Raymond Toy's fix to CMU CL)
2337 * began to systematize and improve MOP conformance in PCL (thanks to
2338 Nathan Froyd, Gerd Moellman and Pierre Mai):
2339 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2341 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2342 ** FINALIZE-INHERITANCE is now called on class finalization;
2343 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2344 argument precedence order.
2345 * fixed bug 202: The compiler no longer fails on functions whose
2346 derived types contradict their declared type.
2347 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2348 so it can be non-toplevel.
2349 * The fasl file version number has changed (because of the new
2350 implementation of DEFMACRO).
2351 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2352 safe code, that any length requirement by their type-specifier
2353 argument is valid. The exceptions to this are described in bug
2355 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2356 functions now check the types of their inputs as required by ANSI.
2357 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2358 introduced symbol is DECLAREd to be SPECIAL.
2359 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2360 * fixed bug: :COUNT argument to sequence functions may be negative
2361 * fixed bug: body of DO-SYMBOLS may contain declarations
2362 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2363 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2364 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2365 symbol macro only once
2366 * fixed printing of call frame when argument list is unavailable
2367 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2368 * compiler no longer signals WARNING on unknown keyword
2371 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2372 * A beta-quality port to the mips architecture running Linux,
2373 based on the old CMUCL backend, has been made. It has been tested
2374 on a big-endian kernel, and works sufficiently well to be able to
2375 rebuild itself; it has not been tested in little-endian mode.
2376 * fixed an inconsistency between gencgc.c and purify.c which made
2377 dumping/loading .core files unreliable
2378 * fixed bug 120a: The compiler now deals correctly with IFs where
2379 the consequent is the same as the alternative, instead of
2380 misderiving the return type. (thanks to Alexey Dejneka)
2381 * fixed bug 113: Logical pathnames are now dumpable (the logical
2382 host is resolved at load-time, throwing an error if it is not
2384 * fixed bug 174: FORMAT's error message is slightly clearer when a
2385 non-printing character is used in a format directive.
2386 * fixed several bugs in compiler checking of type declarations, i.e.
2387 violations of the Python "declarations are assertions" principle
2388 (thanks to Alexey Dejneka)
2389 * fixed several bugs in PCL's error checking (thanks to Gerd
2391 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2393 * fixed bug in compilation of functions as first class values
2394 (thanks to Antonio Martinez-Shotton)
2395 * The compiler's handling TYPE-ERRORs which it can prove will
2396 inevitably happen at runtime has been cleaned up and corrected
2397 in several ways. (thanks to Alexey Dejneka)
2398 * improved argument type checking for various basic arithmetic
2399 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2400 lavished on them in the past that they can be compiled in many
2401 ways in different special cases
2402 * fixed bug 181: compiler checks validity of user supplied type
2404 * cleaned up code flushing in optimization: Function calls which
2405 should signal errors for safety purposes (e.g. which ANSI says
2406 should signal errors when their arguments are of incorrect type)
2407 are no longer optimized away.
2408 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2409 * incremented fasl file version number, because changes in the
2410 implementation of sequence functions like COERCE caused
2411 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2412 in old inline expansions) to become undefined. (Actually these
2413 changes were later undone, so we might very well be binary
2414 compatible with 0.7.7 after all, but leaving the version number
2415 incremented seemed like the simplest and most conservative
2418 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2419 * An alpha-quality port to the parisc architecture running Linux,
2420 based on the old CMUCL backend, has been made. This, even more so
2421 than the other backends, should be considered still a work in
2422 progress; known problems include that the Linux kernel in 64-bit
2423 mode does not propagate the correct sigcontext structure to
2424 userspace, and consequently SBCL on a parisc64 kernel will not
2426 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2427 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2428 suppressed.) Also now that the compiler is looking at declarations
2429 in the environment, it checks optimization declarations as well,
2430 and suppresses inlining when (> DEBUG SPEED).
2431 * More fixes have been made to treatment of floating point exception
2432 treatment and other Unix signals. In particular, floating point
2433 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2434 * The detection and handling of control stack exhaustion (infinite
2435 or very deeply nested recursion) has changed. Stack exhaustion
2436 detection is now done by write-protecting pages at the OS level
2437 and applies at all optimization settings; when found, a
2438 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2439 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2440 be caught using IGNORE-ERRORS.
2441 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2442 correctly when transforming recursive LABELS functions to LETs.
2443 (thanks to Alexey Dejneka)
2444 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2445 that are names of constants or global variables.
2446 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2447 alien routines with docstrings.
2448 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2449 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2451 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2452 (thanks to Eric Marsden)
2453 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2454 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2455 * Bug fix: The SPARC backend can now compile functions involving
2456 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2457 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2458 to a functions expecting a PATHNAME-DESIGNATOR.
2459 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2460 lambda lists. (thanks to Alexey Dejneka)
2461 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2462 functions so that EXPORTing the name of the class doesn't cause
2463 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2464 to Antonio Martinez for reporting this)
2465 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2466 constant-folded without failing an assertion. (thanks to Einar
2467 Floystad Dorum for reporting this)
2468 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2469 types are now dealt with more correctly by the compiler.
2470 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2471 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2472 behaviour analogous to RENAME-FILE. This puts its behaviour more
2473 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2475 * The fasl file version number has changed again. (because of the
2476 bug fix involving the names of PCL MAKE-INSTANCE functions)
2478 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2479 * bug fix: Floating point exceptions are treated much more
2480 consistently on the x86/Linux and PPC/Linux platforms.
2481 * Array initialization with :INITIAL-ELEMENT is now much faster for
2482 cases when the compiler cannot open code the array creation, but
2483 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2484 array accesses have also seen a speed increase.
2485 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2486 specified. (This was at the root of some bad interactions between
2487 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2489 * bug fix: Internal error arguments for undefined functions are now
2490 computed correctly on the PPC/Linux platform.
2491 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2492 Raymond Toy's patch for CMU CL)
2493 * Support for the Solaris 9 operating environment has been included
2494 (thanks to Daniel Merritt)
2495 * A very ugly but hopefully complete draft of the missing FFI chapter
2496 of the manual has been created by reformatting the corresponding
2497 CMU CL manual chapter into (currently very ugly and incoherent)
2498 DocBook and bringing it up to date for SBCL behavior. Thus, the
2499 manual is now essentially complete, at least by my extreme
2500 once-and-only-once standards, whereby it's acceptable to refer to
2501 the doc strings of SB-EXT functions as the primary documentation.
2502 * The fasl file version number has changed again, due to cleanup of
2503 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2504 is no longer a static symbol.)
2506 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2507 * SBCL now builds with OpenMCL (version 0.12) as the
2508 cross-compilation host; also, more progress has been made toward
2509 bootstrapping under CLISP.
