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 * new feature: the filesystem location of SBCL's core file is
8 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
9 * minor incompatible change: the contrib modules SB-POSIX and
10 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
11 this change is to make it easier to distribute
12 location-independent binaries.
13 * bug fix: as implied by AMOP, standardized classes no longer have
14 slots named by external symbols of public packages. (reported by
16 * optimization: calling functions via constant symbols -- as in
17 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
18 via the function object as in (FUNCALL #'FOO).
19 * optimization: on x86 and x86-64, the calling convention for the
20 case of dealing with an unknown number of values has been altered
21 to be friendlier to the prediction heuristics implemented,
22 particularly with reference to CALL and RET pairing. (thanks to
24 * optimization: on x86, the code for access to thread-local symbol
25 values and binding/unbinding of thread-local symbols is smaller.
26 (thanks to Alastair Bridgewater)
27 * enhancement: CONSTANTP is now able to determine constantness of
28 more complex forms, including calls to constant-foldable standardized
29 functions and some special forms beyond QUOTE.
30 * fixed bug: occasional GC crashes on Solaris/x86
31 * optimization: x86-64 supports stack allocation of results of simple
32 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
33 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
34 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
35 (thanks to James Knight)
36 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
37 by the standard (thanks to Stephen Compall)
39 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
40 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
41 be used for bundling the runtime and the core file into one
42 executable binary. This feature is not currently supported on all SBCL
43 platforms. (thanks to James Bielman and NIIMI Satoshi)
44 * new feature: a generational or ephemeral garbage collector is now
45 the default on the PowerPC platform (both Linux and Darwin). The
46 old Cheney (stop and copy) collector is a build-time option.
47 (thanks to Cyrus Harmon, after Raymond Toy)
48 * minor incompatible change: the method by which SBCL finds its
49 contributed modules has changed; it no longer relies on symbolic
50 links from an $SBCL_HOME/systems directory, but searches directly
51 in the subdirectories of $SBCL_HOME.
52 * enhancement: the dynamic heap size on the Linux/PPC platform is
53 markedly larger, even using the older Cheney garbage collector.
54 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
55 certain complicated string types. (reported by Gary King)
56 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
57 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
58 * fixed bug: DOCUMENTATION on structures no longer signals an
59 error if no structure type of the right name exists. (reported by
61 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
62 applicable non-standard primary or :AROUND method on
63 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
64 no longer cause unbound variable errors. (reported by Kevin Reid)
65 * fixed bug: metacircle resolution in cases where methods have slots
66 added before the slots from STANDARD-METHOD. (reported by Jean
68 * fixed bug: the Power PC assembly code for calling into lisp
69 sometimes computed its return address wrongly (depending on the
70 15th bit of the address, fixed at link time). This used to
71 manifest itself in a segmentation violation while building PCL.
72 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
73 many others over the years)
74 * fixed bug: ROOM no longer reports silly numbers for stack usage if
75 the stack pointer (treated as a signed integer) is negative.
76 (thanks to Peter van Eynde)
78 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
79 * new platform: experimental support for the Windows operating
80 system has been added. (thanks to Alastair Bridgewater)
81 * fixed several bugs in and robustified the PPC FFI (including
82 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
83 * bug fix: allow non-simple string symbol names (reported by Paul
85 * bug fix: interrupt handling on NetBSD (thanks to Richard M
87 * bug fix: saving a core corrupted callbacks on x86/x86-64
88 * bug fix: closed a loophole in metacircularity detection and
89 grounding in the PCL implementation of CLOS.
90 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
91 (reported by tomppa on #lisp)
92 * bug fix: clear the direction flag on Lisp -> C transitions, as
93 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
94 (reported by Andrej Grozin and Hendrik Maryns)
95 * optimization: major improvements to GC efficiency on GENCGC platforms
96 * optimization: faster implementation of EQUAL
97 * optimization: emit more efficient opcodes for some common
98 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
99 * optimization: several other minor code-generation improvements on x86-64
100 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
102 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
103 * minor incompatible change: (SETF CLASS-NAME) and (SETF
104 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
105 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
106 by ANSI as a generic function, and for consistency (SETF
107 GENERIC-FUNCTION-NAME) is treated likewise.
108 * fixed bug #233.b: make constraint propagation notice when a variable
109 value is changed after it is referenced but before it is used
110 * fixed bug #296: no more arbitrary behaviour from filenames with
111 odd characters as --load arguments. (reported by Adam Warner)
112 * fixed bug #390: :CHARACTER-SET pathname components now work as
113 expected. (reported by Tim Daly Jr)
114 * fixed bug #391: complicated :TYPE intersections in slot
115 definitions no longer cause an error in PCL internals.
116 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
117 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
118 returning the number of octets which would be written to the
119 file-stream. (thanks to Robert J. Macomber)
120 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
121 arguments to the internal stream functions. (thanks to David
123 * bug fix: stack exhaustion is now handled on the x86/SunOS
124 platform. (thanks to Daisuke Homma)
125 * bug fix: more accurate ROOM results on GENCGC platforms
126 * optimization: improved type inference for arithmetic-for
127 index variables in LOOP
128 * optimization: faster floating-point SQRT on x86-64
129 * fixed some bugs revealed by Paul Dietz' test suite:
130 ** DOCUMENTATION returns NIL instead of "" for method combinations
131 that don't have a docstring
133 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
134 * minor incompatible change: (SETF CLASS-NAME) and (SETF
135 GENERIC-FUNCTION-NAME) are no longer generic functions, and
136 therefore are not customizeable by user code (as seems to be at
137 least permitted and maybe required by AMOP). As a consolation,
138 however, the SBCL implementation of these functions now calls
139 REINITIALIZE-INSTANCE as specified by AMOP.
140 * bug fix: slot-definition documentation is propagated as per ANSI
141 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
142 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
143 argument, as do eslotd initialization methods. (from Pascal
144 Costanza's "Closer" project)
145 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
146 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
148 * bug fix: it is now possible to have more than one subclass of
149 STANDARD-GENERIC-FUNCTION without causing stack overflow.
150 (reported by Bruno Haible, Pascal Costanza and others)
151 * bug fix: the dependent update protocol now works for generic
152 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
154 * bug fix: condition-class instances corresponding to
155 DEFINE-CONDITION forms are now created eagerly. (reported by
156 Kalle Olavi Niemitalo on comp.lang.lisp)
157 * bug fix: floating point printing is more accurate in some
158 circumstances. (thanks to Simon Alexander)
159 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
160 merged with *DEFAULT-PATHNAME-DEFAULTS*.
161 * bug fix: callbacks on OS X now preserve stack-alignment.
162 (thanks to Cyrus Harmon)
163 * enhancement: the x86-64 disassembler is much better at
164 disassembling SSE instructions. (thanks to Lutz Euler)
165 * enhancement: CHECK-TYPE on a variable now allows type-inference
166 (thanks to Vincent Arkesteijn)
167 * optimization: improved performance of EUC-JP external format.
168 (thanks to NIIMI Satoshi)
169 * optimization: performance improvements to IO on file streams of
170 :ELEMENT-TYPE CHARACTER
171 * optimization: much faster memory allocation on x86-64
172 * optimization: faster garbage collections (latency and throughput)
174 * optimization: faster bignum subtraction and division on x86 and x86-64
175 (thanks to Lutz Euler)
176 * optimization: more accurate type-inference for loops that use a
177 floating point index variable or a negative step.
179 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
180 * bug fix: add a workaround to SBCL looping infinitely at startup on
181 Linux kernels with apparently buggy implementations of personality().
182 (thanks to Svein Ove Aas)
183 * bug fix: Unicode symbols are correctly printed in LDB backtraces
184 (thanks to David Lichteblau)
185 * bug fix: local bindings shadow global symbol macros for
186 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
187 * optimization: non-open coded uses of numeric comparison operators
188 (e.g. >) no longer cons when called with more than one parameter
189 on platforms supporting dynamic-extent allocation.
190 * enhancement: saving cores with foreign code loaded is now supported
191 on MIPS/Linux in addition to the previously supported platforms.
192 * bug fix: FIND-RESTART now tests for activity, not applicability when given
193 a restart object as identifier. (reported by Helmut Eller for CMUCL)
194 * bug fix: division by zero in sb-sprof when no samples were collected
195 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
196 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
197 (thanks to Alexander Kjeldaas)
198 * new feature: ignoring the return values of destructive functions
199 like NREVERSE signals a compile-time style-warning.
200 (thanks to Kevin Reid)
201 * new feature: the alignment of alien structure fields can be
202 explicitly specified. (thanks to Cyrus Harmon)
203 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
204 and dump core on SIGQUIT
206 ** incompatible change: threads do not inherit values of specials
207 from their parents (see manual)
208 ** bug fix: threads stacks belonging to dead threads are freed by the
209 next exiting thread, no need to gc to collect thread stacks anymore
210 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
211 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
212 ** bug fix: dynamic variable and thread start related gc lossage
213 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
215 * fixed some bugs revealed by Paul Dietz' test suite:
216 ** SUBTYPEP is slightly more accurate on heinously complicated
217 CONS types where some of the members have uncertain (in the
218 NIL, NIL sense) type relationships to each other.
220 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
221 no longer purified unless :PURIFY T is explicitly specified.
222 ** Non-purified cores are significantly smaller than before
224 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
225 * new feature: timers based on Zach Beane's excellent timer package
226 * added support for the following external formats: koi8-u,
227 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
228 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
229 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
230 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
231 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
232 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
233 * incompatible change: a threaded SBCL will no longer revert to
234 non-threaded mode on non-NPTL systems, but refuse to start entirely.
235 * bug fix: interrupts are disabled until startup is complete; no
236 more sigsegvs when receiving a signal to soon
237 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
239 * bug fix: add a workaround for the memory randomization features in
240 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
241 workaround will only be in effect on systems with the proc filesystem
243 * bug fix: printing objects of type HASH-TABLE signals a
244 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
246 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
247 have been running for over 50 days. (reported by Gilbert Baumann)
248 * bug fix: the logic for getting names of functions gets less
249 confused when confronded with alternate-metaclass
250 funcallable-instances. (reported by Cyrus Harmon)
251 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
252 consistent, even on internal alternate-metaclass objects.
253 * bug fix: flush closure information collected by physical
254 environment analysis prepass before the main pass. (bug reported
256 * bug fix: compiler pack phase does not modify a hashtable, which is
257 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
258 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
259 may be expanded. (reported by Alexander <malishev> on c.l.l)
260 * bug fix: no more style-warnings when compiling code using LAST
261 under high-SPEED low-SPACE settings. (reported by David Wallin
263 * fixed some incompatibilities between SBCL's MOP and the MOP
265 ** the METAOBJECT class is now implemented;
266 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
267 STANDARD-OBJECT, as required;
268 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
269 now compatible (as required by VALIDATE-SUPERCLASS); there
270 remains a consistency requirement over the presence or absence
271 of the FUNCTION class in the superclass list of finalized
272 classes; see the manual for more details;
273 ** the :AROUND method for COMPUTE-SLOTS on
274 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
275 requested slot ordering.
277 ** bug fix: parent thread now can be gc'ed even with a live
279 ** bug fix: binding a special with PROGV to no value is not
280 the same as that symbol not having been bound (thanks to
282 ** bug fix: binding specials is thread safe (thanks to
284 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
285 and SB-SPROF do not die with 'no handler for signal XX in
286 interrupt_handle_now(..)' anymore
287 ** bug fix: WITH-TIMEOUT works with multiple running threads
288 * fixed some bugs revealed by Paul Dietz' test suite:
289 ** ENSURE-GENERIC-FUNCTION should take a method class object for
290 the :method-class keyword argument.
292 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
293 * new port: the Solaris operating system on x86 processors is now
294 mostly supported, though some rough edges in the environment
295 remain. (thanks to Daisuke Homma)
296 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
297 space for its heap. (thanks to Thiemo Seufer)
298 * optimization: SBCL on MIPS platforms now supports dynamic-extent
299 closures. (thanks to Thiemo Seufer)
300 * minor incompatible change: eof selects abort in the debugger.
301 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
302 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
304 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
305 deprecated, and will go away in a future revision of SBCL.
306 * minor incompatible change: GC-ON and GC-OFF are no longer
307 implemented with a counter, it does not matter how many times gc
308 is switched on or off
309 * bug fix: discriminating functions for generic function classes
310 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
311 make invalid assumptions about method precedence order. (reported
313 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
314 platforms. (thanks to Thiemo Seufer)
315 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
316 * bug fix: correct pathname printing with printer escaping is on.
317 (thanks to Kevin Reid)
318 * bug fix: complex VOP definitions in "user-space" no longer trigger
319 package locks. (reported by Zach Beane)
320 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
321 for funcallable-instance functions, and is no different from
323 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
324 characters in the current readtable. (reported by Nicholas Neuss)
325 * bug fix: SUBTYPEP on various CONS types returns more a more
326 accurate acknowledgment of its certainty.
327 * optimizations: REMOVE-DUPLICATES now runs in linear time on
328 lists in some cases. This partially fixes bug 384.
329 * flush all standard streams before prompting in the REPL and the
331 * bug fix: signal handling and triggering gc do not conflict
332 directly anymore, in particular a high frequency sb-sprof does
333 not prevent gc from running
334 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
335 approximation for timezone and DST information between the
336 universal time epoch and the smallest negative 32-bit time_t.
337 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
338 year 1899 so long as the encoded time is non-negative.
339 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
340 the correct answer for non-constant rotate amounts when performing
341 an inline 32-bit rotation.
343 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
344 there is only one thread in the session
345 ** bug fix: memory leak for streams created in one thread and
346 written to in another
347 ** bug fix: lockup when compiled with gcc4
348 ** bug fix: race that allows the gc to be triggered when gc is
350 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
351 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
353 ** bug fix: finished threads are gc'ed properly
354 * fixed some bugs revealed by Paul Dietz' test suite:
355 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
356 has arguments with a different set of applicable methods from
357 the orignal arguments.
358 ** The type error thrown by MAP now has a correct expected-type
360 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
361 name a compiled function.
362 ** Three MISC tests where a large bignum was improperly coerced to
363 a float that couldn't represent that bignum during type
364 derivation were fixed.
365 ** SUBTYPEP can now handle types involving the negation of a
366 list-form FUNCTION type.
367 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
368 as (COMPLEX (AND RATIO (NOT FIXNUM))).
369 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
371 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
372 * New feature: Experimental support for bivalent streams: streams
373 opened with :element-type :default now allow character and binary
374 (unsigned-byte 8) I/O
375 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
376 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
378 * bug fix: no more highly sporadic "couldn't check whether ~S is
379 readable" when reading a stream and an interrupt hits in the middle
380 of a select system call
381 * compiler better recognizes complex arrays (reported by Eduardo
383 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
384 for variables on SBCL built with linkage-tables. (reported by Luis
386 * various error reporting improvements.
387 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
388 (Thanks to Thiemo Seufer)
389 * bug fix: nested reader invokations work correctly
390 * bug fix: it is possible to have simultaneous references to foreign
391 code and foreign data with the same name.
393 ** added x86-64 support
394 ** incompatible change: the threading api now works with thread
395 objects instead of thread ids
396 ** bug fix: threads are protected from signals and interruption when
397 starting up or going down
398 ** bug fix: a race where an exiting thread could lose its stack to gc
399 ** bug fix: don't halt on infinite error in threads if possible
400 ** fixed numerous gc deadlocks introduced in the pthread merge
401 ** bug fix: fixed thread safety issues in read and print
402 ** bug fix: debugger doesn't hang on session lock if interrupted at
403 an inappropriate moment
404 ** bug fix: run-program is now thread safe(r)
405 ** bug fix: inner with-recursive-lock no longer releases the mutex
406 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
407 * fixed some bugs revealed by Paul Dietz' test suite:
408 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
409 EXPECTED-TYPE slots filled.
410 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
412 ** MAKE-CONDITION accepts classes as type-designators.
413 ** COMPILE may never return NIL.
414 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
415 range before calling Unix time functions
417 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
418 * numerous signal handling fixes to increase stability
419 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
420 * minor incompatible change: we now correctly canonize default
421 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
422 as per the MOP, rather than the historical (INITARG INITFUNCTION
423 INITFORM). (reported by Bruno Haible)
424 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
425 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
426 for more information.
427 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
428 pathname is a directory pathname.
429 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
430 * dynamic space size on PPC has been increased to 768Mb. (thanks to
432 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
433 :METACLASS argument in addition to a class name. (reported by
434 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
435 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
436 (reported by Bruno Haible; patch from Gerd Moellmann)
437 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
439 * bug fix: more cleanups to the floating point exception handling on
440 x86-64 (thanks to James Knight)
441 * bug fix: the compiler does not try to constant fold calls of
442 COERCE and BIT-* functions when they return freshly constructed
443 objects. (reported by Nikodemus Siivola and Paul Dietz)
444 * optimization: improved the MIPS versions of generic (in the
445 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
446 * optimization: direct conversion of (unsigned-byte 32) to floats on
447 the PowerPC platform.
448 * optimization: structure instances with raw slots now use less
449 memory, and probably show better memory locality. (thanks to
451 * optimization: DYNAMIC-EXTENT declarations for lists and closures
452 are treated as requests for stack allocation on the x86-64,
453 Alpha-32, PPC and SPARC platforms.
454 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
455 the wrong file descriptor; implementation of SOCKET-OPEN-P.
456 (thanks to Tony Martinez)
458 ** gcing a dead thread can no longer lead to lockups
459 ** threads block signals until they are set up properly
460 ** errno is no longer shared by threads
461 ** interrupt-thread restores the eflags register on x86
462 ** fixed some lockups due to gc/thread interaction
463 * fixed some bugs revealed by Paul Dietz' test suite:
464 ** invalid dotted lists no longer raise a read error when
466 ** signal an error if a symbol that names a declaration is used
467 as the name of a type, or vice versa
468 ** allow using the (declare (typespec var*)) abbreviation for
469 (declare (type typespec var*)) with all type specifiers
470 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
471 ** always return NIL from PPRINT-POP when OBJECT is NIL
472 ** don't signal errors when pretty-printing malformed LABELS,
473 FLET or MACROLET forms
474 ** declarations in a DOLIST body are also in scope for the
476 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
478 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
481 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
482 * fixed cross-compiler leakages that prevented building a 32-bit
483 target with a 64-bit host compiler.
484 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
485 opened with :IF-EXISTS :OVERWRITE.
486 * fixed bug 281: error for an invalid qualifier in a short-form method
487 combination method is not signalled until the faulty method is called.
488 * bug fix: iteration variable type inferrer failed to deal with open
489 intervals. (reported by Alan Shields)
490 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
491 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
492 or more runtime options were provided to the sbcl binary.
493 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
494 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
495 * optimization: there's now a fast-path for fixnum arguments in the
496 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
497 * optimization: the code generated on x86-64 is more compact thanks
498 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
499 * bug fix: floating-point exception handling now partly works on
500 x86-64. (Thanks to James Knight)
501 * improvement to the MIPS backend from Thiemo Seufer: C-style
502 64-bit long long arguments and return values to alien functions
504 * fixed some bugs revealed by Paul Dietz' test suite:
505 ** the type-error signalled from WARN has a filled-in DATUM slot.
506 ** the type-error required when a stream is not associated with
507 a file has the stream as its datum.
508 ** type-errors on single-floats on x86-64 no longer have
509 :INVALID-OBJECT as the datum
510 ** the type-errors signalled for invalid function names now have
511 a correct expected type
512 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
513 for structures defined with a :TYPE. Documentation strings for
514 typed structures are no longer immediately discarded
515 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
516 broadcast streams and synonym streams. FILE-LENGTH now also works
517 on broadcast streams.
519 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
520 * incompatible change: the --noprogrammer option, deprecated since
521 version 0.7.5, has been removed. Please use the equivalent
522 --disable-debugger option instead.
523 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
525 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
526 inoperational for a while now) has been completely removed.
527 * null lexical environments are now printed as #<NULL-LEXENV>,
528 significantly reducing the amount of clutter in typical backtraces.
529 * documentation on weak pointers, finalization, and after GC hooks
530 has been added to the manual.
531 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
532 as other specialized array types, is much faster. SUBSEQ and
533 COPY-SEQ on such arrays have also been sped up.
534 * optimization: EQL is now more efficient when at least other argument
535 is known to be of type (OR FIXNUM (NOT NUMBER)).
536 * fixed bug: the runtime is now less vulnerable to changes in the
537 size of the SBCL object on OS X, and virtual memory is reserved for
538 all spaces that need to be at a fixed address.
539 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
540 * fixed bug: finalizers and after GC hooks that cause consing are now
542 * fixed bug: compiler error messages and summaries are now printed to
543 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
544 * fixed inference of the upper bound of an iteration variable.
545 (reported by Rajat Datta).
546 * fixed bug 211e: calling local functions with duplicated constant
547 keyword argument no longer causes a bogus style warning about an
549 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
550 ftype declaration to be disregarded. (reported by Dave Roberts)
551 * fixed bug 373: caused by erronous compilation of references to alien
552 variables in the runtime on ppc/darwin.
553 * fixed bug 376: CONJUGATE type deriver.
554 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
555 (reported by Baughn on #lisp)
556 * a cross-compiler bug on non-x86 platforms has been identified and
557 fixed. (thanks to Bruno Haible)
558 * improvements to the MIPS runtime code for increased stability.
559 (thanks to Thiemo Seufer)
560 * increased the maximimum compact environment size to allow
561 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
562 * partial workaround for bug 135: don't name the function we're
563 calling for hairy cases of EVAL, so as not to accumulate one environment
564 entry per call to EVAL. (reported by Kevin Reid)
565 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
566 * optimization: type testing for non-vector arrays should be faster.
567 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
569 * bug fix: setting 31st element of a bit vector to zero did not work
571 * bug fix: redefining a class definition which failed due to a
572 previous accessor / function clash now works (but see BUGS entry
573 #380 for more problems in this area). (thanks to Zach Beane)
574 * the long form of DEFINE-METHOD-COMBINATION disables method group
575 checking when given a single method group with pattern *. (thanks
577 * on x86 compiler supports stack allocation of results of simple
578 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
579 * support for building and running on Mac OS X 10.4 "Tiger" has been added
580 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
581 * fixed some bugs related to Unicode integration:
582 ** the restarts for recovering from input and output encoding
583 errors only appear when there is in fact such an error to
585 * fixed some bugs revealed by Paul Dietz' test suite:
586 ** MISC.549 and similar: late transformation of unsafe type
587 assertions into derived types caused unexpected code
589 ** SCALE-FLOAT type deriver is less wrong.
590 ** type derivers for EXP, LOG and similar functions compute result
591 types for complex arguments better.
592 ** (MISC.563) CONJUGATE type deriver works for very restricted
594 ** out-of-line type testers for character strings are available.
595 ** EQUAL compiler transform understands specialness of objects
597 ** accessing double-floats stored on the stack now works on x86-64.
598 ** debugger internals could sometimes create invalid lispobjs,
599 resulting in GC crashes.
600 ** MISC.548: type check weakening can convert required type into
602 ** initialization forms for bindings are not in scope of free special
605 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
606 * incompatible change: thread support for non-NPTL systems has
607 been removed - locking is buggy and unreliable. A threaded
608 SBCL build will now warn at startup and refuse to create
609 new threads, unless futex support is detected in the kernel
610 * incompatible change: the top level REPL now has only an ABORT
611 restart associated with it, not TOPLEVEL and ABORT as it used to.
612 TOP and TOPLEVEL are now available as debugger commands for
613 returning to the top level.
614 * incompatible change: forms evaluated in the REPL now use the
615 global optimization policy.
616 * incompatible change: user- and system-initialization files are
617 no longer processed with LOAD, but by READ and EVAL; hence the
618 global optimization policy, startup package, readtable, etc,
620 * The .fasl file format number has been incremented because of
621 various incompatible changes.
622 * internal entry point details and argument counts no longer appear
623 in backtraces unless explicitly requested by setting
624 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
625 * built-in and standard functions no longer have names like "top
626 level local call to FOO".
627 * fixed bug 32: functions defined in non-null lexical environments
628 now have more legible printed representation
629 * fixed bug 33: functions defined in non-null lexical environemnts
630 are now more amenable to inspection by INSPECT.
631 * workaround for bug 354: XEPs no longer appear in backtraces unless
632 explicitly requested.
633 * fixed bug: receiving the signal which results from attempting to
634 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
635 SIGBUS on other *BSDs) on architectures where the C stack is also
636 the Lisp stack (x86 and x86-64) from foreign code no longer leads
637 to debugger confusion or wild instability. (reported by Cheuksan
639 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
640 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
641 (reported by Lutz Euler)
642 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
643 compiled code. (reported by Paul Dietz)
644 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
645 specialized parameter to a method to something that is not TYPEP
646 the specializer is now possible.
647 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
648 face of package deletion.
649 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
650 pathname. (reported by David Tolpin on c.l.l)
651 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
652 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
653 STRING-STREAM and FILE-STREAM, respectively.
654 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
655 than previously; in addition, its test suite is now run on build.
656 (thanks to Luke Gorrie)
657 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
658 * optimization: added a immediate representation for single-floats
660 * fixed some bugs revealed by Paul Dietz' test suite:
661 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
662 ** attempting to create a package with a colliding nickname causes
663 correctable errors to be signalled.
664 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
665 ** several x86-64 backend bugs related to sign-extension of immediate
668 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
669 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
670 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
672 * fixed loading of multiply forward-referenced layouts.
673 (thanks to Cheuksan Wang)
674 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
675 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
676 (see COMPILE-FILE documentation for details of :PRINT option)
677 * fixed bugs 19 and 317: fixed-format floating point printing is
678 more accurate. This also fixes a bug reported by Adam Warner
679 related to the ~@F format directive.
680 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
682 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
683 dispatch table if the from-char is a dispatch macro character.
684 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
685 by the wordsize no longer ignore the last word. (reported by Lutz
687 * fixed bug in type checking of dynamic-extent variables. (reported
689 * optimization: sequence traversal functions use their freedom to
690 coerce function designators to functions.
691 * optimization: code with many calls to CLOS methods specialized on
692 CLOS classes has had redundant type checks removed.
693 * optimization: added declarations to speed up operations that access
694 the internal character database (for example STRING-UPCASE)
695 * optimization: comparison operations between floats and sufficiently small
696 fixnums no longer create extra rationals
697 * fixed some bugs related to Unicode integration:
698 ** portions of multibyte characters at the end of buffers for
699 character-based file input are correctly transferred to the
700 start of the buffer at the next read.
701 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
702 passing it through to OPEN.
703 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
704 argument given to internal calls to OPEN.
705 ** the built-in comment readers (introduced by character sequences
706 ";" and "#|") are more forgiving to encoding errors; they will
707 STYLE-WARN and then attempt to resync the stream at a character
708 boundary. (thanks to Teemu Kalvas)
709 * fixed some bugs in the x86-64 port:
710 ** Negative short int return values from c-calls are sign-extended
712 ** The stack is aligned to 16-bytes for c-calls, as required by
713 the ABI. (thanks to Cheuksan Wang)
714 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
715 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
716 for enums. (thanks to Vincent Arkesteijn)
717 ** Multiple small optimizations and bugfixes for floating point
719 * fixed some bugs revealed by Paul Dietz' test suite:
720 ** Space, Tab, Linefeed, Return and Page have the invalid
721 secondary constituent character trait.
722 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
724 ** WITH-INPUT-FROM-STRING should only update the index place on
726 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
728 ** Bit-array operations (BIT-AND and similar) worked incorrectly
729 with one-dimensional arrays with fill pointers.
730 ** TYPE-OF failed on a complex with an integer realpart and a
732 ** compiler failure during type inference for the code of form
733 (IF (EQL X (THE ...)) ...) (MISC.535).
735 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
736 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
737 platform. The port supports 61-bit fixnums, large memory spaces
738 and reloading shared object files.
739 * enhancement: saving cores with foreign code loaded is now
740 supported on ppc/Darwin in addition to the previously supported
742 * enhancement: the statistical profiler now walks deeper into the
743 call stack for more meaningful call-graphs and accrued time
744 reports (x86/x86-64 only). It also now reports time spent in
746 * enhancement: it is now possible to trace most individual methods
747 of a generic function in addition to tracing the generic function
749 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
751 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
752 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
753 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
754 produces an error. (thanks to Vincent Arkesteijn)
755 * bug fix: NAMESTRING on pathnames with :WILD components in their
756 directories works correctly. (thanks to Artem V. Andreev)
757 * fixed bug 125: compiler preserves identity of closures. (reported
759 * bug fixed: functions with &REST arguments sometimes failed with
760 "Undefined function" when compiled with (DEBUG 3). (reported by
762 * bug fix: overflow during compiling of setting element of a bit
763 vector with constant index and value. (reported by Timmy Douglas)
764 * build fix: fixed the dependence on *LOAD-PATHNAME* and
765 *COMPILE-FILE-PATHNAME* being absolute pathnames.
766 * on x86 compiler partially supports stack allocation of dynamic-extent
768 * GO and RETURN-FROM do not check the extent of their exit points
769 when compiled with SAFETY 0.
770 * fixed some bugs related to Unicode integration:
771 ** encoding and decoding errors are now much more robustly
772 handled; it should now be possible to recover even from invalid
773 input or output to the terminal. (thanks to Teemu Kalvas)
774 ** provided a first cut at implementing STRING-TO-OCTETS and
775 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
776 ** altered the SB-MD5 contributed module slightly, changing the
777 interface just enough for it to be supportable for builds where
778 lisp characters are not eight bits.
779 * fixed some bugs revealed by Paul Dietz' test suite:
780 ** the FORMATTER-generated functions for ~V[ conditionals require
781 the correct number of arguments.
782 ** READ-FROM-STRING returns the mandated second value when applied
783 to displaced strings.
784 ** the #\Rubout and #\Backspace characters are treated as invalid
785 constituent characters by the tokenizer.
787 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
788 * new feature: reloading changed shared object files with
789 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
790 * new feature: references to foreign variables and functions
791 can now be compiled and loaded before the corresponding shared
792 object file is loaded, as long as the foreign definitions are
793 available at runtime.
794 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
795 supported. (thanks to Dan Debertin)
796 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
797 just keywords. (thanks to Vincent Arkesteijn)
798 * fixed bug #331: structure-class instances corresponding to
799 DEFSTRUCT forms are now created eagerly.
800 * fixed bug #345: backtraces from calls to undefined functions work
801 on x86 as well. Related bug #61 is now also partially fixed on x86
802 (backtraces from throws to unknown catch tags.)
803 * bug fix: lambda-list parsing is now stricter vrt. order and number
804 of lambda-list keywords.
805 * bug fix: as specified by AMOP, an error is signalled if a
806 class-option appears multiple times in a DEFCLASS form. (reported
808 * bug fix: RANDOM can be compiled when the compiler derives the type
809 of its numeric argument as a disjoint set of small integers.
810 (reported by Paul Dietz)
811 * bug fix: starting a core saved with shared objects loaded when
812 those objects are not available no longer causes threaded SBCL to
813 hang. (reported by Sean Ross)
814 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
816 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
817 argument, which must always be NIL. (reported by Kalle Niemitalo)
818 * bug fix: printing 1.0d+23 no longer results in an error.
819 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
820 * bug fix: structure slot setters preserve evaluation order. (thanks
822 * fixed some bugs related to Unicode integration:
823 ** RUN-PROGRAM can allow its child to take input from a Lisp
824 stream. (reported by Stefan Scholl)
825 ** ASDF-INSTALL successfully downloads .tar.gz files in all
826 locales. (reported by Ken Causey)
827 * fixed some bugs revealed by Paul Dietz' test suite:
828 ** INCF, DECF and REMF evaluate their place form as specified in
830 ** FORMATTER expands ~{ ~} iteration directives with V or #
831 parameters correctly.
832 ** FORMATTER deals with the ~@[ ~] conditional directive where the
833 consequent uses no arguments correctly.
834 ** the system has a partial understanding of the (COMPLEX RATIO)
836 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
839 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
840 * new feature: a build-time option (controlled by the :SB-UNICODE
841 keyword feature, enabled by default) for building the system with
842 support for the entire 21-bit character space defined by the
844 * new feature: the system now has rudimentary external-format
845 support; the primary user-visible change at this time is that
846 characters with the high bit set (such as lower-case-e-acute) will
847 print correctly to a terminal in a UTF-8 environment.
848 * minor incompatible change: BASE-CHAR no longer names a class;
849 however, CHARACTER continues to do so, as required by ANSI.
850 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
851 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
852 has been moved to the SB-EXT package (temporarily re-exported from
854 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
855 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
857 * fixed bug #351: better error-handlind and reporting for malformed
859 * fixed bug #350: bignum-printing is now more memory-efficient,
860 allowing printing of very large bignums, eg. (expt 2 10000000).
861 (reported by Bruno Haible)
862 * fixed bug #302: better primitive-type selection for intersection
864 * fixed bug #308: non-graphic characters now all have names, as
865 required. (reported by Bruno Haible)
866 * bug fix: redefining a class with different superclasses now correctly
867 removes it from the direct-subclasses of its previous superclasses.
868 (reported by David Morse)
869 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
870 new value now works. (reported by Bruno Haible)
871 * bug fix: correct canonicalization of multiple non-standard slot
872 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
873 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
874 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
875 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
876 now exists, an signals an error.
877 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
878 during macroexpansion for non-iteration variables as well. (reported
879 by Bruno Haible for CMUCL)
880 * bug fix: Cyclic structures and unprintable objects in compiler
881 messages no longer cause errors. (reported by Bruno Haible)
882 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
883 and READ-FROM-STRING all now return a primary value of NIL if
884 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
885 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
886 (reported by Bruno Haible for CMUCL)
887 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
888 do not designate a condition. (reported by Bruno Haible for
890 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
891 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
892 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
893 permitting resolution in favour of any of the conflicting symbols.