2510 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2512 * bug 158 fixed: The compiler can now deal with integer loop
2513 increments different from 1; fixing this turned out also to fix
2515 * bug 169 fixed: no more bogus warnings about using lexical bindings
2516 despite the presence of perfectly good SPECIAL declarations (thanks
2517 to David Lichteblau)
2518 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2519 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2520 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2522 * bug 180 fixed: Method combination specifications no longer ignore
2523 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2524 * bug fix: Structure type predicate functions now check their argument
2525 count as they should.
2526 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2527 correctly. (thanks to Pierre Mai)
2528 * minor incompatible change: The --noprogrammer option is deprecated
2529 in favor of the new --disable-debugger option, which is very similar.
2530 (The major difference is that it takes effect at a slightly different
2531 time at startup, causing handling of errors in --sysinit and
2532 --userinit files will be affected differently.) The
2533 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2534 been added to allow this functionality to be controlled from ordinary
2535 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2536 maintainers, who might want to run non-interactive scripts to
2537 build SBCL cores which will later be used interactively.)
2538 * minor incompatible change: The LOAD function no longer, when given
2539 a wild pathname to load, loads all files matching that pathname.
2540 Instead, an error of type FILE-ERROR is signalled.
2542 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2543 * bug 147 fixed: The compiler preserves its block link/count
2544 invariants more correctly now so that it doesn't crash. (thanks
2546 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2548 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2549 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2550 no-value-for-this-character as NIL (as specified by ANSI).
2551 * HOST-NAMESTRING on physical pathnames now returns a string that is
2552 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2553 (thanks to Christophe Rhodes)
2554 * The Alpha port handles icache flushing more correctly. (thanks to
2556 * More progress has been made toward bootstrapping under CLISP. (thanks
2557 to Christophe Rhodes)
2558 * The fasl file format has changed again, because dynamic loading
2559 on OpenBSD (which has non-ELF object files) motivated some cleanups
2560 in the way that foreign symbols are transformed and passed around.
2561 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2562 is no longer treated as whitespace by the reader, but instead as
2563 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2564 |A
\7fB|, instead of A as it used to.
2566 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2567 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2568 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2569 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2570 this as of 0.7.1.45, but was left out of the previous news section
2571 (thanks to Dan Barlow)
2572 * SBCL now runs on the Solaris operating system on SPARC architectures
2573 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2574 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2575 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2576 * SPARC backend cleanups, allowing builds of cores optimized for V8
2577 and V9 SPARCS, and also emission of code targeted to a particular
2578 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2580 * SBCL is closer to bootstrapping under CLISP, thanks to various
2581 fixes by Christophe Rhodes.
2582 * The fasl file format has changed again, to allow the compiler's
2583 INFO database to support symbol macros.
2584 * The user manual (in doc/) is formatted into HTML more nicely.
2585 (thanks to coreythomas)
2586 * The system is smarter about SUBTYPEP relationships, especially
2587 those involving NOT types (including types such as ATOM which are
2588 represented internally using NOT types). Thus SUBTYPEP is less
2589 likely to return (VALUES NIL NIL) in general, and in particular
2590 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2592 * The fasl file format has changed again, because the internal
2593 representation of types now includes a new slot to support the new
2594 SUBTYPEP-of-NOT-types logic.
2595 * (not a change in the main branch of SBCL, but a related prototype
2596 which can hopefully be merged into the main branch of SBCL in the
2597 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2598 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2599 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2600 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2601 dispatch macro character combinations. (thanks to Alexey Dejneka)
2602 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2603 that superficially look like logical namestrings correctly.
2604 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2606 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2607 default supplied-p) for &optional and &key arguments. (thanks to
2610 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2611 * incompatible change: The compiler is now less aggressive about
2612 tail call optimization, doing it only when (> SPACE DEBUG) or
2613 (> SPEED DEBUG). (This is an incompatible change because there are
2614 programs which relied on the old CMU-CL-style behavior to optimize
2615 away their unbounded recursion which will now die of stack overflow.)
2616 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2617 for non-GENCGC systems has been increased to 20M (since that
2618 seems much closer to the likely performance optimum for modern
2619 systems than the old 4M value was)
2620 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2621 and *DEBUG-PRINT-LEVEL*
2622 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2623 of CMU CL's support for SPARC, and various endianness and other
2624 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2625 * new syntactic sugar for the Unix command line: --load foo.bar is now
2626 an alternate notation for --eval '(load "foo.bar")'.
2628 ** The system now detects stack overflow and handles it gracefully,
2629 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2630 optimization settings. (This is a good thing in general, and
2631 its introduction in this version should be particularly timely
2632 for anyone whose code fails because of suppression of tail
2634 ** The system now hunts for the C variable "environ" in a more
2635 devious way, to avoid segfaults when the C library version differs
2636 between compile time and run time. (thanks to Christophe Rhodes)
2637 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2638 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2639 ** The compiler no longer issues bogus style warnings for undefined
2640 classes in the same source file as the DEFCLASSes which defined
2641 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2643 ** fixes in CONDITION class precedence list for undefined function
2644 errors (thanks to Alexei Dejneka)
2645 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2646 correctly. (thanks to Dan Barlow)
2647 ** portability fixes aiming at bootstrapping under CLISP (thanks
2648 to Dave McDonald and Christophe Rhodes)
2649 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2650 ** fixes in type translation and and type inference (thanks to
2652 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2653 ** various fixes in the new ports (thanks to Dan Barlow)
2654 * several changes related to debugging:
2655 ** suppression of tail recursion, as noted above
2656 ** stack overflow detection, as noted above
2657 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2658 is now the default. (For some time encapsulation has been more
2659 reliable than the breakpoint-based :ENCAPSULATE NIL
2660 implementation, at least on X86 systems; and I just noticed that
2661 encapsulation also seems closer to the spirit of the ANSI
2664 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2666 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2667 up properly again. (There was a packaging bug in 0.7.0 which
2668 left their definitions in SB-SYS::LOAD-FOREIGN and
2669 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2670 vital for most things which interface to C-level interfaces,
2671 like extensions working with sockets or databases or
2672 Perl-compatible regexes or whatever, and the need to fix
2673 this bug is the main reason that 0.7.1 was released so
2675 ** DEFGENERIC is now choosier about the methods it redefines, so that
2676 reLOADing a previously-LOADed file containing DEFGENERICs does
2677 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2678 cycle now works as it should. (thanks to Alexey Dejneka)
2679 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2680 fixed by Christophe Rhodes. (He actually submitted this patch
2681 months ago, and I delayed until after 0.7.0.)
2682 ** Bug 111 (internal compiler confusion about runtime checks on
2683 FUNCTION types) was fixed by Alexey Dejneka.
2684 * Some internal cleanups (getting rid of variables which aren't
2685 needed now that the byte interpreter is gone) caused the fasl
2686 file format number to change again.
2688 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2689 * major incompatible change: The default fasl file extension, i.e. the
2690 default extension for files produced by COMPILE-FILE, has changed
2691 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2693 ** There are many changes in the implementation of the compiler.