894 (reported by Bruno Haible for CMUCL)
895 * bug fix: EQUAL compiler optimizations is less aggressive on
896 strings which can potentially compare true despite having distinct
897 specialized array element types.
898 * bug fix: unit enumerations can be defined without dividing by
899 zero. (thanks to Vincent Arkesteijn)
900 * FORMAT compile-time argument count checking has been enhanced.
901 (report from Bruno Haible for CMUCL)
902 * a partial workaround for the bug 262: the compiler does not try to
903 inline-expand a local function doing RETURN-FROM from a deleted
904 BLOCK. (thanks to Peter Denno for the bug report and to David
905 Wragg for the simple test case)
906 * fixed some bugs revealed by Paul Dietz' test suite:
907 ** RENAME-PACKAGE allows all package designators as new package
909 ** constraint propagation and lambda variable substitution are
910 more cautious in dealing with partially deleted code.
911 ** compiler sometimes lost reoptimization passes.
912 ** CERROR, given a condition as condition designator, passes its
913 remaining arguments to the continue format control without
915 ** Case-altering FORMAT directives work correctly on non-ASCII
917 ** The REAL type specifier handles bounds outside the floating
918 point ranges without signalling FLOATING-POINT-OVERFLOW.
919 ** Functions with IR1-transformations can create intercomponent
920 references to global functions.
921 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
923 ** FORMAT ~R treats a nil value for its first parameter correctly.
925 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
926 * enhancement: saving cores with foreign code loaded is now
927 supported on x86/NetBSD and sparc/Linux in addition to the previously
929 * bug fix: on some platforms repeated installations caused multiple
930 copies of HTML documentation to be installed -- should not happen
931 any more. (reported by Stefan Scholl)
932 * bug fix: parsing self-recursive alien record types multiple times
933 no longer causes infinite recursion. (reported by Thomas F. Burdick,
934 original patch by Helmut Eller for CMUCL)
935 * bug fix: stack-exhaustion detection works now on NetBSD as well.
936 (thanks to Richard Kreuter)
937 * bug fix: defining classes whose accessors are methods on existing
938 generic functions in other (locked) packages no longer signals
939 bogus package lock violations. (reported by François-René Rideau)
940 * bug fix: special variables as DEFMETHOD parameters no longer have
941 associated bogus type declarations. (reported by David Wragg and
943 * bug fix: read-write consistency on streams of element-type
944 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
945 * bug fix: redefiniton of the only method of a generic function with
946 no DEFGENERIC no longer emits a full WARNING. In addition,
947 redefinition of generic functions with no DEFGENERIC to an
948 incompatible lambda list now signals an error. (thanks to Zach
950 * bug fix: DEFGENERIC now works even when there's a function of the
951 same name in an enclosing lexical environment. (thanks to Zach
953 * fixed compiler failure, caused by instrumenting code during
954 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
955 * optimization: added loop analysis and improved register allocation
956 * fixed some bugs revealed by Paul Dietz' test suite:
957 ** POSITION on displaced vectors with non-zero displacement
958 returns the right answer.
959 ** (SIMPLE-STRING) is a valid type specifier for sequence
961 ** *PRINT-LEVEL* handling for slotless structures is pedantically
963 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
964 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
966 ** The justification version of the FORMAT ~< directive treats
967 non-zero minpad parameter correctly.
969 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
970 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
971 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
972 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
973 the supported interface.
974 * new feature: Single-stepping of code compiled with DEBUG 2 or
975 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
976 * new feature: saving cores with foreign code loaded is now
977 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
978 Timothy Moore's work for CMUCL)
979 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
980 parameters to * instead of NIL if no initform is supplied.
981 (reported by Johan Bockgaard)
982 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
983 correctly when FUNCALL appears as the car of the form. Note:
984 despite this FUNCALL forms are not currently subject to
985 compiler-macro expansion. (port of Raymond Toy's fix for the
986 same from CMUCL, reported by Johan Bockgaard)
987 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
988 (thanks for Teemu Kalvas)
989 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
990 case instead of terminating the iteration (thanks for Julian
991 Squires, Sean Champ and Raymond Toy)
992 * bug fix: incorrect expansion of defgeneric that caused a style
993 warning. (thanks for Zach Beane)
994 * bug fix: slot accessor effective method computation works properly
995 for classes with multiple non-standard applicable methods on
996 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
997 * on x86 compiler supports stack allocation of results of LIST and
998 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
999 CMUCL implementation by Gerd Moellmann)
1000 * fixed some bugs revealed by Paul Dietz' test suite:
1001 ** FORMAT strings with both the ~<~:;~> form of the justification
1002 directive and pretty-printing directives cause an error.
1004 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1005 * incompatible change: the internal functions
1006 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1007 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1008 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1009 instead of the old functions.
1010 * new feature: on platforms where "dladdr" is available foreign
1011 function names now appear in backtraces. (based on Helmut Eller's
1013 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1014 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1016 * documentation: ASDF manual is now distributed with SBCL.
1017 * bug fix: SBCL can now load its contributed modules with REQUIRE
1018 even if the system-provided entries have been removed from
1019 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1021 * bug fix: backtraces involving undefined functions or assembly
1022 routines are more informative. (thanks to Brian Downing)
1023 * bug fix: mutually referent alien structures now work correctly.
1024 (reported by Rick Taube)
1025 * bug fix: structures defined by WITH-ALIEN can be referred to
1026 within other definitions in the same WITH-ALIEN.
1027 * bug fix: division operators (MOD, TRUNCATE and the like) with
1028 constant zero divisors and integer dividends no longer generate
1030 * bug fix: provide default methods for INPUT-STREAM-P and
1031 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1032 * bug fix: improve the detection and resolution of MOP metacycles
1033 (where effective method computation depends on the generic function
1034 having its effective method computed). (reported by Bruno Haible)
1035 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1036 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1037 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1038 * bug fix: pathnames with relative directory components can be
1039 represented relative to default pathnames.
1040 * optimization: in taking the GCD of bignums, reduce the two bignums
1041 to approximately the same size (using Euclid's algorithm) before
1042 applying the more sophisticated binary GCD. (thanks to Juho
1044 * optimization: COUNT on bitvectors now operates word-at-a-time.
1045 * optimization: ASH with a positive, but not necessarily constant,
1046 (leftwards) shift, when in a modular context, is compiled to a
1048 * fixed some bugs revealed by Paul Dietz' test suite:
1049 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1050 if the corresponding argument is NIL.
1051 ** FORMAT directives accepting a minpad parameter treat negative
1053 ** literal commas as character parameters to FORMAT directives are
1055 ** literal spaces directly after ~<Newline> directives within a
1056 format-logical-block (~:< ~@:>) do not induce :FILL-style
1057 conditional newlines.
1058 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1059 as specified: it no longer includes conditional newlines).
1060 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1062 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1063 :PER-LINE-PREFIX argument does not evaluate to a string.
1065 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1066 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1067 the manual for details; for now, package locks can be disabled by
1068 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1069 if no major problems are found then it is likely that they will be
1070 compiled in unconditionally.
1071 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1072 unsupported operators on all platforms. To load a shared library
1073 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1074 object file, link it into a shared library outside of SBCL and
1075 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1077 * bug workaround: ROOM T and the GC were not working together
1078 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1079 reporting facility aren't preserved. That reporting has been
1080 disabled, so now until and unless someone figures out how to make it
1081 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1082 * minor incompatible change: as threatened around sbcl-0.8.0, the
1083 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1084 The symbols are also exported from SB-PCL for backwards
1085 compatibility, but more so than before SB-PCL should be treated as
1086 an implementation-internal package.
1087 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1089 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1090 It is known as of this release not to work on the Alpha, however.
1091 * fixed bug #167: errors signalled due to illegal syntax in method
1092 bodies are now more legible.
1093 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1094 designators and can hence be used with TYPEP.
1095 * fixed bug #333: CHECK-TYPE now ensures that the type error
1096 signalled, if any, has the right object to be accessed by
1097 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1098 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1099 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1101 * fixed bug #334: programmatic addition of slots using specialized
1102 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1103 and :ALLOCATION :CLASS slots.
1104 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1105 float exponents. (rereported by Peter Seibel)
1106 * fixed a bug: #\Space (and other whitespace characters) are no
1107 longer considered to be macro characters in standard syntax by
1108 GET-MACRO-CHARACTER.
1109 * fixed bug: initialization of condition class metaobjects no longer
1110 causes an instance of the condition to be created. (reported by
1112 * fixed bug: it is now possible to have slots such that
1113 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1114 description is neither :INSTANCE nor :CLASS.
1115 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1116 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1117 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1118 non-standard SLOT-DEFINITION-ALLOCATION values.
1119 * fixed bugs: various race conditions handling exiting threads.
1120 CL-PPCRE's thread torture test now passes.
1121 * fixed arguably-a-bug: GC time proportional to number of threads in
1122 system even when most of them are idle
1123 * optimization: improved performance of BIT and SBIT on bit-vectors.
1124 * .fasl file incompatibility: The fasl file version number has
1125 been incremented because of changes associated with package locks.
1127 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1128 * minor incompatible change: the system no longer provides
1129 optimization hints (conditions which are TYPEP
1130 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1131 mode; these hints are still emitted when higher SPEED optimization
1133 * new contrib module: a sampling profiler (profiling by statistical
1134 sampling, rather than by instrumenting functions) is available as
1135 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1136 * the behaviour of the standard function ED is now customizeable by
1137 third parties through a hook variable: see ED's documentation
1138 string for information on the protocol.
1139 * the compiler no longer emits efficiency notes for (FUNCALL X)
1140 when the type of X is uncertain under default optimization
1142 * fixed bug 276: mutating a binding of a specialized parameter to a
1143 method to something that is not TYPEP the specializer is now
1145 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1146 exhibit the required behaviour on the x86 platform. (thanks to
1147 Peter van Eynde, Eric Marsden and Bruno Haible)
1148 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1149 even for difficult arguments. (reported by Peter Graves)
1150 * fixed bug 141a: the backquote printer now descends quoted
1152 * fixed another bug in backquote printing: no more destructive
1153 modification of the form's list structure. (reported by Brian
1155 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1156 move between its address being taken and the call to
1157 interrupt_thread, fixing a crashing race condition.
1158 * the SB-POSIX contrib implementation has been adjusted so that it
1159 no longer exhibits ridiculously poor performance when constructing
1160 instances corresponding to C structs.
1162 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1163 * minor incompatible change: the sb-grovel contrib now treats C
1164 structures as alien (in the sense of SB-ALIEN) objects rather than
1165 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1166 has implications for memory management of client code
1167 (sb-grovel-returned objects must now be manually managed) and for
1168 type safety (alien objects now have full types).
1169 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1170 used to control emission of compiler diagnostics, rather than the
1171 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1172 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1173 quality should be considered deprecated.
1174 * install.sh now installs the user manual as well
1175 * (not quite a new documentable feature, but worth considering in
1176 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1177 beginnings of a semantically meaningful condition hierarchy is
1178 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1180 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1181 designator as the defaults argument.
1182 * fixed bug: Displaced arrays whose displaced-to array has become
1183 too small now cause ARRAY-DIMENSION to signal an error, providing
1184 for safer bounds-checking. (reported by Bruno Haible)
1185 * fixed bug: DEFCLASS slot definitions with identical :READER and
1186 :WRITER names now signal a reasonable error. (reported by Thomas
1188 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1190 * fixed bug: Invalid slot specification errors now print correctly.
1191 (thanks to Zach Beane)
1192 * fixed bug 320: Shared to local slot value transfers in class
1193 redefinitions now happen corrently from superclasses as
1194 well. (reported by Bruno Haible)
1195 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1197 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1198 improper lists correctly. (reported by Bruno Haible)
1199 * fixed bug 313: source-transform for <fun-name> was erroneously
1200 applied to a call of a value of a variable with name <fun-name>.
1201 (reported by Antonio Menezes Leitao)
1202 * fixed bug 307: The obsolete instance protocol ensures that
1203 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1204 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1205 remove dead unknown-values globs from the middle of the stack.
1206 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1207 easier to resume long computations after using *BREAK-ON-SIGNALS*
1208 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1209 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1210 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1211 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1212 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1213 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1215 * fixed a (fairly theoretical) bug in string printing: if
1216 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1217 does not have array-element-type equal to the most general string
1219 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1220 function name. (thanks to Nikodemus Siivola)
1221 * fixed bug: compile-time format string checker failed on
1222 non-closed ~{. (reported by Thomas F Burdick)
1223 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1224 don't warn on legal loop code involving more than one aggregate
1225 boolean. (thanks to Nikodemus Siivola)
1226 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1227 in keyword position to MAKE-PACKAGE was not being reported as
1229 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1230 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1232 * fixed bugs 280 and 312: the checking for multiple definitions in a
1233 file is less likely to become confused by uses of inline
1235 * fixed bug: the #S reader macro performs the keyword coercion
1236 specified for slot names. (reported by Kalle Niemitalo)
1237 * fixed bug: lambda lists may contain symbols whose names start with
1238 & but are not lambda-list-keywords; their occurrence triggers a
1240 * fixed bug 321: define-method-combination argument lambda lists do
1241 not cause invalid code to be generated when &optional or &aux
1242 variables have default values. (reported by Bruno Haible)
1243 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1244 have CLOS classes; this appears also to have fixed reported
1245 instability in stack exhaustion detection.
1246 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1247 MUFFLE-WARNING when no associated restart is present is now
1249 * optimization: rearranged the expansion of various defining macros
1250 so that each expands into only one top-level form in a
1251 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1253 * optimization: used a previously little-used slot in symbols to
1254 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1256 * fixed some bugs revealed by Paul Dietz' test suite:
1257 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1259 ** Fixed an optimization bug related to inheritance of initforms
1260 from local to shared slots.
1261 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1262 ** CAST optimizer forgot to flush argument derived type.
1263 ** print/read consistency on floats is now orders of magnitude
1264 more likely. (thanks also to Bruno Haible for a similar report
1266 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1267 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1268 for objects of type REAL. Make it so.
1269 ** SXHASH is specified (infelicitously) to respect similarity,
1270 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1271 it so. (thanks to Markus Ziegler)
1272 ** on the Alpha, the compiler succeeds in compiling functions
1273 returning a known number of arguments greater than 63.
1274 ** fixed handling of invalid NIL arguments in keyword position
1276 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1278 ** provided a readably-printable representation for RANDOM-STATE
1280 ** ensured that pathnames, where they have a namestring, always
1281 print using #P"..." syntax.
1283 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1284 * Support for the forthcoming 2.0 version of the NetBSD kernel
1285 running on x86 hardware has been added. (thanks to Perry
1286 E. Metzger most immediately, and others for their past work)
1287 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1288 current release version; SBCL's OpenBSD support had been broken
1289 since about the time of OpenBSD's switch to ELF binary format.)
1290 * [placeholder for DX summary]
1291 ** user code with &REST lists declared dynamic-extent, under high
1292 speed or space and low safety and debug optimization policy.
1293 * The manual has been converted to Texinfo format and the debugger
1294 chapter from the cmucl manual has been added.
1295 * A facility has been added to extract documentation strings from
1296 sbcl and store them as Texinfo-formatted snippets for inclusion in
1297 the manual (via Texinfo's @include directive)
1298 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1299 the test case to Dave Roberts)
1300 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1301 pointers. (reported by Sean Ross)
1302 * bug fix: PROFILE output is printed nicely even for large numerical
1303 values. (thanks to Zach Beane)
1304 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1305 greater than 32 handle EOF correctly.
1306 * bug fix: on X86 an immediate argument of the IMUL instruction is
1307 correctly printed in disassembly. (thanks to Lutz Euler)
1308 * bug fix: class slots in redefined classes preserve their old
1309 values. (thanks to Bruno Haible and Nikodemus Siivola)
1310 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1311 succeeds. (reported by Marco Baringer)
1312 * fixed some bugs revealed by Paul Dietz' test suite:
1313 ** READ-SEQUENCE now works on ECHO-STREAMs.
1314 ** RATIONALIZE works more according to its specification. (thanks
1317 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1318 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1319 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1320 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1321 should matter to you only if you rebind the printer control
1322 variables and then find you want different bindings in the
1323 debugger than in the ordinary execution of your program.)
1324 * The runtime build system has been tweaked to support building
1325 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1326 assembler and linker. (thanks to Nikodemus Siivola)
1327 * Unbound, undefined, undeclared variables now trigger full
1328 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1329 more often programmer error than deliberate exploitation of undefined
1331 * optimization: the hash algorithm for strings has changed to one
1332 that is less vulnerable to spurious collisions. (thanks to Juho
1334 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1335 needless bounds checking. (thanks to Juho Snellman)
1336 * optimization: implemented multiplication as a modular
1337 (UNSIGNED-BYTE 32) operation on the PPC backend.
1338 * fixed some bugs revealed by Paul Dietz' test suite:
1339 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1341 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1342 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1343 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1345 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1347 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1349 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1350 their output stream on EOF from read.
1351 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1352 have been read to end-of-file.
1353 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1355 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1356 description of determination of which consecutive characters
1358 ** Printing the "Space" character with escaping on now yields "#\\ ",
1359 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1360 ** Reading floating-point numbers with *READ-BASE* set to a number
1361 less than 10 works correctly.
1362 ** Reading floating-point numbers with *READ-BASE* set to a number
1363 more than 10 works correctly.
1364 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1365 the readtable currently in effect.
1367 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1368 * minor incompatible change: parsing of namestrings on a physical
1369 (Unix) host has changed; numbers after the final #\. in a
1370 namestring are no longer interpreted as a version field. This is
1371 intented to be largely invisible to the user, except that the
1372 meaning of the namestring "*.*.*" has changed: it now refers to a
1373 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1374 should usually be replaced by
1375 (make-pathname :name :wild :type :wild :version :wild)
1376 with the added benefit that this is more likely to be portable.
1377 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1378 option to OPEN now signals an error if the file being opened
1379 exists; this may have an impact on existing code.
1380 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1381 signals received, on the PowerPC platforms (both Linux and
1382 Darwin). (thanks to Pierre Mai for pointing out the location of
1384 * several fixes on OS X: The system now builds and runs cleanly on
1385 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1386 (thanks to Brian Mastenbrook)
1387 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1388 second-resolution: integer multiples of 1/3600 between -24 and 24.
1389 (thanks to Vincent Arkesteijn)
1390 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1391 type when called with 1 argument; PEEK-CHAR checked type of
1392 PEEK-TYPE only after having read first character from a
1393 stream. (reported by Peter Graves)
1394 * bug fix: the garbage collector now has much better locality
1395 behaviour, and in particular no longer treats all memory as being
1396 exhausted when many small objects point to each other in a deeply
1398 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1399 recognized as being TYPEP their class.
1400 * bug fix: the PUSHNEW documentation string has been corrected.
1401 (thanks to Vincent Arkesteijn)
1402 * bug fix: defaulting of the value for the last of an atypically
1403 large number of multiple values being bound was not being
1404 performed correctly on the Alpha or PPC platforms
1405 * optimization: implemented multiplication as a modular
1406 (UNSIGNED-BYTE 32) operation on the x86 backend.
1407 * optimization: SEARCH on simple-base-strings can now be open-coded.
1408 (see also contrib/compiler-extras.lisp for inspiration for
1409 teaching the compiler about the Boyer-Moore algorithm).
1410 * value, returned by MAX (and MIN) called with several EQUALP, but
1411 not EQL, arguments now does not depend on compiler settings.
1412 * fixed some bugs revealed by Paul Dietz' test suite:
1413 ** in stack analysis liveness information is propagated from
1414 non-local entry points.
1415 ** pathwise CAST removing failed when the CAST node did not start
1417 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1418 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1420 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1421 host is already defined.
1422 ** RENAME-FILE works on streams instead of signalling an internal
1424 ** PEEK-CHAR uses the current readtable when determining whether
1425 or not a character is whitespace.
1426 ** MERGE-PATHNAMES handles the case when the pathname does not
1427 specify a name while the default-pathname specifies a version.
1428 ** Pathnames now stand a better chance of respecting print/read
1430 ** Attempting to use standardized file system operators with a
1431 pathname with invalid :DIRECTORY components signals a
1433 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1435 ** DIRECTORY on logical pathnames is more correct.
1436 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1437 signal a TYPE-ERROR if their argument is not a stream.
1438 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1439 designator argument does not designate a stream.
1440 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1441 examining the synonym.
1442 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1444 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1445 element-type larger than ([UN]SIGNED-BYTE 32).
1447 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1448 * When built with the :SB-FUTEX feature, threaded builds now take
1449 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1450 for faster/more reliable mutex and condition variable support.
1451 * Incompatible change (but one you probably shouldn't have been using
1452 anyway): the interface and code for arbitrating between multiple
1453 threads in the same user session has been redesigned.
1454 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1455 error when called without an explicit environment argument.
1456 (thanks to Nikodemus Siivola)
1457 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1458 confused about their position. (thanks to Adam Warner and Gerd
1460 * bug fix: Pretty printing backquoted forms with unquotations in the
1461 argument list position of various code constructs such as LAMBDA
1462 now works correctly. (reported by Paul Dietz)
1463 * bug fix: Pretty printing unquotations no longer loses all
1464 stream position information.
1465 * optimization: performance of string output streams is now less
1466 poor for multiple small sequence writes.
1467 * optimization: performance of CSUBTYPEP in the presence of complex
1468 expressions involving CONS and NOT many times has been improved.
1469 (reported by Paul Dietz)
1470 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1472 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1473 streams for the REPL, for the debugger, and so on. (thanks to
1475 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1476 * fixed some bugs revealed by Paul Dietz' test suite:
1477 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1478 is ignored if it is too small, rather than propagating through
1479 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1480 ** extremely complex negations of CONS types were not being
1481 sufficiently canonicalized, leading to inconsistencies in
1483 ** VALUES tranformer lost derived type.
1485 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1486 * fixed a bootstrapping bug: the build process no longer assumes
1487 that the various BOOLE-related constants have the same value in
1488 host and target lisps. (noted by Paul Dietz' test suite on an
1489 SBCL binary built from CLISP)
1490 * The system can now be dynamically linked on the MIPS platform,
1491 which enables dynamic loading of foreign code from Lisp. (thanks
1492 to Ralf Baechle for discussions on the MIPS ABI)
1493 * The system now records debugging information for its own source
1494 files in a filesystem-position-independent manner, relative to
1495 the "SYS" logical host.
1496 * fixed a compiler bug: MV-LET convertion did not check references
1497 to the "max args" entry point. (reported by Brian Downing)
1498 * tweaked disassembly notes to be less confident about proclaiming
1499 some instruction as an LRA. (thanks to Brian Downing)
1500 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1501 now each have their own history, command character, and other
1502 characteristics. (thanks to David Lichteblau)
1503 * fixed some bugs revealed by Paul Dietz' test suite:
1504 ** compiler failure in compiling LOGAND expressions including a
1506 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1507 shift greater than 32.
1508 ** FLUSH-DEST did not mark blocks for type check regeneration.
1509 ** HANDLER-CASE failed to accept declarations in handler clauses
1510 in some circumstances.
1512 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1513 * New code in contrib/sb-introspect (still probably not entirely
1514 stable yet) provides some support for smart Lisp development
1515 environments like SLIME.
1516 * The conditions signalled for errors occurring when loading .fasl
1517 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1518 in a way which should help ASDF recover gracefully.
1519 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1520 stuff has been cleaned up. If you code contrib/ stuff, this might
1521 affect you, and you can look at contrib/README, contrib/STANDARDS,
1522 and/or the 0.8.4.27 diff to check.
1523 * In full calls the compiler now does not generate checks for declared
1524 argument types for all arguments.
1525 * various threading fixes
1526 ** and some experimental patches which didn't make it into
1527 the main tree for this release, but which are shipped in
1528 contrib/experimental-thread.patch as a possible fix for some
1529 failures (deadlock, spinning...) in GC-intensive multithreaded
1531 * fixed PPC build problem (source code incompatibility of different
1532 library versions): added offsetof() hackery which attempts to divine
1533 where glibc maintainers put uc_mcontext today
1534 * fixed bug 282: compiler does not trust type assertions while passing
1535 arguments to a full call.
1536 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1537 &OPTIONAL VALUES type parameter.
1538 * fix bug 214: algorithm for noting rejected templates is now more
1539 similar to that of template seletion. (also reported by rydis on
1541 * fixed bug 141b: printing backquoted information readably and prettily
1542 inserts a space where necessary.
1543 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1544 (CHARACTER 10) are now reported as errors, rather than propagated
1545 as unknown types. (reported by piso on #lisp)
1546 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1547 with values NIL and :ERROR. (thanks to Milan Zamazal)
1548 * fixed bug 191c: CLOS now does proper keyword argument checking as
1549 described in CLHS 7.6.5 and 7.6.5.1.
1550 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1551 longer raise an error; further, using a list as a for-as-arithmetic
1552 counter now raises a meaningful error.
1553 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1554 now understood by sequence creation functions such as MAKE-SEQUENCE
1556 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1557 a STRING-INPUT-STREAM.
1558 * compiler enhancement: SIGNUM is now better able to derive the type
1560 * type declarations inside WITH-SLOTS are checked. (reported by
1562 * fixed some bugs revealed by Paul Dietz' test suite:
1563 ** incorrect optimization of TRUNCATE for a positive first
1564 argument and negative second.
1565 ** compiler failure in let-convertion during flushing dead code.
1566 ** compiler failure while deriving type of TRUNCATE on an
1567 interval, containing 0.
1568 ** ASH of a negative bignum by a negative bignum count now returns
1570 ** intersection of CONS types now canonicalizes properly, fixing
1571 inconsistencies in SUBTYPEP.
1573 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1574 * incompatible change: The --disable-debugger command line
1575 option now clobbers the debugger at a more fundamental
1576 level, by redefining #'INVOKE-DEBUGGER instead of by
1577 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1578 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1579 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1580 enter the debugger REPL and then suffer recursive errors
1581 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1582 while under the new system, BREAK in --disable-debugger mode
1583 terminates the system just as an unhandled error would.
1584 * fixed compiler performance when processing loops with a step >1;
1585 * bug fix: DOCUMENTATION now retrieves generic function
1586 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1587 support has been systematized, and now supports the methods
1588 specified by ANSI, along with a default method and a method for
1589 slot documentation. (reported by Nathan Froyd)
1590 * bug fix: effective methods associated with a generic function are
1591 no longer cached over a change of that generic function's method
1592 combination. (reported by Andreas Fuchs)
1593 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1594 names a non-existent directory. (thanks to Andreas Fuchs)
1595 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1596 the PPC platform, lead to stack corruption; this has been fixed.
1597 (reported by Rainer Joswig)
1598 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1599 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1600 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1601 arithmetic in general, is now much more reliable on the PPC
1603 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1604 platform now returns the right answer.
1605 * optimization: restored some effective method precomputation in
1606 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1607 precomputation is now tunable.
1608 * optimization: compiler-internal data structure use has been
1609 reviewed, and changes have been made that should improve the
1610 performance of the compiler by about 20%.
1611 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1612 simple-base-strings and simple-bit-vectors is improved.
1613 * optimization: the optimization of 32-bit logical and arithmetic
1614 functions introduced in version 0.8.3 on the x86 has been
1615 implemented on the mips, ppc and sparc platforms; an
1616 implementation of the same facility, but for 64-bit arithmetic,
1617 has been added for the alpha.
1618 * microoptimization: the compiler is better able to make use of the
1619 x86 LEA instruction for multiplication by constants.
1620 * bug fix: in some situations compiler did not report usage of
1621 generic arithmetic in (SPEED 3) policy.
1622 * bug 145b fix: compiler used wrong type specifier while converting
1623 MEMBER-types to numeric.
1624 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1626 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1627 index argument now works properly on the Alpha platform.
1628 * bug fix: floating point exception treatment on the Alpha platform
1630 * bug fix: FILE-POSITION works much better on string input and
1631 output streams. (thanks to Nikodemus Siivola)
1632 * bug fix: many threading/garbage collection symptoms sorted.
1633 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1634 might be pseudo-atomic.
1635 * internal change: Stopping for GC is now done with signals not
1636 ptrace. GC is now done in whichever thread wanted it, instead of
1638 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1640 * bug fix: result form in DO is not contained in the implicit
1642 * incompatible change: ICR structure is changed; the value part of
1643 CONTINUATION is now called LVAR; corresponding functions are
1644 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1645 * added type deriver for ISQRT (thanks to Robert E. Brown).
1646 * bug fix: better support for loading from the command line when an
1647 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1649 * fixed some bugs revealed by Paul Dietz' test suite:
1650 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1651 ** ROUND and FROUND now give the right answer when given very
1652 small float arguments.
1653 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1655 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1656 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1657 ** LCM should return a non-negative integer.
1658 ** PARSE-INTEGER returned the index of a terminator instead of the
1659 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1660 ** PARSE-INTEGER returned an incorrect index being applied to a
1662 ** LCM with two arguments of 0 returns 0 rather than signalling
1664 ** unsigned addition of a 32-bit constant with the high bit set no
1665 longer causes an internal compiler error on the x86.
1666 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1667 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1668 with negative last argument.
1669 ** byte specifiers with zero size and position no longer cause
1670 an error during type derivation.
1671 ** bignum multiplication on the Alpha platform now returns the
1673 * porting: The system now builds on SuSE AMD64, although it still
1674 generates a 32-bit binary.
1675 * .fasl file incompatibility: The fasl file version number has
1676 been incremented (because of the changes to internal compiler
1677 data structures referred to above).
1679 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1680 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1681 more accurately, on the Darwin kernel running on PowerPC hardware.
1682 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1683 * Compiler code deletion notes now signal a condition of type
1684 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1685 an associated MUFFLE-WARNING restart.
1686 * The compiler now performs limited argument count validation of
1687 constant format strings in FORMAT, and where appropriate in ERROR,
1688 CERROR and WARN. (thanks to Gerd Moellmann)
1689 * New ASDF-INSTALL contrib can be used for automatic download and
1690 installation of third-party Lisp code from CCLAN or other sites
1692 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1693 which forces another thread to execute a function supplied by the
1695 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1696 now accept and act upon their :ELEMENT-TYPE keyword argument.
1697 (reported by Martin Atzmueller, Edi Weitz)
1698 * bug fix: FILE-POSITION now accepts position designators up to
1699 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1700 is the greater. (thanks to Patrik Nordebo)
1701 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1702 Kalle Olavi Niemitalo)
1703 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1704 printable. (reported by Eric Marsden)
1705 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1706 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1707 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1708 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1709 (thanks to Stig Sandoe)
1710 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1711 UNREAD-CHAR now correctly includes the unread character in the
1712 target sequence. (thanks to Gerd Moellmann)
1713 * bug fix in threaded builds: the system can now be suspended and
1714 resumed by shell job control with minimal disruption.
1715 * bug fixes in times and timezones >2038AD
1716 * better handling of "where is GNU make?" problem in build scripts
1717 (thanks to Nikodemus Siivola)
1718 * new optimization: inside a named function any reference to a
1719 function with the same name is considered to be a self-reference;
1720 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1721 optimization quality.
1722 * new optimization on x86: logical functions and + now have
1723 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1724 used when the result is truncated to 32 bits.
1725 * VALUES declaration is partially enabled.
1726 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1727 * fixed some bugs revealed by Paul Dietz' test suite:
1728 ** The system now obeys the constraint imposed by
1729 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1730 types form a lattice under type intersection.
1731 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1732 ** ASSOC now ignores NIL elements in an alist.
1733 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1734 and (1+ MOST-POSITIVE-FIXNUM) answers.
1735 ** The addition of a method with invalid qualifiers to a generic
1736 function does not cause an error to be signalled immediately;
1737 a warning is signalled, and the error is generated only on
1738 calling the generic function.
1739 * changed .fasl file version number, in part to add type codes for
1740 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1741 obscure ANSI requirements
1743 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1744 * fixed bug 148: failure to inline-expand a local function left
1745 garbage, confusing the compiler.
1746 * fixed bugs 3cd: structure slot readers perform type check if the
1747 slot can have an invalid value (i.e. it is either not initialized
1748 or can be written with a less specific slot writer).
1749 * bug fix: the compiler now traps array references to elements off
1750 the end of an array; previously, the bounds checking in some
1751 circumstances could go off-by-one.
1752 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1754 * type declarations for array element types now obey the description
1755 on the CLHS page "Declaration TYPE", as per discussions on
1756 sbcl-help around 2003-05-08. This means that a declaration
1757 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1758 declaration, all references to BAR will be asserted or assumed
1759 (with THE, so dependent on compiler policy) to involve objects of
1760 type FOO. Note that no such declaration is implied in
1761 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1762 * declared types of functions from the "Conditions"
1763 chapter. (reported by Paul Dietz)
1764 * bug fix: CERROR accepts a function as its first argument.
1765 * bug fix: NTH an NTHCDR accept a bignum as index
1766 arguments. (reported by Adam Warner)
1767 * optimization: character compare routines now optimize comparing
1768 against a constant character. (reported by Gilbert Baumann)
1769 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1770 argument now works properly on the MIPS platform.
1771 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1772 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1773 declared to have a functional type, does not cause an error
1775 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1776 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1777 * fixed bug: if last continuation of a deleted block has a
1778 destination, this destination should be deleted too. (reported by
1780 * fixed a bug in the bootstrap process: the host compiler's values
1781 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1782 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1783 test case from Patrik Nordebo)
1784 * improved the ability of the disassembler on the PPC platform to
1785 provide helpful disassembly notes.
1786 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1787 the class in more cases than previously.
1788 * bug fix: FILE-POSITION now understands :START and :END for
1789 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1790 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1791 8) data. (thanks to David Lichteblau)
1792 * bug fix: it is possible to add a method to a generic function
1793 without lambda list.
1794 * bug fix: reader failed to signal END-OF-FILE inside an
1795 object representation. (reported by Nikodemus Siivola)
1796 * fixed some bugs revealed by Paul Dietz' test suite:
1797 ** LAST and [N]BUTLAST should accept a bignum.