2694 SBCL is now essentially a compiler-only implementation of ANSI
2695 Common Lisp. EVAL still "interprets" a few special cases, but
2696 almost all the interesting cases are handled by creating
2697 a LAMBDA expression, calling COMPILE on it, then calling
2698 FUNCALL on the result.
2699 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2700 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2701 Since the code is newer, there might still be some new bugs
2702 (though not as many as before Martin Atzmueller's fixes:-). But
2703 the new code is substantially simpler and clearer, and hopefully
2704 any remaining bugs will be simpler, less fundamental, and more
2705 fixable then the bugs in the old code.
2706 ** The revised compiler is still a little unsteady on its feet.
2708 *** The debugging information it produces (particularly the names
2709 of FUNCTION objects) is sometimes much less useful than what
2710 the old compiler produced.
2711 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2712 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2713 within a MACROLET) has been temporarily weakened.
2714 ** There are new compiler optimizations for various functions:
2715 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2716 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2717 *** the math functions TRUNCATE, FLOOR, and CEILING
2718 *** the function-of-all-trades COERCE
2719 Mostly these should be transparent, but there's one
2720 potentially-annoying problem (bug 117): when the compiler
2721 inline-expands a function and does type analysis on the result,
2722 it can create control paths which have type mismatches, and
2723 when it can't prove that those control paths aren't taken,
2724 it will issue WARNINGs about the type mismatches. This is
2725 a particular problem in practice for the new sequence functions.
2726 It's not clear how this should be fixed, and for now, a
2727 workaround is given in the entry for 117 in the BUGS file.
2728 ** (Because of the interaction between the two previous items --
2729 occasional inlining problems and new inline expansions -- some
2730 of the new sequence function optimizations won't really kick in
2731 completely until debugging information, and then inlining, are
2732 straightened out in some future version.)
2733 * minor incompatible changes:
2734 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2735 DIRECTORY no longer implicitly promotes NIL slots of its
2736 pathname argument to :WILD. In particular, when you ask for the
2737 contents of a directory (which you used to be able to do without
2738 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2739 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2740 ** changes in behavior that ANSI explicitly defines to be
2741 implementation dependent:
2742 *** The new compiler-only implementation still conforms with ANSI,
2743 but acts a little different than before. Besides the obvious
2744 changes in performance tradeoffs (that the cost per form passed
2745 to EVAL has gone up, and the cost per form executed by EVAL
2746 has gone down), the behavior of the system changes a little
2747 because there are no longer any interpreted function objects.
2748 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2749 e.g. doing COMPILE on the output of interactive DEFUN is
2751 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2753 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2754 has changed from (:CL) to NIL.
2755 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2756 e.g. #\Newline and #\Tab, don't have names specified in the
2757 ANSI Common Lisp standard, is now based on their ASCII symbolic
2758 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2759 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2760 are no longer used for output.
2761 ** changes in internal implementation constants:
2762 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2763 4 million. (If your application spends a lot of time GCing and
2764 you have a lot of RAM, you might want to experiment with
2765 increasing it even more.)
2766 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2767 However, almost all old code should still continue to work without
2768 immediate update, as SB-C-CALL is now a (deprecated) nickname
2770 ** Old operator names in the style DEF-FOO are now deprecated in
2771 favor of new corresponding names DEFINE-FOO, for consistency with
2772 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2773 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2774 internal symbols, but a few supported extensions like
2775 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2776 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2777 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2778 etc. as you get deeper into recursive calls to the debugger
2779 command loop, instead of the old "5]", "5]]", "5]]]"
2780 sequence. (I was motivated to do this when squabbles between
2781 ILISP and SBCL left me very deeply nested in the debugger. In the
2782 short term, this change will probably provoke more ILISP/SBCL
2783 squabbles, but hopefully it will be an improvement in the long run.)
2784 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2785 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2786 function, since declaiming return types involving aliens is
2787 (1) annoyingly messy to do by hand and (2) vital to efficient
2788 compilation of code which calls such functions.
2789 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2790 longer reexported by the SB-EXT package. They're solely useful
2791 for alien code, so it seems more logical that you should get
2792 them from the SB-ALIEN package, not in SB-EXT.
2793 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2794 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2795 features. Instead, the code that they used to control is always
2796 built into the system.
2797 * many other bug fixes
2798 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2799 advantage of the new EVAL-WHEN stuff and to clean them up in
2800 general, and they are now more ANSI-compliant in a number of
2801 ways. Martin Atzmueller is responsible for a lot of this.
2802 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2804 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2805 *** correct ERROR type for various file operations
2806 *** some fixes for Lisp streams
2807 *** DEFMETHOD syntax checking
2808 *** changing old weird representation of debug information as
2809 strings (which, among their other deficiencies, don't transform
2810 correctly when you rename packages, and don't change their
2811 print representation when you change things like *PACKAGE*
2812 and *PRINT-LENGTH*) to symbols and lists of symbols
2813 He also made several improvements and fixed several bugs in DESCRIBE.
2814 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2816 *** misbehavior of WRITE-STRING/WRITE-LINE
2817 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2818 and several other LOOP problems as well
2819 *** DIRECTORY when similar filenames are present
2820 *** DEFGENERIC with :METHOD options
2821 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2822 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2823 *** argument ordering in FIND with :TEST option
2824 *** mishandled package designator argument in APROPOS-LIST
2825 *** various problems in the backquote readmacro
2826 *** a bug in APROPOS
2827 *** probably some others that I'm not describing very well here,
2828 since the CVS log documents them by reference to sbcl-devel
2829 messages, and the SourceForge archives aren't working well.:-(
2830 ** Dan Barlow improved the Alpha port (and is making progress on the
2831 PPC port, for those of you who think different).
2832 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2833 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2834 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2835 continues to work on the SPARC port (for those of you in a position
2836 to look down upon our little PC-compatible boxes from a great height).
2837 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2838 and uses that copy, rather than the current dynamic value, when
2839 it's trying to decide whether to truncate output. Thus e.g.
2840 (let ((*print-lines* 50))
2841 (pprint-logical-block (stream nil)
2843 (let ((*print-lines* 8))
2844 (print (aref possiblybigthings i) stream)))))
2845 should now truncate the logical block only at 50 lines, instead of
2846 often truncating it at 8 lines, as it did before.
2847 * The doc/cmucl/ directory, containing old CMU CL documentation
2848 from the time of the fork, is no longer part of the base system.
2849 SourceForge has shut down its anonymous FTP service, and with it
2850 my original plan for distributing the old CMU CL documentation
2851 there. For now, if you need these files you can download an old
2852 SBCL source release and extract them from it.
2853 * The fasl file version number changed again, for dozens of reasons,
2854 some of which are apparent above.
2856 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2857 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2858 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2859 patch, so that DISASSEMBLE output is much nicer.
2860 * The code in the SB-PROFILE package now seems reasonably stable.