1798 ** condition slot accessors are methods.
1799 ** (VECTOR NIL) is a subtype of STRING.
1801 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1802 * minor incompatible change: some nonsensical specialized lambda
1803 lists (used in DEFMETHOD) which were previously ignored now signal
1805 * minor incompatible change: the system is now aware of the types of
1806 variables in the COMMON-LISP package, and will signal errors for
1807 most violations of these type constraints (where previously they
1808 were silently accepted).
1809 * minor incompatible change: COMPILE-FILE now uses the freedom
1810 afforded (ANSI 3.2.2.3) to use derived function types for
1811 functions defined in the same file. This also permits the system
1812 to warn on static type mismatches and function
1813 redefinition. (Currently it does not work with high DEBUG level.)
1814 * minor incompatible change: VALUES declaration is disabled.
1815 * When issuing notes, the compiler now signals a condition of type
1816 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1817 restart for use in user handlers. It is expected that the
1818 COMPILER-NOTE condition will eventually become a condition
1819 supertype to a hierarchy of note types, which will then be
1820 handleable in a similar fashion. However, at the moment, no such
1821 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1822 but it's an implementation detail, not a classification for the
1824 * Changes in type checking closed the following bugs:
1825 ** type checking of unused values (192b, 194d, 203);
1826 ** template selection based on unsafe type assertions (192c, 236);
1827 ** type checking in branches (194bc).
1828 * A short form of VALUES type specifier has ANSI meaning (it has
1829 increased the number of situations when SBCL cannot perform type
1831 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1832 keywords or constants is permissible.
1833 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1834 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1835 classes. (thanks to Antonio Martinez)
1836 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1837 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1838 argument to a pretty-print stream. Instead, it leaves any such
1839 operations to DESCRIBE-OBJECT methods.
1840 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1842 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1843 lambda lists. (thanks to David Lichteblau)
1844 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1845 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1846 (thanks to Gerd Moellmann)
1847 * fixed bug 246: increased compilation speed of long
1848 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1850 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1852 * DEFCONSTANT now throws a condition of type
1853 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1854 constant to a non-EQL value; CONTINUE and ABORT restarts
1855 respectively change and preserve the value.
1856 * fixed bug 63: The code walker, part of the implementation of CLOS,
1857 is now better at handling symbol macros.
1858 * bug fix: There is no longer an internal implementation type named
1859 CL:LENGTH. (reported by Raymond Toy)
1860 * bug fix: In macro-like defining macros/special operators the
1861 implicit block does not enclose the lambda list.
1862 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1863 enclosed in parentheses) are not suitable as type specifiers, and
1864 their use properly signals an error now.
1865 * bug fix: An argument count mismatch for a type specifier in code
1866 being compiled no longer causes an unhandled error at compile
1867 time, but signals a compile-time warning.
1868 * fixed simple vector readable printing
1869 * bug fix: DESCRIBE takes more care over whether the class
1870 precedence list slot of a class is bound before accessing it.
1871 (reported by Markus Krummenacker)
1872 * bug fix: FORMATTER can successfully compile pretty-printer format
1873 strings which use variants of the ~* directive inside.
1874 * bug fix: SEARCH now applies its TEST predicate to the elements of
1875 the arguments in the correct order. (thanks to Wolfhard Buss)
1876 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1877 in type inference. (thanks to Robert E. Brown)
1878 * bug fix: Reading in symbols with an explicit package name of ""
1879 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1880 * fixed some bugs revealed by Paul Dietz' test suite:
1881 ** NIL is now allowed as a structure slot name.
1882 ** Arbitrary numbers, not just REALs, are allowed in certain
1883 circumstances in LOOP for-as-arithmetic clauses.
1884 ** Multiple class redefinitions before slot access no longer
1885 causes a type error.
1886 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1887 association between the name and a class.
1888 ** Generic functions with non-standard method-combination and over
1889 five methods all of which return constants no longer return NIL
1890 after the first few invocations. (thanks to Gerd Moellmann)
1891 ** CALL-NEXT-METHOD with no arguments now passes the original
1892 values of the arguments, even in the presence of assignment.
1893 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1895 ** DISASSEMBLE works with closures and funcallable instances.
1896 ** ADD-METHOD now returns the generic function, not the new method.
1897 ** FIND-METHOD signals an error if the lengths of the specializers
1898 is incompatible with the generic function, even if the ERRORP
1900 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1901 which its argument is a member.
1902 ** DEFCLASS only redefines the class named by its class-name
1903 argument if that name is the proper name of the class;
1904 otherwise, it creates a new class.
1905 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1906 of the UNBOUND-SLOT condition to the name of the slot.
1907 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1908 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1909 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1910 SLOT-MAKUNBOUND in the specified fashion.
1912 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1913 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1914 cross-compilation host. As a consequence, we can now bootstrap our
1915 way up to SBCL starting with a bare gcc toolchain and human-readable
1916 source code (first the source to CLISP, then the source to SBCL).
1917 * A contributed module containing a partial implementation of the
1918 simple-streams interface has been included. (thanks to Rudi
1920 * A contributed module implementing the RFC1321 Message Digest
1921 Algorithm, known as MD5, has been included.
1922 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1923 no longer has any effect, as the code controlled by this feature
1924 has been deleted. (As far as we know, no-one has ever built using
1925 this feature, and its semantics were confused in any case).
1926 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1927 work on conditions (as required by the ANSI specification),
1928 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1929 expected behaviour on conditions. Users should note, however,
1930 that such behaviour is not required by the ANSI specification,
1931 and so use of this behaviour may render their code unportable.
1932 * fixed some bugs revealed by Paul Dietz' test suite:
1933 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1935 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1936 ** SXHASH on condition objects no longer returns NIL.
1937 ** :ALLOCATION :CLASS slots are better treated; their values are
1938 updated on class redefinition, and initforms inherited from
1939 superclasses are applied.
1940 ** REMOVE-METHOD returns its generic function argument even when
1941 no method was removed.
1942 ** SHARED-INITIALIZE now initializes the values of the requested
1943 slots, including those with :ALLOCATION :CLASS.
1944 ** ALLOCATE-INSTANCE now works on structure classes defined via
1945 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1947 ** SLOT-EXISTS-P now works on conditions, as well as structures
1949 ** MAKE-LOAD-FORM now has the required methods on
1950 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1951 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1952 keyword, and now implements the SLOT-NAMES argument.
1953 ** methods with &OPTIONAL arguments no longer allow too many
1954 arguments to be passed in the call without error.
1955 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1956 option is consistent with the required arguments of the generic
1957 function lambda list.
1958 * bug fix: REQUIRE accepts a string designator. (Thanks to
1960 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1961 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1962 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1963 * bug fix: The compiler now checks for duplicated variables in macro
1965 * bug fix: SETQ on globals returns the correct value.
1966 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1967 not a COMPILER-ERROR (followed by some other strange error on
1968 choosing the CONTINUE restart).
1969 * bug fix: make.sh and friends are now more consistent in the way that
1970 they look for GNU "make".
1972 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1973 * experimental native threads support (on x86 Linux >=2.4 only).
1974 This is not compiled in by default: you need to add :SB-THREAD to
1975 the target features. See the "Beyond ANSI" chapter of the manual
1977 * fix for longstanding nonANSIism: The old distinction between
1978 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1979 The return value from CL:FIND-CLASS is now a CLOS class, and
1980 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1981 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1982 * An interface to the MetaObject Protocol, as described in Kiczales,
1983 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1984 MIT Press, 1991, is available from the SB-MOP package.
1985 * incompatible change: the SB-PCL package should now be considered
1986 a private implementation detail, and no longer a semi-private MOP
1988 * minor incompatible change: due to rearrangement for threads, the
1989 control stack and binding stack are now allocated at arbitrary
1990 addresses instead of being hardcoded per-port. Users affected by
1991 this probably have to be doing advanced things with shared
1992 libraries, and will know who they are.
1993 * minor incompatible change: Previously, all --eval forms used were
1994 processed with READ before any of them were processed with EVAL.
1995 Now each --eval form is processed with both READ and EVAL before
1996 the next --eval form is processed. (Thus package operations like
1997 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1998 work as the user might reasonably expect.)
1999 * minor incompatible change: *STANDARD-INPUT* is now only an
2000 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2002 * minor incompatible change: Y-OR-N-P is now character-oriented,
2003 not line oriented. Also, YES-OR-NO-P now works without errors.
2004 (thanks to Antonio Martinez)
2005 * sb-aclrepl module improvements: an integrated inspector, added
2006 repl features, and a bug fix to :trace command.
2007 * Known functions, which cannot be open coded by the backend, are
2008 considered to be able to check types of their arguments. (fixing
2009 a bug report by Nathan J. Froyd)
2010 * fixed a bug in computing method discriminating functions: It is
2011 now possible to define methods specialized on classes which have
2012 forward-referenced superclasses. (thanks to Gerd Moellmann)
2013 * fixed evaluation order in optional entries (reported by Gilbert
2015 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2016 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2017 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2018 required-by-AMOP NAME argument, as well as CLASS and
2019 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2020 * fixed bug 20: DEFMETHOD can define methods using names that are
2021 not the proper names of classes to designate class specializers.
2022 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2023 Unix file descriptors, instead of blowing up. (thanks to Antonio
2025 * Garbage collection refactoring: user-visible change is that a
2026 call to the GC function during WITHOUT-GCING will not do garbage
2027 collection until the end of the WITHOUT-GCING. If you were doing
2028 this you were probably losing anyway.
2029 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2030 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2031 TYPEP the latter but not the former.
2032 * The compiler issues a full WARNING for calls to undefined functions
2033 with names from the CL package.
2034 * MAP-INTO for a vector destination is open coded. (reported by
2035 Brian Downing on c.l.l)
2036 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2037 documentation string.
2038 * fixed some bugs revealed by Paul Dietz' test suite:
2039 ** COPY-ALIST now signals an error if its argument is a dotted
2041 ** Condition slots are now accessed more correctly in the presence
2042 of multiple initargs for a given slot.
2043 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2044 exclude restarts of the same name associated with a different
2046 ** DEFCLASS of forward-referenced classes with another
2047 forward-referenced class in the superclasses list no longer
2049 ** Condition slots are now initialized once each, not multiple
2050 times. (thanks to Gerd Moellmann)
2051 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2053 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2054 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2055 arguments contain duplicated elements.
2056 ** RESTART-CASE understands local macros.
2057 ** RESTART-CASE associates exactly its own restarts with a condition.
2058 ** ENDP in safe mode checks its argument to be of type LIST.
2059 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2060 in question is unbound.
2061 ** Optimized MAKE-INSTANCE functions no longer cause internal
2062 assertion failures in the presence of duplicate initargs.
2063 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2064 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2065 propagated even in the ctor optimized implementation of
2067 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2069 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2070 * a better implementation of SXHASH on (simple) bit vectors,
2071 measured both in execution speed and in distribution of results
2072 over the positive fixnums, has been installed. Likewise, a better
2073 implementation of EQUAL for simple bit vectors is now available.
2074 * fixed CEILING optimization for a divisor of form 2^k.
2075 * fixed bug 240 (emitting extra style warnings "using the lexical
2076 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2077 by Antonio Martinez)
2078 * fixed SXHASH, giving different results for NIL depending on type
2079 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2080 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2081 variable. (found by Rolf Wester)
2082 * SBCL does not ignore type declarations for special
2083 variables. (reported by rif on c.l.l 2003-03-05)
2084 * some bug fixes in contrib/sb-aclrepl/
2085 * fixed some bugs revealed by Paul Dietz' test suite:
2086 ** a bug in the CONS type specifier, whereby the CAR and CDR
2087 types got intertwined, has been fixed;
2088 ** the type system is now able to reason about the interaction
2089 between INTEGER and RATIO types more completely;
2090 ** APPEND, [N]REVERSE and NRECONC check that those their
2091 arguments, which must be proper lists, are really so;
2092 ** An array specialized to be unable to hold elements has been
2093 implemented, as required -- yes, really -- by ANSI;
2094 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2095 on malformed property lists;
2097 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2098 * incompatible packaging change: in line with Unix convention,
2099 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2100 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2101 * REQUIRE and PROVIDE are now optionally capable of doing something
2102 useful. See the documentation string for REQUIRE.
2103 * infrastructure for a managed SBCL contrib system: contributed
2104 modules in this release include:
2105 ** the ASDF system definition facility;
2106 ** an interface to the BSD Sockets API;
2107 ** an ACL-like convenience interface to the repl;
2108 (thanks to Kevin Rosenberg)
2109 ** an implementation of ROTATE-BYTE, with efficient implementation
2111 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2112 gives a closer approximation to the right answer than previously.
2113 (thanks to Raymond Toy)
2114 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2115 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2116 optional environment arguments, as required by ANSI.
2117 * fixed bugs in other functions taking environment objects, allowing
2118 calls with an explicit NIL environment argument to be compiled
2120 * fixed bug 228: primary return values from
2121 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2122 COMPILE or FUNCTION.
2123 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2124 :TYPE will no longer signal errors on innocuous objects.
2125 * fixed bug 231b: SETQ is better at respecting type declarations in
2126 the lexical environment.
2127 * fixed a bug in DEFCLASS: classes named by symbols with no or
2128 unprintable packages can now be defined.
2129 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2130 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2131 * fixed compiler failure related to checking types of functions.
2132 (reported by Robert E. Brown)
2133 * the compiler is now much more consistent in its error-checking
2134 treatment of bounding index arguments to sequence functions: in
2135 (SAFETY 3) code, errors will be signalled in almost all cases if
2136 invalid sequence bounding indices are passed to functions defined
2137 by ANSI to operate on sequences.
2138 * fixed a bug in the build procedure: documentation of SBCL-specific
2139 packages is now preserved and available in the final Lisp image.
2140 * lifted FDEFINITION lookup out of loops in the implementation of
2141 many list operations. (thanks to Robert E. Brown)
2142 * fixed a bug in the reader: the #n# reader macro now works for
2143 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2144 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2145 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2146 E. Brown and Tony Martinez respectively)
2147 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2148 SLOT-BOUNDP: in particular, it is now able to optimize them much
2149 better, and is now not vulnerable to having packages renamed.
2150 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2151 always called when a slot is not present in an instance. (thanks
2153 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2154 longer signal an error on unreading a character following EOF on
2155 the previous constituent stream. (thanks to Tony Martinez)
2156 * fixed some bugs revealed by Paul Dietz' test suite:
2157 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2158 not just nonnegative fixnums;
2159 ** the logical bit-array operators such as BIT-AND now accept an
2160 explicit NIL for their "opt-arg" argument (to indicate a
2161 freshly-consed result bit-array);
2162 ** ELT now signals an error on an invalid sequence index in safe
2164 ** the type system is now cleverer about negations of numeric
2165 types, and consequently understands the BIGNUM and RATIO types
2167 ** the type system is now cleverer about the interaction between
2168 INTEGER and RATIO types: while bugs still remain, many more
2169 cases are accurately computed;
2170 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2171 if it is in the last clause;
2172 ** CONSTANTLY now correctly returns a side-effect-free function in
2174 ** DECLARE is no longer treated as a special-operator; in
2175 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2176 * incremented fasl file version number due to the change in the
2177 DEFSTRUCT-SLOT-DESCRIPTION structure.
2179 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2180 * minor incompatible change: code processed by the "interpreter" or
2181 EVAL now has a compilation optimization policy of (DEBUG 2)
2182 (changed from (DEBUG 1)) to improve debuggability of interactive
2183 development, and to allow the use of the debug RETURN command in
2185 * an experimental implementation of the RETURN command for the
2186 debugger has been included. (thanks to Frederik Kuivinen)
2187 * fixed bug 62: constraints were not propagated into a loop.
2188 * fixed bug in embedded calls of SORT (reported and investigated by
2190 * fixed some bugs revealed by Paul Dietz' test suite:
2191 ** printing and reading of arrays with some dimensions having
2192 length 0 (thanks to Gerd Moellmann);
2193 ** BOA constructor with &AUX argument without a default value does
2194 not cause a type error;
2195 ** CONSTANTP now returns true for all self-evaluating objects.
2197 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2198 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2199 accessors that are related by inheritance, as specified in the
2200 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2201 Valtteri Vuorikoski)
2202 * The compiler is now able to inline functions that were defined in
2203 a complex lexical environment (e.g. inside a MACROLET).
2204 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2206 * Support for the upcoming FreeBSD-5.0 release has been included.
2207 (thanks to Dag-Erling Smorgrav)
2208 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2209 effect when it is not in a toplevel context.
2210 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2211 stand a better chance of being correct. (thanks to Gerd
2213 * fixed bug in COERCE, which now signals an error on coercing a
2214 rational to a bounded real type which excludes the expected
2216 * The compiler is now able to derive types more accurately from the
2217 COERCE and COMPILE functions.
2218 * fixed bug 223: functional binding is considered to be constant
2219 only for symbols in the CL package.