2861 I still haven't decided what the final interface should look like
2862 (I'd like PROFILE to interact cleanly with TRACE, since both
2863 facilities use function encapsulation) but if you have a need
2864 for profiling now, you can probably use it successfully with
2865 the current CMU-CL-style interface.
2866 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2867 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2868 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2869 link now returns the dangling link itself, and for similar
2870 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2871 link itself. (In these cases the old code signalled an error and
2872 looped endlessly, respectively.) Thus, DIRECTORY now works even
2873 in the presence of dangling and cyclic symbolic links.
2874 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2875 is now a supported extension again, since the consensus on
2876 sbcl-devel was that it can be useful for ordinary development
2877 work, not just for debugging SBCL itself.
2878 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2879 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2880 that currently-defined functions might be redefined later with
2881 different return types.
2882 * Hash tables can be printed readably, as inspired by CMU CL code
2883 of Eric Marsden and SBCL code of Martin Atzmueller.
2884 * better error handling in CLOS method combination, thanks to
2885 Martin Atzmueller porting Pierre Mai's CMU CL patches
2886 * more overflow fixes for >16Mbyte I/O buffers
2887 * A bug in READ has been fixed, so that now a single Ctrl-D
2888 character suffices to cause end-of-file on character streams.
2889 In particular, now you only need one Ctrl-D at the command
2890 line (not two) to exit SBCL.
2891 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2893 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2894 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2895 does the right thing.
2896 * The implementation of some type tests, especially for CONDITION
2897 types, is now tidier and maybe faster, due to CMU CL code
2898 originally by Douglas Crosher, ported by Martin Atzmueller.
2899 * Some math functions have been fixed, and there are new
2900 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2901 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2902 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2903 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2904 likely to be built into sbcl-0.7.0.)
2905 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2906 CL, ported to SBCL by Martin Atzmueller)
2907 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2908 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2909 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2910 * There's a new slam.sh hack to shorten the edit/compile/debug
2911 cycle for low-level changes to SBCL itself, and a new
2912 :SB-AFTER-XC-CORE target feature to control the generation of
2913 the after-xc.core file needed by slam.sh.
2914 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2915 COMPILE-FILE is no longer supported, so that now every function
2916 gets an entry point, so that block compilation looks a little
2917 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2918 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2919 returns the number of bytes consed since the system started,
2920 rather than the number consed since the first time the function
2921 was called. (The new definition parallels ANSI functions like
2922 CL:GET-INTERNAL-RUN-TIME.)
2923 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2924 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2925 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2926 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2927 * Fasl file version numbers are now independent of the target CPU,
2928 since historically most system changes which required version
2929 number changes have affected all CPUs equally. Similarly,
2930 the byte fasl file version is now equal to the ordinary
2933 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2934 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2935 is no longer recognized. I apologize for this, because it was
2936 listed in SB-EXT as a supported extension, but I found that
2937 its existing behavior was poorly specified, as well as incorrectly
2938 specified, and it looked like too much of a mess to straighten it
2939 out. I have enough on my hands trying to get ANSI stuff to work..
2940 * many patches ported from CMU CL by Martin Atzmueller, with
2941 half a dozen bug fixes in pretty-printing and the debugger, and
2942 half a dozen others elsewhere
2943 * fixed bug 13: Floating point infinities are now supported again.
2944 They might still be a little bit flaky, but thanks to bug reports
2945 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2946 as flaky as they were.
2947 * The --noprogrammer command line option is now supported. (Its
2948 behavior is slightly different in detail from what the old man
2949 page claimed it would do, but it's still appropriate under the
2950 same circumstances that the man page talks about.)
2951 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2952 are now supported, and enabled by default. Thus, the compiler can
2953 handle many floating point and complex operations much less
2954 inefficiently. (Thus e.g. you can implement a complex FFT
2956 * The compiler now detects type mismatches between DECLAIM FTYPE
2957 and DEFUN better, and implements CHECK-TYPE more correctly, and
2958 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2959 thanks to patches from Martin Atzmueller.
2960 * various fixes to make the cross-compiler more portable to
2961 ANSI-conforming-but-different cross-compilation hosts (notably
2962 Lispworks for Windows, following bug reports from Arthur Lemmens)
2963 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2964 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2965 fixed thanks to Pierre Mai's CMU CL patches.
2966 * fixes to keep the system from overflowing internal counters when
2967 it tries to use i/o buffers larger than 16M bytes
2968 * fixed bug 45a: Various internal functions required to support
2969 complex special functions have been merged from CMU CL sources.
2970 (When I was first setting up SBCL, I misunderstood a compile-time
2971 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2972 * improved support for type intersection and union, fixing bug 12
2973 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2974 more obscure bugs as well
2975 * some steps toward byte-compiling non-performance-critical
2976 parts of the system, courtesy of patches from Martin Atzmueller
2977 * Christophe Rhodes has made some debian packages of sbcl at
2978 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2979 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2980 stable, but are nonetheless usable. When he's ready, I'd be happy
2981 to add them to the SourceForge "File Releases" section. (And if
2982 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2983 * new fasl file format version number (because of changes in
2984 internal representation of (OR ..) types to accommodate the new
2985 support for (AND ..) types, among other things)
2987 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2988 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2990 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2992 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2993 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2994 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2995 * DESCRIBE now works on structure objects again.
2996 * Most function call argument type mismatches are now handled as
2997 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2998 know whether the function will be redefined before the call is
2999 executed. (The compiler could flag local calls with full WARNINGs,
3000 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3001 it doesn't keep track of enough information to know whether calls
3002 are local in this sense.)
3003 * Compiler output is now more verbose, with messages truncated
3004 later than before. (There should be some supported way for users
3005 to override the default verbosity, but I haven't decided how to
3006 provide it yet, so this behavior is still controlled by the internal
3007 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3008 src/compiler/ir1util.lisp.)
3009 * Fasl file format version numbers have increased again, because
3010 support for the Gray streams extension changes the layout of the
3011 system's STREAM objects.
3012 * The Gray subclassable streams extension now works, thanks to a
3013 patch from Martin Atzmueller.
3014 * The full LOAD-FOREIGN extension (not just the primitive
3015 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3016 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3017 but like most other programs, it defaults to copying the Unix
3018 environment from the original process instead of starting the
3019 new process in an empty environment.
3020 * Extensions which manipulate the Unix environment now support
3021 an :ENVIRONMENT keyword option which doesn't smash case or
3022 do other bad things. The CMU-CL-style :ENV option is retained
3023 for porting convenience.
3024 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3025 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3027 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3029 * A patch from Martin Atzmueller seems to have solved the SIGINT
3030 problem, and as far as we know, signal-handling now works cleanly.
3031 (If you find any new bugs, please report them!)
3032 * The system no longer defaults Lisp source file names to types
3033 ".l", ".cl", or ".lsp", but only to ".lisp".
3034 * The compiler no longer uses special default file extensions for
3035 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3036 seems to expect a single default extension for all compiled code,
3037 and there's no compelling reason to try to stretch the standard
3038 to allow two different extensions.) Instead, byte-compiled files
3039 default to the same extension as native-compiled files.
3040 * Fasl file format version numbers have increased again, because
3041 a rearrangement of internal implementation packages made some
3042 dumped symbols in old fasl files unreadable in new cores.
3043 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3044 many fewer weird special cases.
3045 * Bug #17 (differing COMPILE-FILE behavior between logical and
3046 physical pathnames) has been fixed, and some related misbehavior too,
3047 thanks to a patch from Martin Atzmueller.
3048 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3049 by Tim Moore, ported to SBCL by Martin Atzmueller.
3050 * Martin Atzmueller fixed several filesystem-related problems,
3051 including bug #36, in part by porting CMU CL patches, which were
3052 written in part by Paul Werkowski.
3053 * More compiler warnings in src/runtime/ are gone, thanks to
3054 more patches from Martin Atzmueller.
3055 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3058 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3060 * DESCRIBE now works on CONDITION objects.
3061 * The debugger now handles errors which arise when trying to print
3062 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3064 * The build system now uses an additional file, customize-target-features.lisp,
3065 to allow local modifications to the target *FEATURES* list. (The point of
3066 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3067 debugging features enabled, without having to worry about propagating it
3068 into everyone's system when I do a "cvs update".) When no
3069 customize-target-features.lisp file exists, the target *FEATURES* list
3070 should be constructed the same way as before.
3071 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3072 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3073 warn about behavior which is undefined under the ANSI spec. Note: This
3074 is specified by ANSI, but it's not very popular with programmers.
3075 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3076 macro in the SBCL sources for an example of a workaround which you
3077 might use to make portable ANSI-standard code which does what you want.
3078 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3079 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3080 when it's used as a non-toplevel form. (This is required in order
3081 to implement the DEFCONSTANT-EQX macro.)
3082 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3083 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3084 funny things, due to bugs in the implementation of EVAL-WHEN
3085 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3086 non-toplevel DEFCONSTANTs are uncommon.)
3087 * The core file version number and fasl file version number have been
3088 incremented, because the old noncompliant DEFCONSTANT behavior involved
3089 calling functions which no longer exist, and because I also took the
3090 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3091 * fixed bug 1 (error handling before read-eval-print loop starts), and
3092 redid debugger restarts and related debugger commands somewhat while
3094 ** The QUIT debugger command is gone, since it did something
3095 rather different than the SB-EXT:QUIT command, and since it never
3096 worked properly outside the main toplevel read/eval/print loop.
3097 Invoking the new TOPLEVEL restart provides the same functionality.
3098 ** The GO debugger command is also gone, since you can just invoke
3099 the CONTINUE restart directly instead.
3100 ** The TOP debugger command is also gone, since it's redundant with the
3101 FRAME 0 command, and since it interfered with abbreviations for the
3103 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3105 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3106 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3107 some porting work by Martin Atzmueller)
3108 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3109 with a boatload of patches which he ported to SBCL. Now that those
3111 ** The system tries to make sure that its low-priority messages
3112 are prefixed by semicolons, to help people who like to use
3113 syntax highlighting in their ILISP buffer. (This patch
3114 was originally due to Raymond Toy.)
3115 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3116 patches originally written by Raymond Toy.
3117 ** The compiler understands coercion between single-value and
3118 multiple-VALUES type expressions better, getting rid of some very
3119 weird behavior, thanks to patches originally by Robert MacLachlan
3120 and Douglas Crosher.
3121 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3122 lambda lists, thanks to a patch originally by Pierre Mai.
3123 ** The system no longer bogusly warns about "abbreviated type
3125 ** The compiler gets less confused by inlining and RETURN-FROM,
3126 thanks to some patches originally by Tim Moore.
3127 ** The system no longer hangs when dumping circular lists to fasl
3128 files, thanks to a patch originally from Douglas Crosher.
3129 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3130 undefined function error.
3131 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3132 in structure slot initforms) for now, documented workaround instead:-|
3133 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3134 * fixed bug 5: added stubs for various Gray stream functions called
3135 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3136 installed, at least appropriate type errors are generated
3137 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3138 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3139 consistently in DEFMETHOD forms.
3140 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3141 it doesn't seem to affect SBCL after all
3142 * The C runtime system now builds with better optimization and many
3143 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3145 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3147 * The system is now under CVS at SourceForge (instead of the
3148 CVS repository on my home machine).
3149 * The new signal handling code has been tweaked to treat register
3150 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3151 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3152 downstream, e.g. in debug-int.lisp, has implicit dependencies
3153 on the unsignedness of integer representation of machine words,
3154 and that caused the system to bomb out with infinite regress
3155 when trying to recover from type errors involving signed values,
3156 e.g. (BUTLAST '(1 2 3) -1).)
3157 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3158 (This was one of the bugs Peter Van Eynde reported back in July.)
3159 * The system now uses code inspired by Colin Walters' O(N)
3160 implementation of MAP (from the cmucl-imp@cons.org mailing
3161 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3162 inline the MAP operation, and there is more than one
3163 sequence argument to the MAP call (so that it can't just
3164 do ETYPECASE once and for all based on the type of the
3165 single sequence argument). (The old non-inline implementation
3166 of the general M-argument sequence-of-length-N case required
3167 O(M*N*N) time when any of the sequence arguments were LISTs.)
3168 * The QUIT :UNIX-CODE keyword argument has been renamed to
3169 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3170 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3172 * Raymond Wiker's patches to port dynamic loading from Linux to
3173 FreeBSD have been added.
3174 * The BUGS file is now more nearly up to date, thanks in large part
3175 to Martin Atzmueller's review of it.
3176 * The debugger now flushes standard output streams before it begins
3177 its output ("debugger invoked" and so forth).
3178 * The core version number and fasl file version number have both
3179 been incremented, because of incompatible changes in the layout
3181 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3182 used to not be called for a saved Lisp image.)
3183 * Martin Atzmueller's version of a patch to fix a compiler crash,
3184 as posted on sbcl-devel 13 September 2000, has been installed.
3185 * Instead of installing Martin Atzmueller's patch for the
3186 compiler transform for SUBSEQ, I deleted the compiler transform,
3187 and transforms for some similar consing operations.
3188 * A bug in signal handling which kept TRACE from working on OpenBSD
3190 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3191 away by constant folding
3192 * The system now defines its address space constants in one place
3193 (in the Lisp sources), and propagates them automatically elsewhere
3194 (through GENESIS and the sbcl.h file). Therefore, patching the
3195 address map is less unnecessarily tedious and error-prone. The
3196 Lisp names of address space constants have also been systematized.
3197 * CVS tags like dollar-Header-dollar have been removed from
3198 the sources, because they have never saved me trouble and
3199 they've been source of trouble working with patches and other
3200 diff-related operations.