2220 * fixed bug 231: SETQ did not check the type of a variable being set
2221 (reported by Robert E. Brown)
2222 * A new optimization for MAKE-INSTANCE has been included, fixing
2223 various bugs (including relating to :ALLOCATION :CLASS slots and
2224 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2226 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2227 ** As required by ANSI, LOOP now disallows anonymous collection
2228 clauses such as COLLECT I in conjunction with aggregate boolean
2229 clauses such as THEREIS (= I 1);
2230 ** LOOP now signals an error when any variable is reused in the
2231 same loop (including the potentially useful construct analogous
2232 to WITH A = 1 WITH A = (1+ A);
2233 ** IT is only a special loop symbol within the first clause of a
2234 conditional loop clause;
2235 ** LOOP with a typed iteration variable over a hashtable now
2236 signals a type error iff it should.
2237 * fixed some other bugs revealed by Paul Dietz' test suite:
2238 ** FILE-STREAM now names the class previously known as FD-STREAM;
2239 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2240 argument) no longer signals an error;
2241 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2242 of the slot symbol, rather than using the current package
2243 ((:CONC-NAME "") continues to intern the slot's name in the
2245 * incremented fasl file version number, because of the incompatible
2246 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2247 of the new implementation of DEFINE-COMPILER-MACRO.
2249 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2250 * Support for building SBCL for MIPS platforms running in
2251 little-endian mode has now been checked in, and basic
2252 functionality on said platforms verified.
2253 * minor incompatible change: PCL now records the pathname of a file
2254 in which methods and the like are defined, rather than its
2256 * minor incompatible change: TRUENAME now considers the truename of
2257 a file naming a directory to be the pathname with :DIRECTORY
2258 component indicating that directory.
2259 * minor incompatible change: a NAMED clause in the extended form of
2260 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2261 reason for the previous behaviour is unclear.
2262 * more systematization and improvement of CLOS and MOP conformance
2263 in PCL (thanks to Gerd Moellman and Pierre Mai):
2264 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2266 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2267 primary methods with no specializers;
2268 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2270 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2271 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2272 FORWARD-REFERENCED-CLASSes; error reporting on
2273 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2275 ** SXHASH on CLOS instances now uses a slot internal to the
2276 instance to return different numbers on distinct instances,
2277 while preserving the same return value through invocations of
2279 ** DEFMETHOD signals errors when methods with longer incongruent
2280 lambda lists are added to generic functions;
2281 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2282 CLASS, as specified in AMOP;
2283 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2284 on the order returned by the primary method for classes of
2285 class STANDARD-CLASS;
2286 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2287 * fixed some bugs shown by Paul Dietz' test suite:
2288 ** DOLIST puts its body in TAGBODY;
2289 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2291 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2292 value producing form;
2293 ** if more variables are given to PROGV than values, extra
2294 variables are bound and made to have no value;
2295 ** NSUBSTITUTE on list arguments gets the right answer with
2297 ** ELT signals an error of type TYPE-ERROR when the index argument
2298 is not a valid sequence index;
2299 ** LOOP signals (at macroexpansion time) an error of type
2300 PROGRAM-ERROR when duplicate variable names are found;
2301 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2302 ** FUNCALL of special-operators now cause an error of type
2304 ** PSETQ now works as required in the presence of side-effecting
2305 symbol-macro places;
2306 ** NCONC accepts any object as its last argument;
2307 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2309 ** loop-for-as-package does not require a package to be explicitely
2311 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2313 * fixed bug 166: compiler preserves "there is a way to go"
2314 invariant when deleting code.
2315 * fixed bug 172: macro lambda lists with required arguments after
2316 &REST arguments now cause an error to be signalled. (thanks to
2318 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2319 bug. (thanks to Matthew Danish)
2320 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2322 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2323 arguments are lexically rebound. (thanks to Gerd Moellmann and
2325 * fixed bug 194: error messages are now more informative when there
2326 is no primary method applicable in a call to a generic
2327 function. (thanks to Gerd Moellmann)
2328 * fixed bug in command line argument checking (thanks to Julian
2330 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2331 (thanks to Matthew Danish)
2332 * incremented fasl file version number, because of the
2333 SXHASH-related changes in the layout of CLOS data structures
2335 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2336 * minor incompatible change: The runtime (the Unix executable named
2337 "sbcl") is now much pickier about the .core files it will load.
2338 Essentially it now requires .core files to descend from the same
2339 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2340 as the runtime does. (The intent is to prevent the crashes which
2341 can occur, and which can even be reported as mysterious failures,
2342 when people patch the sources or change the build parameters
2343 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2344 sbcl and .core files.)
2345 * fixed bug: VALUES-LIST is no longer optimized away.
2346 * fixed bug 142: The FFI conversion of C string values to Lisp
2347 string values no longer conses excessively. (thanks to Nathan
2348 Froyd porting Raymond Toy's fix to CMU CL)
2349 * began to systematize and improve MOP conformance in PCL (thanks to
2350 Nathan Froyd, Gerd Moellman and Pierre Mai):
2351 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2353 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2354 ** FINALIZE-INHERITANCE is now called on class finalization;
2355 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2356 argument precedence order.
2357 * fixed bug 202: The compiler no longer fails on functions whose
2358 derived types contradict their declared type.
2359 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2360 so it can be non-toplevel.
2361 * The fasl file version number has changed (because of the new
2362 implementation of DEFMACRO).
2363 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2364 safe code, that any length requirement by their type-specifier
2365 argument is valid. The exceptions to this are described in bug
2367 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2368 functions now check the types of their inputs as required by ANSI.
2369 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2370 introduced symbol is DECLAREd to be SPECIAL.
2371 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2372 * fixed bug: :COUNT argument to sequence functions may be negative
2373 * fixed bug: body of DO-SYMBOLS may contain declarations
2374 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2375 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2376 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2377 symbol macro only once
2378 * fixed printing of call frame when argument list is unavailable
2379 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2380 * compiler no longer signals WARNING on unknown keyword
2383 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2384 * A beta-quality port to the mips architecture running Linux,
2385 based on the old CMUCL backend, has been made. It has been tested
2386 on a big-endian kernel, and works sufficiently well to be able to
2387 rebuild itself; it has not been tested in little-endian mode.
2388 * fixed an inconsistency between gencgc.c and purify.c which made
2389 dumping/loading .core files unreliable
2390 * fixed bug 120a: The compiler now deals correctly with IFs where
2391 the consequent is the same as the alternative, instead of
2392 misderiving the return type. (thanks to Alexey Dejneka)
2393 * fixed bug 113: Logical pathnames are now dumpable (the logical
2394 host is resolved at load-time, throwing an error if it is not
2396 * fixed bug 174: FORMAT's error message is slightly clearer when a
2397 non-printing character is used in a format directive.
2398 * fixed several bugs in compiler checking of type declarations, i.e.
2399 violations of the Python "declarations are assertions" principle
2400 (thanks to Alexey Dejneka)
2401 * fixed several bugs in PCL's error checking (thanks to Gerd
2403 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2405 * fixed bug in compilation of functions as first class values
2406 (thanks to Antonio Martinez-Shotton)
2407 * The compiler's handling TYPE-ERRORs which it can prove will
2408 inevitably happen at runtime has been cleaned up and corrected
2409 in several ways. (thanks to Alexey Dejneka)
2410 * improved argument type checking for various basic arithmetic
2411 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2412 lavished on them in the past that they can be compiled in many
2413 ways in different special cases
2414 * fixed bug 181: compiler checks validity of user supplied type
2416 * cleaned up code flushing in optimization: Function calls which
2417 should signal errors for safety purposes (e.g. which ANSI says
2418 should signal errors when their arguments are of incorrect type)
2419 are no longer optimized away.
2420 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2421 * incremented fasl file version number, because changes in the
2422 implementation of sequence functions like COERCE caused
2423 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2424 in old inline expansions) to become undefined. (Actually these
2425 changes were later undone, so we might very well be binary
2426 compatible with 0.7.7 after all, but leaving the version number
2427 incremented seemed like the simplest and most conservative
2430 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2431 * An alpha-quality port to the parisc architecture running Linux,
2432 based on the old CMUCL backend, has been made. This, even more so
2433 than the other backends, should be considered still a work in
2434 progress; known problems include that the Linux kernel in 64-bit
2435 mode does not propagate the correct sigcontext structure to
2436 userspace, and consequently SBCL on a parisc64 kernel will not
2438 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2439 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2440 suppressed.) Also now that the compiler is looking at declarations
2441 in the environment, it checks optimization declarations as well,
2442 and suppresses inlining when (> DEBUG SPEED).
2443 * More fixes have been made to treatment of floating point exception
2444 treatment and other Unix signals. In particular, floating point
2445 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2446 * The detection and handling of control stack exhaustion (infinite
2447 or very deeply nested recursion) has changed. Stack exhaustion
2448 detection is now done by write-protecting pages at the OS level
2449 and applies at all optimization settings; when found, a
2450 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2451 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2452 be caught using IGNORE-ERRORS.
2453 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2454 correctly when transforming recursive LABELS functions to LETs.
2455 (thanks to Alexey Dejneka)
2456 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2457 that are names of constants or global variables.
2458 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2459 alien routines with docstrings.
2460 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2461 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2463 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2464 (thanks to Eric Marsden)
2465 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2466 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2467 * Bug fix: The SPARC backend can now compile functions involving
2468 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2469 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2470 to a functions expecting a PATHNAME-DESIGNATOR.
2471 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2472 lambda lists. (thanks to Alexey Dejneka)
2473 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2474 functions so that EXPORTing the name of the class doesn't cause
2475 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2476 to Antonio Martinez for reporting this)
2477 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2478 constant-folded without failing an assertion. (thanks to Einar
2479 Floystad Dorum for reporting this)
2480 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2481 types are now dealt with more correctly by the compiler.
2482 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2483 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2484 behaviour analogous to RENAME-FILE. This puts its behaviour more
2485 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2487 * The fasl file version number has changed again. (because of the
2488 bug fix involving the names of PCL MAKE-INSTANCE functions)
2490 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2491 * bug fix: Floating point exceptions are treated much more
2492 consistently on the x86/Linux and PPC/Linux platforms.
2493 * Array initialization with :INITIAL-ELEMENT is now much faster for
2494 cases when the compiler cannot open code the array creation, but
2495 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2496 array accesses have also seen a speed increase.
2497 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2498 specified. (This was at the root of some bad interactions between
2499 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2501 * bug fix: Internal error arguments for undefined functions are now
2502 computed correctly on the PPC/Linux platform.
2503 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2504 Raymond Toy's patch for CMU CL)
2505 * Support for the Solaris 9 operating environment has been included
2506 (thanks to Daniel Merritt)
2507 * A very ugly but hopefully complete draft of the missing FFI chapter
2508 of the manual has been created by reformatting the corresponding
2509 CMU CL manual chapter into (currently very ugly and incoherent)
2510 DocBook and bringing it up to date for SBCL behavior. Thus, the
2511 manual is now essentially complete, at least by my extreme
2512 once-and-only-once standards, whereby it's acceptable to refer to
2513 the doc strings of SB-EXT functions as the primary documentation.
2514 * The fasl file version number has changed again, due to cleanup of
2515 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2516 is no longer a static symbol.)
2518 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2519 * SBCL now builds with OpenMCL (version 0.12) as the
2520 cross-compilation host; also, more progress has been made toward
2521 bootstrapping under CLISP.
2522 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2524 * bug 158 fixed: The compiler can now deal with integer loop
2525 increments different from 1; fixing this turned out also to fix
2527 * bug 169 fixed: no more bogus warnings about using lexical bindings
2528 despite the presence of perfectly good SPECIAL declarations (thanks
2529 to David Lichteblau)
2530 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2531 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2532 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2534 * bug 180 fixed: Method combination specifications no longer ignore
2535 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2536 * bug fix: Structure type predicate functions now check their argument
2537 count as they should.
2538 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2539 correctly. (thanks to Pierre Mai)
2540 * minor incompatible change: The --noprogrammer option is deprecated
2541 in favor of the new --disable-debugger option, which is very similar.
2542 (The major difference is that it takes effect at a slightly different
2543 time at startup, causing handling of errors in --sysinit and
2544 --userinit files will be affected differently.) The
2545 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2546 been added to allow this functionality to be controlled from ordinary
2547 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2548 maintainers, who might want to run non-interactive scripts to
2549 build SBCL cores which will later be used interactively.)
2550 * minor incompatible change: The LOAD function no longer, when given
2551 a wild pathname to load, loads all files matching that pathname.
2552 Instead, an error of type FILE-ERROR is signalled.
2554 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2555 * bug 147 fixed: The compiler preserves its block link/count
2556 invariants more correctly now so that it doesn't crash. (thanks
2558 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2560 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2561 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2562 no-value-for-this-character as NIL (as specified by ANSI).
2563 * HOST-NAMESTRING on physical pathnames now returns a string that is
2564 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2565 (thanks to Christophe Rhodes)
2566 * The Alpha port handles icache flushing more correctly. (thanks to
2568 * More progress has been made toward bootstrapping under CLISP. (thanks
2569 to Christophe Rhodes)
2570 * The fasl file format has changed again, because dynamic loading
2571 on OpenBSD (which has non-ELF object files) motivated some cleanups
2572 in the way that foreign symbols are transformed and passed around.
2573 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2574 is no longer treated as whitespace by the reader, but instead as
2575 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2576 |A
\7fB|, instead of A as it used to.
2578 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2579 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2580 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2581 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2582 this as of 0.7.1.45, but was left out of the previous news section
2583 (thanks to Dan Barlow)
2584 * SBCL now runs on the Solaris operating system on SPARC architectures
2585 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2586 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2587 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2588 * SPARC backend cleanups, allowing builds of cores optimized for V8
2589 and V9 SPARCS, and also emission of code targeted to a particular
2590 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2592 * SBCL is closer to bootstrapping under CLISP, thanks to various
2593 fixes by Christophe Rhodes.
2594 * The fasl file format has changed again, to allow the compiler's
2595 INFO database to support symbol macros.
2596 * The user manual (in doc/) is formatted into HTML more nicely.
2597 (thanks to coreythomas)
2598 * The system is smarter about SUBTYPEP relationships, especially
2599 those involving NOT types (including types such as ATOM which are
2600 represented internally using NOT types). Thus SUBTYPEP is less
2601 likely to return (VALUES NIL NIL) in general, and in particular
2602 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2604 * The fasl file format has changed again, because the internal
2605 representation of types now includes a new slot to support the new
2606 SUBTYPEP-of-NOT-types logic.
2607 * (not a change in the main branch of SBCL, but a related prototype
2608 which can hopefully be merged into the main branch of SBCL in the
2609 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2610 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2611 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2612 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2613 dispatch macro character combinations. (thanks to Alexey Dejneka)
2614 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2615 that superficially look like logical namestrings correctly.
2616 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2618 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2619 default supplied-p) for &optional and &key arguments. (thanks to
2622 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2623 * incompatible change: The compiler is now less aggressive about
2624 tail call optimization, doing it only when (> SPACE DEBUG) or
2625 (> SPEED DEBUG). (This is an incompatible change because there are
2626 programs which relied on the old CMU-CL-style behavior to optimize
2627 away their unbounded recursion which will now die of stack overflow.)
2628 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2629 for non-GENCGC systems has been increased to 20M (since that
2630 seems much closer to the likely performance optimum for modern
2631 systems than the old 4M value was)
2632 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2633 and *DEBUG-PRINT-LEVEL*
2634 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2635 of CMU CL's support for SPARC, and various endianness and other
2636 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2637 * new syntactic sugar for the Unix command line: --load foo.bar is now
2638 an alternate notation for --eval '(load "foo.bar")'.
2640 ** The system now detects stack overflow and handles it gracefully,
2641 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2642 optimization settings. (This is a good thing in general, and
2643 its introduction in this version should be particularly timely
2644 for anyone whose code fails because of suppression of tail
2646 ** The system now hunts for the C variable "environ" in a more
2647 devious way, to avoid segfaults when the C library version differs
2648 between compile time and run time. (thanks to Christophe Rhodes)
2649 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2650 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2651 ** The compiler no longer issues bogus style warnings for undefined
2652 classes in the same source file as the DEFCLASSes which defined
2653 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2655 ** fixes in CONDITION class precedence list for undefined function
2656 errors (thanks to Alexei Dejneka)
2657 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2658 correctly. (thanks to Dan Barlow)
2659 ** portability fixes aiming at bootstrapping under CLISP (thanks
2660 to Dave McDonald and Christophe Rhodes)
2661 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2662 ** fixes in type translation and and type inference (thanks to
2664 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2665 ** various fixes in the new ports (thanks to Dan Barlow)
2666 * several changes related to debugging:
2667 ** suppression of tail recursion, as noted above
2668 ** stack overflow detection, as noted above
2669 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2670 is now the default. (For some time encapsulation has been more
2671 reliable than the breakpoint-based :ENCAPSULATE NIL
2672 implementation, at least on X86 systems; and I just noticed that
2673 encapsulation also seems closer to the spirit of the ANSI
2676 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2678 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2679 up properly again. (There was a packaging bug in 0.7.0 which
2680 left their definitions in SB-SYS::LOAD-FOREIGN and
2681 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2682 vital for most things which interface to C-level interfaces,
2683 like extensions working with sockets or databases or
2684 Perl-compatible regexes or whatever, and the need to fix
2685 this bug is the main reason that 0.7.1 was released so
2687 ** DEFGENERIC is now choosier about the methods it redefines, so that
2688 reLOADing a previously-LOADed file containing DEFGENERICs does
2689 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2690 cycle now works as it should. (thanks to Alexey Dejneka)
2691 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2692 fixed by Christophe Rhodes. (He actually submitted this patch
2693 months ago, and I delayed until after 0.7.0.)
2694 ** Bug 111 (internal compiler confusion about runtime checks on
2695 FUNCTION types) was fixed by Alexey Dejneka.
2696 * Some internal cleanups (getting rid of variables which aren't
2697 needed now that the byte interpreter is gone) caused the fasl
2698 file format number to change again.
2700 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2701 * major incompatible change: The default fasl file extension, i.e. the
2702 default extension for files produced by COMPILE-FILE, has changed
2703 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2705 ** There are many changes in the implementation of the compiler.
2706 SBCL is now essentially a compiler-only implementation of ANSI
2707 Common Lisp. EVAL still "interprets" a few special cases, but
2708 almost all the interesting cases are handled by creating
2709 a LAMBDA expression, calling COMPILE on it, then calling
2710 FUNCALL on the result.
2711 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2712 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2713 Since the code is newer, there might still be some new bugs
2714 (though not as many as before Martin Atzmueller's fixes:-). But
2715 the new code is substantially simpler and clearer, and hopefully
2716 any remaining bugs will be simpler, less fundamental, and more
2717 fixable then the bugs in the old code.
2718 ** The revised compiler is still a little unsteady on its feet.
2720 *** The debugging information it produces (particularly the names
2721 of FUNCTION objects) is sometimes much less useful than what
2722 the old compiler produced.
2723 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2724 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2725 within a MACROLET) has been temporarily weakened.
2726 ** There are new compiler optimizations for various functions:
2727 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2728 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2729 *** the math functions TRUNCATE, FLOOR, and CEILING
2730 *** the function-of-all-trades COERCE
2731 Mostly these should be transparent, but there's one
2732 potentially-annoying problem (bug 117): when the compiler
2733 inline-expands a function and does type analysis on the result,
2734 it can create control paths which have type mismatches, and
2735 when it can't prove that those control paths aren't taken,
2736 it will issue WARNINGs about the type mismatches. This is
2737 a particular problem in practice for the new sequence functions.
2738 It's not clear how this should be fixed, and for now, a
2739 workaround is given in the entry for 117 in the BUGS file.
2740 ** (Because of the interaction between the two previous items --
2741 occasional inlining problems and new inline expansions -- some
2742 of the new sequence function optimizations won't really kick in
2743 completely until debugging information, and then inlining, are
2744 straightened out in some future version.)
2745 * minor incompatible changes:
2746 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2747 DIRECTORY no longer implicitly promotes NIL slots of its
2748 pathname argument to :WILD. In particular, when you ask for the
2749 contents of a directory (which you used to be able to do without
2750 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2751 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2752 ** changes in behavior that ANSI explicitly defines to be
2753 implementation dependent:
2754 *** The new compiler-only implementation still conforms with ANSI,
2755 but acts a little different than before. Besides the obvious
2756 changes in performance tradeoffs (that the cost per form passed
2757 to EVAL has gone up, and the cost per form executed by EVAL
2758 has gone down), the behavior of the system changes a little
2759 because there are no longer any interpreted function objects.
2760 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2761 e.g. doing COMPILE on the output of interactive DEFUN is
2763 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2765 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2766 has changed from (:CL) to NIL.
2767 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2768 e.g. #\Newline and #\Tab, don't have names specified in the
2769 ANSI Common Lisp standard, is now based on their ASCII symbolic
2770 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2771 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2772 are no longer used for output.
2773 ** changes in internal implementation constants:
2774 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2775 4 million. (If your application spends a lot of time GCing and
2776 you have a lot of RAM, you might want to experiment with
2777 increasing it even more.)
2778 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2779 However, almost all old code should still continue to work without
2780 immediate update, as SB-C-CALL is now a (deprecated) nickname
2782 ** Old operator names in the style DEF-FOO are now deprecated in
2783 favor of new corresponding names DEFINE-FOO, for consistency with
2784 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2785 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2786 internal symbols, but a few supported extensions like
2787 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2788 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2789 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2790 etc. as you get deeper into recursive calls to the debugger
2791 command loop, instead of the old "5]", "5]]", "5]]]"
2792 sequence. (I was motivated to do this when squabbles between
2793 ILISP and SBCL left me very deeply nested in the debugger. In the
2794 short term, this change will probably provoke more ILISP/SBCL
2795 squabbles, but hopefully it will be an improvement in the long run.)
2796 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2797 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2798 function, since declaiming return types involving aliens is
2799 (1) annoyingly messy to do by hand and (2) vital to efficient
2800 compilation of code which calls such functions.
2801 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2802 longer reexported by the SB-EXT package. They're solely useful
2803 for alien code, so it seems more logical that you should get
2804 them from the SB-ALIEN package, not in SB-EXT.
2805 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2806 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2807 features. Instead, the code that they used to control is always
2808 built into the system.
2809 * many other bug fixes
2810 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2811 advantage of the new EVAL-WHEN stuff and to clean them up in
2812 general, and they are now more ANSI-compliant in a number of
2813 ways. Martin Atzmueller is responsible for a lot of this.
2814 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2816 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2817 *** correct ERROR type for various file operations
2818 *** some fixes for Lisp streams
2819 *** DEFMETHOD syntax checking
2820 *** changing old weird representation of debug information as
2821 strings (which, among their other deficiencies, don't transform
2822 correctly when you rename packages, and don't change their
2823 print representation when you change things like *PACKAGE*
2824 and *PRINT-LENGTH*) to symbols and lists of symbols
2825 He also made several improvements and fixed several bugs in DESCRIBE.
2826 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2828 *** misbehavior of WRITE-STRING/WRITE-LINE
2829 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2830 and several other LOOP problems as well
2831 *** DIRECTORY when similar filenames are present
2832 *** DEFGENERIC with :METHOD options
2833 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2834 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2835 *** argument ordering in FIND with :TEST option
2836 *** mishandled package designator argument in APROPOS-LIST
2837 *** various problems in the backquote readmacro
2838 *** a bug in APROPOS
2839 *** probably some others that I'm not describing very well here,
2840 since the CVS log documents them by reference to sbcl-devel
2841 messages, and the SourceForge archives aren't working well.:-(
2842 ** Dan Barlow improved the Alpha port (and is making progress on the
2843 PPC port, for those of you who think different).
2844 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2845 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2846 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2847 continues to work on the SPARC port (for those of you in a position
2848 to look down upon our little PC-compatible boxes from a great height).
2849 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2850 and uses that copy, rather than the current dynamic value, when
2851 it's trying to decide whether to truncate output. Thus e.g.
2852 (let ((*print-lines* 50))
2853 (pprint-logical-block (stream nil)
2855 (let ((*print-lines* 8))
2856 (print (aref possiblybigthings i) stream)))))
2857 should now truncate the logical block only at 50 lines, instead of
2858 often truncating it at 8 lines, as it did before.
2859 * The doc/cmucl/ directory, containing old CMU CL documentation
2860 from the time of the fork, is no longer part of the base system.
2861 SourceForge has shut down its anonymous FTP service, and with it
2862 my original plan for distributing the old CMU CL documentation
2863 there. For now, if you need these files you can download an old
2864 SBCL source release and extract them from it.
2865 * The fasl file version number changed again, for dozens of reasons,
2866 some of which are apparent above.
2868 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2869 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2870 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2871 patch, so that DISASSEMBLE output is much nicer.
2872 * The code in the SB-PROFILE package now seems reasonably stable.
2873 I still haven't decided what the final interface should look like
2874 (I'd like PROFILE to interact cleanly with TRACE, since both
2875 facilities use function encapsulation) but if you have a need
2876 for profiling now, you can probably use it successfully with
2877 the current CMU-CL-style interface.
2878 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2879 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2880 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2881 link now returns the dangling link itself, and for similar
2882 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2883 link itself. (In these cases the old code signalled an error and
2884 looped endlessly, respectively.) Thus, DIRECTORY now works even
2885 in the presence of dangling and cyclic symbolic links.
2886 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2887 is now a supported extension again, since the consensus on
2888 sbcl-devel was that it can be useful for ordinary development
2889 work, not just for debugging SBCL itself.
2890 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2891 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2892 that currently-defined functions might be redefined later with
2893 different return types.
2894 * Hash tables can be printed readably, as inspired by CMU CL code
2895 of Eric Marsden and SBCL code of Martin Atzmueller.
2896 * better error handling in CLOS method combination, thanks to
2897 Martin Atzmueller porting Pierre Mai's CMU CL patches
2898 * more overflow fixes for >16Mbyte I/O buffers
2899 * A bug in READ has been fixed, so that now a single Ctrl-D
2900 character suffices to cause end-of-file on character streams.
2901 In particular, now you only need one Ctrl-D at the command
2902 line (not two) to exit SBCL.
2903 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2905 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2906 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2907 does the right thing.
2908 * The implementation of some type tests, especially for CONDITION
2909 types, is now tidier and maybe faster, due to CMU CL code
2910 originally by Douglas Crosher, ported by Martin Atzmueller.
2911 * Some math functions have been fixed, and there are new
2912 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2913 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2914 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2915 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2916 likely to be built into sbcl-0.7.0.)
2917 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2918 CL, ported to SBCL by Martin Atzmueller)
2919 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2920 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2921 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2922 * There's a new slam.sh hack to shorten the edit/compile/debug
2923 cycle for low-level changes to SBCL itself, and a new
2924 :SB-AFTER-XC-CORE target feature to control the generation of
2925 the after-xc.core file needed by slam.sh.
2926 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2927 COMPILE-FILE is no longer supported, so that now every function
2928 gets an entry point, so that block compilation looks a little
2929 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2930 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2931 returns the number of bytes consed since the system started,
2932 rather than the number consed since the first time the function
2933 was called. (The new definition parallels ANSI functions like
2934 CL:GET-INTERNAL-RUN-TIME.)
2935 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2936 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2937 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2938 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2939 * Fasl file version numbers are now independent of the target CPU,
2940 since historically most system changes which required version
2941 number changes have affected all CPUs equally. Similarly,
2942 the byte fasl file version is now equal to the ordinary
2945 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2946 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2947 is no longer recognized. I apologize for this, because it was
2948 listed in SB-EXT as a supported extension, but I found that
2949 its existing behavior was poorly specified, as well as incorrectly
2950 specified, and it looked like too much of a mess to straighten it
2951 out. I have enough on my hands trying to get ANSI stuff to work..
2952 * many patches ported from CMU CL by Martin Atzmueller, with
2953 half a dozen bug fixes in pretty-printing and the debugger, and
2954 half a dozen others elsewhere
2955 * fixed bug 13: Floating point infinities are now supported again.
2956 They might still be a little bit flaky, but thanks to bug reports
2957 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2958 as flaky as they were.
2959 * The --noprogrammer command line option is now supported. (Its
2960 behavior is slightly different in detail from what the old man
2961 page claimed it would do, but it's still appropriate under the
2962 same circumstances that the man page talks about.)
2963 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2964 are now supported, and enabled by default. Thus, the compiler can
2965 handle many floating point and complex operations much less
2966 inefficiently. (Thus e.g. you can implement a complex FFT
2968 * The compiler now detects type mismatches between DECLAIM FTYPE
2969 and DEFUN better, and implements CHECK-TYPE more correctly, and
2970 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2971 thanks to patches from Martin Atzmueller.
2972 * various fixes to make the cross-compiler more portable to
2973 ANSI-conforming-but-different cross-compilation hosts (notably
2974 Lispworks for Windows, following bug reports from Arthur Lemmens)
2975 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2976 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2977 fixed thanks to Pierre Mai's CMU CL patches.
2978 * fixes to keep the system from overflowing internal counters when
2979 it tries to use i/o buffers larger than 16M bytes
2980 * fixed bug 45a: Various internal functions required to support
2981 complex special functions have been merged from CMU CL sources.
2982 (When I was first setting up SBCL, I misunderstood a compile-time
2983 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2984 * improved support for type intersection and union, fixing bug 12
2985 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2986 more obscure bugs as well
2987 * some steps toward byte-compiling non-performance-critical
2988 parts of the system, courtesy of patches from Martin Atzmueller
2989 * Christophe Rhodes has made some debian packages of sbcl at
2990 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2991 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2992 stable, but are nonetheless usable. When he's ready, I'd be happy
2993 to add them to the SourceForge "File Releases" section. (And if
2994 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2995 * new fasl file format version number (because of changes in
2996 internal representation of (OR ..) types to accommodate the new
2997 support for (AND ..) types, among other things)
2999 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3000 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3002 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3004 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3005 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3006 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3007 * DESCRIBE now works on structure objects again.
3008 * Most function call argument type mismatches are now handled as
3009 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3010 know whether the function will be redefined before the call is
3011 executed. (The compiler could flag local calls with full WARNINGs,
3012 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3013 it doesn't keep track of enough information to know whether calls
3014 are local in this sense.)
3015 * Compiler output is now more verbose, with messages truncated
3016 later than before. (There should be some supported way for users
3017 to override the default verbosity, but I haven't decided how to
3018 provide it yet, so this behavior is still controlled by the internal
3019 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3020 src/compiler/ir1util.lisp.)
3021 * Fasl file format version numbers have increased again, because
3022 support for the Gray streams extension changes the layout of the
3023 system's STREAM objects.
3024 * The Gray subclassable streams extension now works, thanks to a
3025 patch from Martin Atzmueller.
3026 * The full LOAD-FOREIGN extension (not just the primitive
3027 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3028 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3029 but like most other programs, it defaults to copying the Unix
3030 environment from the original process instead of starting the
3031 new process in an empty environment.
3032 * Extensions which manipulate the Unix environment now support
3033 an :ENVIRONMENT keyword option which doesn't smash case or
3034 do other bad things. The CMU-CL-style :ENV option is retained
3035 for porting convenience.
3036 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3037 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3039 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3041 * A patch from Martin Atzmueller seems to have solved the SIGINT
3042 problem, and as far as we know, signal-handling now works cleanly.
3043 (If you find any new bugs, please report them!)
3044 * The system no longer defaults Lisp source file names to types
3045 ".l", ".cl", or ".lsp", but only to ".lisp".
3046 * The compiler no longer uses special default file extensions for
3047 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3048 seems to expect a single default extension for all compiled code,
3049 and there's no compelling reason to try to stretch the standard
3050 to allow two different extensions.) Instead, byte-compiled files
3051 default to the same extension as native-compiled files.
3052 * Fasl file format version numbers have increased again, because
3053 a rearrangement of internal implementation packages made some
3054 dumped symbols in old fasl files unreadable in new cores.
3055 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3056 many fewer weird special cases.
3057 * Bug #17 (differing COMPILE-FILE behavior between logical and
3058 physical pathnames) has been fixed, and some related misbehavior too,
3059 thanks to a patch from Martin Atzmueller.
3060 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3061 by Tim Moore, ported to SBCL by Martin Atzmueller.
3062 * Martin Atzmueller fixed several filesystem-related problems,
3063 including bug #36, in part by porting CMU CL patches, which were
3064 written in part by Paul Werkowski.
3065 * More compiler warnings in src/runtime/ are gone, thanks to
3066 more patches from Martin Atzmueller.
3067 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3070 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3072 * DESCRIBE now works on CONDITION objects.
3073 * The debugger now handles errors which arise when trying to print
3074 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3076 * The build system now uses an additional file, customize-target-features.lisp,
3077 to allow local modifications to the target *FEATURES* list. (The point of
3078 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3079 debugging features enabled, without having to worry about propagating it
3080 into everyone's system when I do a "cvs update".) When no
3081 customize-target-features.lisp file exists, the target *FEATURES* list
3082 should be constructed the same way as before.
3083 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3084 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3085 warn about behavior which is undefined under the ANSI spec. Note: This
3086 is specified by ANSI, but it's not very popular with programmers.
3087 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3088 macro in the SBCL sources for an example of a workaround which you
3089 might use to make portable ANSI-standard code which does what you want.
3090 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3091 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3092 when it's used as a non-toplevel form. (This is required in order
3093 to implement the DEFCONSTANT-EQX macro.)
3094 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3095 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3096 funny things, due to bugs in the implementation of EVAL-WHEN
3097 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3098 non-toplevel DEFCONSTANTs are uncommon.)
3099 * The core file version number and fasl file version number have been
3100 incremented, because the old noncompliant DEFCONSTANT behavior involved
3101 calling functions which no longer exist, and because I also took the
3102 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3103 * fixed bug 1 (error handling before read-eval-print loop starts), and
3104 redid debugger restarts and related debugger commands somewhat while
3106 ** The QUIT debugger command is gone, since it did something
3107 rather different than the SB-EXT:QUIT command, and since it never
3108 worked properly outside the main toplevel read/eval/print loop.