3201 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3202 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3204 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3206 * The system has been ported to OpenBSD.
3207 * The system now compiles with a simple "sh make.sh" on the systems
3208 that it's supported on. I.e., now you no longer need to tweak
3209 text in the target-features.lisp-expr and symlinks in src/runtime/
3210 by hand, the make.sh takes care of it for you.
3211 * The system is no longer so grossly inefficient when compiling code
3212 involving vectors implemented as general (not simple) vectors (VECTOR T),
3213 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3214 which dares to use the various sequence functions on non-simple
3215 vectors, takes less of a performance hit.
3216 * There is now a primitive type predicate VECTOR-T-P
3217 to test for the (VECTOR T) type, so that e.g.
3218 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3219 can now be compiled with some semblance of efficiency. (The old code
3220 turned the type declaration into a full call to %TYPEP at runtime!)
3221 * AREF on (VECTOR T) is still not fast, since it's still compiled
3222 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3223 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3224 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3225 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3226 without making full calls to SUBTYPEP at runtime.
3227 (Some analogous efficiency issues for non-simple vectors specialized to
3228 element types other than T, or for non-simple multidimensional arrays,
3229 have not been addressed. They could almost certainly be handled the
3230 same way if anyone is motivated to do so.)
3231 * The changes in array handling break binary compatibility, so
3232 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3233 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3234 instead of (VALUES T T).
3235 * By following the instructions that Dan Barlow posted to sbcl-devel
3236 on 2 July 2000, I was able to enable primitive dynamic object
3237 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3238 functionality is not implemented (since it calls ld to resolve
3239 library references automatically, requiring RUN-PROGRAM for its
3240 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3241 try to resolve library references) is now supported.
3242 * The system now flushes the standard output streams when it terminates,
3243 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3244 them at several other probably-convenient times, e.g. in each pass of
3245 the toplevel read-eval-print loop, and after evaluating a form given
3246 as an "--eval" command-line option. (These changes were motivated by a
3247 discussion of stream flushing issues on cmucl-imp in August 2000.)
3248 * The source transform for TYPEP of array types no longer assumes
3249 that an array whose element type is a not-yet-defined type
3250 is implemented as an array of T, but instead punts, so that the
3251 type will be interpreted at runtime.
3252 * There is now some support for cross-compiling in make.sh: each of
3253 the phases of make.sh has its own script. (This should be transparent
3254 to people doing ordinary, non-cross-compile builds.)
3255 * Since my laptop doesn't have hundreds of megabytes of memory like
3256 my desktop machine, I became more motivated to do some items on
3257 my to-do list in order to reduce the size of the system a little:
3258 ** Arrange for various needed-only-at-cold-init things to be
3259 uninterned after cold init. To support this, those things have
3260 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3261 symbols with such names are now uninterned after cold init).
3262 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3263 fasl files for cold load.
3264 ** Remove the old compiler structure pooling code (which used to
3265 be conditional on the target feature :SB-ALLOC) completely.
3266 ** Redo the representation of some data in cold init to be more compact.
3267 (I also looked into supporting byte compiled code at bootstrap time,
3268 which would probably reduce the size of the system a lot, but that
3269 looked too complicated, so I punted for now.)
3270 * The maximum signal nesting depth in the src/runtime/ support code has
3271 been reduced from 4096 to 256. (I don't know any reason for the very
3272 large old value. If the new smaller value turns out to break something,
3273 I'll probably just bump it back up.)
3274 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3276 * Many, many bugs reported by Peter Van Eynde have been added to
3277 the BUGS list; some have even been fixed.
3278 * While enabling dynamic object file loading, I tried to make the
3279 code easier to understand, renaming various functions and variables
3280 with less ambiguous names, and changing some function calling
3281 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3282 * While trying to figure out how to do the OpenBSD port, I tried to
3283 clean up some of the code in src/runtime/. In particular, I dropped
3284 support for non-POSIX signal handling, added various comments,
3285 tweaked the code to reduce the number of compilation warnings, and
3286 renamed some files to increase consistency.
3287 * To support the new automatic configuration functionality in make.sh,
3288 the source file target-features.lisp-expr has been replaced with the
3289 source file base-target-features.lisp-expr and the machine-generated
3290 file local-target-features.lisp-expr.
3291 * fixed a stupid quoting error in make.sh so that using CMU CL
3292 "lisp -batch" as cross-compilation host works again
3294 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3296 * DESCRIBE no longer tries to call itself recursively to describe
3297 bound/fbound values, so that it no longer fails on symbols which are
3298 bound to themselves (like keywords, T, and NIL).
3299 * DESCRIBE now works on generic functions.
3300 * The printer now prints less-screwed-up representations of closures
3301 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3302 * A private symbol is used instead of the :EMPTY keyword previously
3303 used to mark empty slots in hash tables. Thus
3304 (DEFVAR *HT* (MAKE-HASH-TABLE))
3305 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3306 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3307 now does what ANSI says that it should. (You can still get
3308 similar noncompliant behavior if bang on the hash table
3309 implementation with all the symbols you get back from
3310 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3311 This breaks binary compatibility, since tests for equality to
3312 :EMPTY are wired into things like the macroexpansion of
3313 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3315 * There's now a minimal placeholder implementation for CL:STEP,
3316 as required by ANSI.
3317 * An obscure bug in the interaction of the normal compiler, the byte
3318 compiler, inlining, and structure predicates has been patched
3319 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3320 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3322 * Missing ordinary arguments in a macro call are now detected even
3323 when the macro lambda list contains &KEY or &REST.
3324 * The debugger no longer complains about encountering the top of the
3325 stack when you type "FRAME 0" to explicitly instruct it to go to
3326 the top of the stack. And it now prints the frame you request even
3327 if it's the current frame (instead of saying "You are here.").
3328 * As specified by ANSI, the system now always prints keywords
3329 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3331 * The default initial SIZE of HASH-TABLEs is now smaller.
3332 * Type information from CLOS class dispatch is now propagated
3333 into DEFMETHOD bodies, so that e.g.
3334 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3336 is now basically equivalent to
3337 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3338 (DECLARE (TYPE SINGLE-FLOAT X))
3340 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3341 operation, without having to do run-time type dispatch.
3342 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3343 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3344 * Since I don't seem to be making big file reorganizations very often
3345 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3346 is overflowing my ability to conveniently back them up), I've finally
3347 checked the system into CVS. (The CVS repository is on my home system,
3348 not at SourceForge -- putting it on SourceForge might come later.)
3349 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3350 high-level GC-NOTIFY-FOO functions send their output. (There's
3351 still very little control of where low-level verbose GC functions
3352 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3353 less than it used to -- the GC-NOTIFY-FOO functions are now under
3354 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3355 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3356 in only one place in the source code, and propagates it automatically
3357 everywhere that it's needed. Thus e.g. when I bump the version from
3358 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3359 * The C source files now include boilerplate legalese and documentation
3360 at the head of each file (just as the Lisp source files already did).
3361 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3362 to his page will be replaced with a link to his new CLiki service.
3364 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3366 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3367 * The build process now looks for GNU make under the default name "gmake",
3368 instead of "make" as it used to. If GNU make is not available as "gmake"
3369 on your system, you can change this default behavior by setting the
3370 GNUMAKE environment variable.
3371 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3372 can build without error under CMU CL.
3374 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3376 * There is now a partial SBCL user manual (with some new text and some
3377 text cribbed from the CMU CL manual).
3378 * The beginnings of a profiler have been added (starting with the
3379 CMU CL profiler and simplifying and cleaning up). Eventually the
3380 main interface should be through the TRACE macro, but for now,
3381 it's still accessed through vaguely CMU-CL-style functions and macros
3382 exported from the package SB-PROFILE.
3383 * Some problems left over from porting CMU CL to the new
3384 cross-compilation bootstrap process have been cleaned up:
3385 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3386 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3387 part when getting rid of the compiler *BACKEND* stuff.)
3388 ** The value of *NULL-TYPE* was screwed up, because it was
3389 being initialized before the type system knew the final
3390 definition of the 'NULL type. This screwed up several key
3391 optimizations in the compiler, causing inefficiency in all sorts
3392 of places. (I found it because I wanted to understand why
3393 GET-INTERNAL-RUN-TIME was consing.)
3394 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3395 PROCLAIM FTYPE information. Unfortunately this broke binary
3396 compatibility again, since now the forms output by DEFGENERIC
3397 to refer to functions which didn't exist in 0.6.3.
3398 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3399 can use the new (as of 0.6.3) transform for SOME into MAP into
3401 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3402 values to (INTEGER 0 1000000), so that the time code will no longer
3403 occasionally get blown up by Linux returning 1000000 microseconds
3404 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3405 its output conform to the ANSI spec. (Alas, this makes its output
3406 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3407 * A full call to MAP NIL with a single sequence argument no longer conses.
3408 * fixes to problems pointed out by Martin Atzmueller:
3409 * The manual page no longer talks about multiprocessing as though
3410 it were currently supported.
3411 * The ILISP support patches have been removed from the distribution,
3412 because as of version 5.10.1, ILISP now supports SBCL without us
3413 having to maintain patches.
3414 * added a modified version of Raymond Toy's recent CMU CL patch for
3415 EQUALP comparison of HASH-TABLE
3417 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3419 * The system still can't cross-compile itself with
3420 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3421 entails), but at least it can compile more of itself that way
3422 than it used to be able to, and various buglets which were uncovered
3423 by trying to cross-compile itself that way have now been fixed.
3424 * This release breaks binary compatibility again. This time
3425 at least I've incremented the FASL file format version to 2, so that the
3426 problem can be detected reliably instead of just causing weird errors.
3427 * various new style warnings:
3428 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3429 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3430 ** using the lexical binding of a variable named in the *FOO* style
3431 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3432 as specified by ANSI.
3433 * *RANDOM-STATE* is no longer automatically initialized from
3434 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3435 default behavior of the system is to repeat its behavior every time
3436 it's run. If you'd like to change this behavior, you can always
3437 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3438 old convention there was no comparably easy way to get the system to
3439 repeat its behavior every time it was run.
3440 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3441 FTYPE declarations has been removed, in favor of their ANSI
3442 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3443 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3444 the types of their sequence arguments can be determined at compile time.
3445 This is done through a new open code expansion for MAP which eliminates
3446 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3447 argument types can be determined at compile time.
3448 * The optimizer now transforms COERCE into an identity operation when it
3449 can prove that the coerced object is already of the correct type. (This
3450 can be a win for machine generated code, including the output of other
3451 optimization transforms, such as the MAP transform above.)
3452 * Credit information has been moved from source file headers into CREDITS.
3453 * Source file headers have been made more standard.
3454 * The CASE macro now compiles without complaining even when it has
3457 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3459 * (Note that the way that the PCL macroexpansions were rewritten
3460 to accommodate the change in DEFGENERIC below breaks binary
3461 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3462 not run under sbcl-0.6.2. Once we get out of alpha releases,
3463 i.e. hit release 1.0.0, we'll probably try to maintain binary
3464 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3465 and sbcl-1.4.4. Until then, however, it might be fairly common
3466 for maintenance releases to break binary compatibility.)
3467 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3469 * The reporting of unhandled signals has been changed to print some
3470 explanatory text as well as the report form. (Previously only
3471 the report form was printed.)
3472 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3473 it defines, so that the compiler no longer issues undefined function
3474 warnings for compiled-but-not-yet-loaded generic functions.
3475 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3476 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3477 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3478 section of the ANSI spec are supported.
3479 * The "" nickname for the "KEYWORD" package has been removed.
3480 The reader still handles symbol tokens which begin with a package marker
3481 as keywords, but it doesn't expose its mechanism for doing so in the
3482 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3483 * The system now issues STYLE-WARNINGs for contradictory TYPE
3484 proclamations. (Warnings for contradictory FTYPE proclamations would
3485 be nice too, but those can't be done usefully unless the type system
3486 is made smarter about FUNCTION types.)
3487 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3488 systematized, so that "*target-*.lisp is supposed to exist only on the
3489 target and imply that there's a related file which exists on the
3490 host, and *host-*.lisp is supposed to exist only on the host and imply
3491 that there's a related file which exists on the target. This involves a
3492 lot of renaming. Hopefully the acute confusion caused by the renaming
3493 will be justified by the reduction in chronic confusion..
3494 ** runtime-type.lisp -> early-target-type.lisp
3495 ** target-type.lisp -> late-target-type.lisp
3496 ** early-host-format.lisp -> early-format.lisp
3497 ** late-host-format.lisp -> late-format.lisp
3498 ** host-error.lisp -> misc-error.lisp
3499 ** early-error.lisp -> early-target-error.lisp
3500 ** late-error.lisp -> late-target-error.lisp
3501 ** host-defboot.lisp -> early-defboot.lisp
3502 ** code/misc.lisp -> code/target-misc.lisp
3503 ** code/host-misc.lisp -> code/misc.lisp
3504 ** code/numbers.lisp -> code/target-numbers.lisp
3505 ** code/early-numbers.lisp -> numbers.lisp
3506 ** early-host-type.lisp -> early-type.lisp
3507 ** late-host-type.lisp -> late-type.lisp
3508 ** host-typep.lisp -> typep.lisp
3509 ** load.lisp -> target-load.lisp
3510 ** host-load.lisp -> load.lisp
3511 ** host-disassem.lisp -> disassem.lisp
3512 ** host-insts.lisp -> insts.lisp
3513 ** byte-comp.lisp -> target-byte-comp.lisp
3514 ** host-byte-comp.lisp -> byte-comp.lisp
3515 ** host-signal.lisp -> signal.lisp
3516 ** host-defstruct.lisp -> defstruct.lisp
3517 ** late-target-type.lisp -> deftypes-for-target.lisp
3518 Furthermore, several other previously target-only files foo.lisp (e.g.
3519 hash-table.lisp and random.lisp) have been split into a target-and-host
3520 foo.lisp file and a target-only target-foo.lisp file, with their key type
3521 definitions in the target-and-host part, so that the cross-compiler will
3522 know more about target types.
3523 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3524 gone away. In their place are various *BACKEND-FOO* variables
3525 corresponding to the slots of the old structure.
3526 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3527 into the target SBCL has been fixed.
3528 * The chill.lisp system for loading cold code into a running SBCL
3530 * Support for the CMU CL "scavenger hook" extension has been removed.
3531 (It was undocumented and unused in the CMU CL sources that SBCL was
3532 derived from, and stale in sbcl-0.6.1.)
3533 * Various errors in the cross-compiler type system were detected
3534 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3535 (enabling various consistency checks). Many of them were fixed,
3536 but some hard problems remain, so the compiler is back to
3537 running without *TYPE-SYSTEM-INITIALIZED* for now.
3538 * As part of the cross-compiler type system cleanup, I implemented
3539 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3540 * I have started adding UNCROSS calls throughout the type system
3541 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3542 on cross-compiler input files will be able to go away, and various
3544 * CONSTANTP now returns true for quoted forms (as explicitly required
3547 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3549 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3550 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3551 (MAKE-LIST -1). In the longer term, it ought to become true
3552 that declarations are assertions even at SAFETY 1. For now, it's not
3553 quite true even at SAFETY 3, but it's at least more nearly true..
3554 (Note that this change seems to increases the size of the system by
3555 O(5%) and to decrease the speed of the compiler by 20% or more.)
3556 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3557 for the problem of printing dozens of lines of distracting information
3558 about low-level system machinery as part of the top stack frame
3559 on entry to the debugger when an undefined function was called.
3560 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3561 is not reset to COMMON-LISP-USER.
3562 * Compilation of stuff related to dyncount.lisp has been made conditional
3563 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3564 smaller. The various dyncount-related symbols have been moved into
3565 a new "SB-DYNCOUNT" package.
3566 * tty-inspect.lisp has been renamed to inspect.lisp.
3567 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3568 feature has gone away. (When we eventually port to other flavors of
3569 libc and/or Unix, we'll try to make the differences between flavors
3570 invisible at the user level.)
3571 * Various other *FEATURES* tags, and/or their associated conditionals,
3572 have been removed if obsolescent, or given better documentation, or
3573 sometimes given more-mnemonic names.
3575 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3577 * tidied up "make.sh" script
3578 * tidied up system directory structure
3579 * better "clean.sh" behavior
3580 * added doc/FOR-CMUCL-DEVELOPERS
3581 * many many small tweaks to output format, e.g. removing possibly-confusing
3582 trailing #\. character in DESCRIBE-INSTANCE
3583 * (EQUALP #\A 'A) no longer signals an error.
3584 * new hashing code, including EQUALP hashing
3585 * tidied up Lisp initialization and toplevel
3586 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3587 * command line argument processing
3588 * added POSIX-GETENV function to deal with Unix-ish environment variables
3589 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3590 terminating SBCL on EOF
3591 * non-verbose GC by default
3592 * There is no more "sbcl" shell script; the sbcl file is now the C
3593 runtime executable (just like CMU CL).
3594 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3596 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3597 symbol and package information as Lisp native symbol and package objects
3598 instead of strings naming symbols and strings naming packages. This way,
3599 whenever packages are renamed (as in warm init), debug information is
3600 transformed along with everything else.
3601 * tweaked the optimization policy declarations which control the building
3602 of SBCL itself. Now, among other things, the system no longer saves
3603 source location debugging information. (This helps two problems at once
3604 by reducing SBCL size and by keeping SBCL from trying to look for its
3605 sources -- which may not exist -- when reporting errors.)
3606 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3607 debugging and testing purposes
3608 * cleaned up printing, making the printer call PRINT-OBJECT for
3609 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3610 methods, giving nearly-ANSI behavior
3611 * converted almost all special variables to use *FOO* naming convention
3612 * deleted PARSE-TIME functionality, since it can be done portably
3613 * moved some files out of cold init into warm init
3614 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3616 * regularized formatting of source files
3617 * added an install.sh script
3618 * fixed ridiculous memory usage of cross-compiler by making
3619 compiler/alloc.lisp not try to do pooling unless it can hook
3620 itself into the GC of the cross-compilation host. Now the system
3621 builds nicely on my old laptop.
3622 * added :SB-ALLOC in target-features.lisp-expr
3623 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3624 was not implemented)
3625 * re-did condition handling and note reporting in the compiler. Notes
3626 are no longer handled by signalling conditions. Style warnings
3627 and warnings are handled more correctly and reported in such a way
3628 that it's easy to find one or the other in your output (so that you
3629 can e.g. figure out which of many problems caused COMPILE-FILE to
3631 * changed the severity of several compiler warnings from full WARNING
3632 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3633 compiler note reporting so that it doesn't use the condition system
3634 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3635 COMPILE-FILE command)
3636 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3637 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3638 no longer supported, since their implementation was deeply intertwingled
3639 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3640 * removed bogus "support" for compiler macros named (SETF FOO), and
3641 removed the compiler macro for SETF INFO (but only after making a fool
3642 of myself on the cmucl-imp mailing list by posting a bogus patch for
3643 DEFINE-COMPILER-MACRO..)
3644 * Compiled files containing forms which have side effects on the Lisp
3645 reader (such as DEFPACKAGE forms) are now handled more correctly.
3646 (Compiler queuing of top level lambdas has been suppressed by setting
3647 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3648 * deleted various currently-unused source files, e.g. gengc.lisp. They
3649 may be added back at some point e.g. when porting to other architectures,
3650 but until they are it's distracting to distribute them and to try to
3652 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3653 there were so many of them they're just distractions, and UNCROSS is
3654 known to be able to handle the current sources
3655 * moved PROFILE functionality into TRACE, so that it will be clear
3656 how the wrapping and unwrapping of functions when you profile them
3657 interacts with the wrapping and unwrapping of functions when you
3658 trace them. (Actually, the functionality isn't there yet, but at least
3659 the interface specification is there. Hopefully, the functionality will
3660 arrive with some future maintenance release.)
3661 * removed host-oops.lisp
3662 * changed signature of QUIT function to allow UNIX-CODE argument
3663 * fixed READ-SEQUENCE bug
3664 * tweaked verbose GC output so that it looks more like the progress
3665 output that ANSI specifies for functions like LOAD
3666 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3667 * added <http://sbcl.sourceforge.com> to the banner information printed by