3109 Invoking the new TOPLEVEL restart provides the same functionality.
3110 ** The GO debugger command is also gone, since you can just invoke
3111 the CONTINUE restart directly instead.
3112 ** The TOP debugger command is also gone, since it's redundant with the
3113 FRAME 0 command, and since it interfered with abbreviations for the
3115 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3117 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3118 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3119 some porting work by Martin Atzmueller)
3120 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3121 with a boatload of patches which he ported to SBCL. Now that those
3123 ** The system tries to make sure that its low-priority messages
3124 are prefixed by semicolons, to help people who like to use
3125 syntax highlighting in their ILISP buffer. (This patch
3126 was originally due to Raymond Toy.)
3127 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3128 patches originally written by Raymond Toy.
3129 ** The compiler understands coercion between single-value and
3130 multiple-VALUES type expressions better, getting rid of some very
3131 weird behavior, thanks to patches originally by Robert MacLachlan
3132 and Douglas Crosher.
3133 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3134 lambda lists, thanks to a patch originally by Pierre Mai.
3135 ** The system no longer bogusly warns about "abbreviated type
3137 ** The compiler gets less confused by inlining and RETURN-FROM,
3138 thanks to some patches originally by Tim Moore.
3139 ** The system no longer hangs when dumping circular lists to fasl
3140 files, thanks to a patch originally from Douglas Crosher.
3141 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3142 undefined function error.
3143 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3144 in structure slot initforms) for now, documented workaround instead:-|
3145 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3146 * fixed bug 5: added stubs for various Gray stream functions called
3147 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3148 installed, at least appropriate type errors are generated
3149 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3150 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3151 consistently in DEFMETHOD forms.
3152 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3153 it doesn't seem to affect SBCL after all
3154 * The C runtime system now builds with better optimization and many
3155 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3157 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3159 * The system is now under CVS at SourceForge (instead of the
3160 CVS repository on my home machine).
3161 * The new signal handling code has been tweaked to treat register
3162 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3163 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3164 downstream, e.g. in debug-int.lisp, has implicit dependencies
3165 on the unsignedness of integer representation of machine words,
3166 and that caused the system to bomb out with infinite regress
3167 when trying to recover from type errors involving signed values,
3168 e.g. (BUTLAST '(1 2 3) -1).)
3169 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3170 (This was one of the bugs Peter Van Eynde reported back in July.)
3171 * The system now uses code inspired by Colin Walters' O(N)
3172 implementation of MAP (from the cmucl-imp@cons.org mailing
3173 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3174 inline the MAP operation, and there is more than one
3175 sequence argument to the MAP call (so that it can't just
3176 do ETYPECASE once and for all based on the type of the
3177 single sequence argument). (The old non-inline implementation
3178 of the general M-argument sequence-of-length-N case required
3179 O(M*N*N) time when any of the sequence arguments were LISTs.)
3180 * The QUIT :UNIX-CODE keyword argument has been renamed to
3181 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3182 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3184 * Raymond Wiker's patches to port dynamic loading from Linux to
3185 FreeBSD have been added.
3186 * The BUGS file is now more nearly up to date, thanks in large part
3187 to Martin Atzmueller's review of it.
3188 * The debugger now flushes standard output streams before it begins
3189 its output ("debugger invoked" and so forth).
3190 * The core version number and fasl file version number have both
3191 been incremented, because of incompatible changes in the layout
3193 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3194 used to not be called for a saved Lisp image.)
3195 * Martin Atzmueller's version of a patch to fix a compiler crash,
3196 as posted on sbcl-devel 13 September 2000, has been installed.
3197 * Instead of installing Martin Atzmueller's patch for the
3198 compiler transform for SUBSEQ, I deleted the compiler transform,
3199 and transforms for some similar consing operations.
3200 * A bug in signal handling which kept TRACE from working on OpenBSD
3202 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3203 away by constant folding
3204 * The system now defines its address space constants in one place
3205 (in the Lisp sources), and propagates them automatically elsewhere
3206 (through GENESIS and the sbcl.h file). Therefore, patching the
3207 address map is less unnecessarily tedious and error-prone. The
3208 Lisp names of address space constants have also been systematized.
3209 * CVS tags like dollar-Header-dollar have been removed from
3210 the sources, because they have never saved me trouble and
3211 they've been source of trouble working with patches and other
3212 diff-related operations.
3213 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3214 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3216 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3218 * The system has been ported to OpenBSD.
3219 * The system now compiles with a simple "sh make.sh" on the systems
3220 that it's supported on. I.e., now you no longer need to tweak
3221 text in the target-features.lisp-expr and symlinks in src/runtime/
3222 by hand, the make.sh takes care of it for you.
3223 * The system is no longer so grossly inefficient when compiling code
3224 involving vectors implemented as general (not simple) vectors (VECTOR T),
3225 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3226 which dares to use the various sequence functions on non-simple
3227 vectors, takes less of a performance hit.
3228 * There is now a primitive type predicate VECTOR-T-P
3229 to test for the (VECTOR T) type, so that e.g.
3230 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3231 can now be compiled with some semblance of efficiency. (The old code
3232 turned the type declaration into a full call to %TYPEP at runtime!)
3233 * AREF on (VECTOR T) is still not fast, since it's still compiled
3234 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3235 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3236 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3237 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3238 without making full calls to SUBTYPEP at runtime.
3239 (Some analogous efficiency issues for non-simple vectors specialized to
3240 element types other than T, or for non-simple multidimensional arrays,
3241 have not been addressed. They could almost certainly be handled the
3242 same way if anyone is motivated to do so.)
3243 * The changes in array handling break binary compatibility, so
3244 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3245 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3246 instead of (VALUES T T).
3247 * By following the instructions that Dan Barlow posted to sbcl-devel
3248 on 2 July 2000, I was able to enable primitive dynamic object
3249 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3250 functionality is not implemented (since it calls ld to resolve
3251 library references automatically, requiring RUN-PROGRAM for its
3252 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3253 try to resolve library references) is now supported.
3254 * The system now flushes the standard output streams when it terminates,
3255 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3256 them at several other probably-convenient times, e.g. in each pass of
3257 the toplevel read-eval-print loop, and after evaluating a form given
3258 as an "--eval" command-line option. (These changes were motivated by a
3259 discussion of stream flushing issues on cmucl-imp in August 2000.)
3260 * The source transform for TYPEP of array types no longer assumes
3261 that an array whose element type is a not-yet-defined type
3262 is implemented as an array of T, but instead punts, so that the
3263 type will be interpreted at runtime.
3264 * There is now some support for cross-compiling in make.sh: each of
3265 the phases of make.sh has its own script. (This should be transparent
3266 to people doing ordinary, non-cross-compile builds.)
3267 * Since my laptop doesn't have hundreds of megabytes of memory like
3268 my desktop machine, I became more motivated to do some items on
3269 my to-do list in order to reduce the size of the system a little:
3270 ** Arrange for various needed-only-at-cold-init things to be
3271 uninterned after cold init. To support this, those things have
3272 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3273 symbols with such names are now uninterned after cold init).
3274 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3275 fasl files for cold load.
3276 ** Remove the old compiler structure pooling code (which used to
3277 be conditional on the target feature :SB-ALLOC) completely.
3278 ** Redo the representation of some data in cold init to be more compact.
3279 (I also looked into supporting byte compiled code at bootstrap time,
3280 which would probably reduce the size of the system a lot, but that
3281 looked too complicated, so I punted for now.)
3282 * The maximum signal nesting depth in the src/runtime/ support code has
3283 been reduced from 4096 to 256. (I don't know any reason for the very
3284 large old value. If the new smaller value turns out to break something,
3285 I'll probably just bump it back up.)
3286 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3288 * Many, many bugs reported by Peter Van Eynde have been added to
3289 the BUGS list; some have even been fixed.
3290 * While enabling dynamic object file loading, I tried to make the
3291 code easier to understand, renaming various functions and variables
3292 with less ambiguous names, and changing some function calling
3293 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3294 * While trying to figure out how to do the OpenBSD port, I tried to
3295 clean up some of the code in src/runtime/. In particular, I dropped
3296 support for non-POSIX signal handling, added various comments,
3297 tweaked the code to reduce the number of compilation warnings, and
3298 renamed some files to increase consistency.
3299 * To support the new automatic configuration functionality in make.sh,
3300 the source file target-features.lisp-expr has been replaced with the
3301 source file base-target-features.lisp-expr and the machine-generated
3302 file local-target-features.lisp-expr.
3303 * fixed a stupid quoting error in make.sh so that using CMU CL
3304 "lisp -batch" as cross-compilation host works again
3306 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3308 * DESCRIBE no longer tries to call itself recursively to describe
3309 bound/fbound values, so that it no longer fails on symbols which are
3310 bound to themselves (like keywords, T, and NIL).
3311 * DESCRIBE now works on generic functions.
3312 * The printer now prints less-screwed-up representations of closures
3313 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3314 * A private symbol is used instead of the :EMPTY keyword previously
3315 used to mark empty slots in hash tables. Thus
3316 (DEFVAR *HT* (MAKE-HASH-TABLE))
3317 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3318 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3319 now does what ANSI says that it should. (You can still get
3320 similar noncompliant behavior if bang on the hash table
3321 implementation with all the symbols you get back from
3322 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3323 This breaks binary compatibility, since tests for equality to
3324 :EMPTY are wired into things like the macroexpansion of
3325 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3327 * There's now a minimal placeholder implementation for CL:STEP,
3328 as required by ANSI.
3329 * An obscure bug in the interaction of the normal compiler, the byte
3330 compiler, inlining, and structure predicates has been patched
3331 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3332 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3334 * Missing ordinary arguments in a macro call are now detected even
3335 when the macro lambda list contains &KEY or &REST.
3336 * The debugger no longer complains about encountering the top of the
3337 stack when you type "FRAME 0" to explicitly instruct it to go to
3338 the top of the stack. And it now prints the frame you request even
3339 if it's the current frame (instead of saying "You are here.").
3340 * As specified by ANSI, the system now always prints keywords
3341 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3343 * The default initial SIZE of HASH-TABLEs is now smaller.
3344 * Type information from CLOS class dispatch is now propagated
3345 into DEFMETHOD bodies, so that e.g.
3346 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3348 is now basically equivalent to
3349 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3350 (DECLARE (TYPE SINGLE-FLOAT X))
3352 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3353 operation, without having to do run-time type dispatch.
3354 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3355 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3356 * Since I don't seem to be making big file reorganizations very often
3357 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3358 is overflowing my ability to conveniently back them up), I've finally
3359 checked the system into CVS. (The CVS repository is on my home system,
3360 not at SourceForge -- putting it on SourceForge might come later.)
3361 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3362 high-level GC-NOTIFY-FOO functions send their output. (There's
3363 still very little control of where low-level verbose GC functions
3364 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3365 less than it used to -- the GC-NOTIFY-FOO functions are now under
3366 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3367 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3368 in only one place in the source code, and propagates it automatically
3369 everywhere that it's needed. Thus e.g. when I bump the version from
3370 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3371 * The C source files now include boilerplate legalese and documentation
3372 at the head of each file (just as the Lisp source files already did).
3373 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3374 to his page will be replaced with a link to his new CLiki service.
3376 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3378 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3379 * The build process now looks for GNU make under the default name "gmake",
3380 instead of "make" as it used to. If GNU make is not available as "gmake"
3381 on your system, you can change this default behavior by setting the
3382 GNUMAKE environment variable.
3383 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3384 can build without error under CMU CL.
3386 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3388 * There is now a partial SBCL user manual (with some new text and some
3389 text cribbed from the CMU CL manual).
3390 * The beginnings of a profiler have been added (starting with the
3391 CMU CL profiler and simplifying and cleaning up). Eventually the
3392 main interface should be through the TRACE macro, but for now,
3393 it's still accessed through vaguely CMU-CL-style functions and macros
3394 exported from the package SB-PROFILE.
3395 * Some problems left over from porting CMU CL to the new
3396 cross-compilation bootstrap process have been cleaned up:
3397 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3398 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3399 part when getting rid of the compiler *BACKEND* stuff.)
3400 ** The value of *NULL-TYPE* was screwed up, because it was
3401 being initialized before the type system knew the final
3402 definition of the 'NULL type. This screwed up several key
3403 optimizations in the compiler, causing inefficiency in all sorts
3404 of places. (I found it because I wanted to understand why
3405 GET-INTERNAL-RUN-TIME was consing.)
3406 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3407 PROCLAIM FTYPE information. Unfortunately this broke binary
3408 compatibility again, since now the forms output by DEFGENERIC
3409 to refer to functions which didn't exist in 0.6.3.
3410 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3411 can use the new (as of 0.6.3) transform for SOME into MAP into
3413 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3414 values to (INTEGER 0 1000000), so that the time code will no longer
3415 occasionally get blown up by Linux returning 1000000 microseconds
3416 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3417 its output conform to the ANSI spec. (Alas, this makes its output
3418 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3419 * A full call to MAP NIL with a single sequence argument no longer conses.
3420 * fixes to problems pointed out by Martin Atzmueller:
3421 * The manual page no longer talks about multiprocessing as though
3422 it were currently supported.
3423 * The ILISP support patches have been removed from the distribution,
3424 because as of version 5.10.1, ILISP now supports SBCL without us
3425 having to maintain patches.
3426 * added a modified version of Raymond Toy's recent CMU CL patch for
3427 EQUALP comparison of HASH-TABLE
3429 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3431 * The system still can't cross-compile itself with
3432 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3433 entails), but at least it can compile more of itself that way
3434 than it used to be able to, and various buglets which were uncovered
3435 by trying to cross-compile itself that way have now been fixed.
3436 * This release breaks binary compatibility again. This time
3437 at least I've incremented the FASL file format version to 2, so that the
3438 problem can be detected reliably instead of just causing weird errors.
3439 * various new style warnings:
3440 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3441 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3442 ** using the lexical binding of a variable named in the *FOO* style
3443 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3444 as specified by ANSI.
3445 * *RANDOM-STATE* is no longer automatically initialized from
3446 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3447 default behavior of the system is to repeat its behavior every time
3448 it's run. If you'd like to change this behavior, you can always
3449 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3450 old convention there was no comparably easy way to get the system to
3451 repeat its behavior every time it was run.
3452 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3453 FTYPE declarations has been removed, in favor of their ANSI
3454 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3455 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3456 the types of their sequence arguments can be determined at compile time.
3457 This is done through a new open code expansion for MAP which eliminates
3458 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3459 argument types can be determined at compile time.
3460 * The optimizer now transforms COERCE into an identity operation when it
3461 can prove that the coerced object is already of the correct type. (This
3462 can be a win for machine generated code, including the output of other
3463 optimization transforms, such as the MAP transform above.)
3464 * Credit information has been moved from source file headers into CREDITS.
3465 * Source file headers have been made more standard.
3466 * The CASE macro now compiles without complaining even when it has
3469 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3471 * (Note that the way that the PCL macroexpansions were rewritten
3472 to accommodate the change in DEFGENERIC below breaks binary
3473 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3474 not run under sbcl-0.6.2. Once we get out of alpha releases,
3475 i.e. hit release 1.0.0, we'll probably try to maintain binary
3476 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3477 and sbcl-1.4.4. Until then, however, it might be fairly common
3478 for maintenance releases to break binary compatibility.)
3479 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3481 * The reporting of unhandled signals has been changed to print some
3482 explanatory text as well as the report form. (Previously only
3483 the report form was printed.)
3484 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3485 it defines, so that the compiler no longer issues undefined function
3486 warnings for compiled-but-not-yet-loaded generic functions.
3487 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3488 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3489 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3490 section of the ANSI spec are supported.
3491 * The "" nickname for the "KEYWORD" package has been removed.
3492 The reader still handles symbol tokens which begin with a package marker
3493 as keywords, but it doesn't expose its mechanism for doing so in the
3494 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3495 * The system now issues STYLE-WARNINGs for contradictory TYPE
3496 proclamations. (Warnings for contradictory FTYPE proclamations would
3497 be nice too, but those can't be done usefully unless the type system
3498 is made smarter about FUNCTION types.)
3499 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3500 systematized, so that "*target-*.lisp is supposed to exist only on the
3501 target and imply that there's a related file which exists on the
3502 host, and *host-*.lisp is supposed to exist only on the host and imply
3503 that there's a related file which exists on the target. This involves a
3504 lot of renaming. Hopefully the acute confusion caused by the renaming
3505 will be justified by the reduction in chronic confusion..
3506 ** runtime-type.lisp -> early-target-type.lisp
3507 ** target-type.lisp -> late-target-type.lisp
3508 ** early-host-format.lisp -> early-format.lisp
3509 ** late-host-format.lisp -> late-format.lisp
3510 ** host-error.lisp -> misc-error.lisp
3511 ** early-error.lisp -> early-target-error.lisp
3512 ** late-error.lisp -> late-target-error.lisp
3513 ** host-defboot.lisp -> early-defboot.lisp
3514 ** code/misc.lisp -> code/target-misc.lisp
3515 ** code/host-misc.lisp -> code/misc.lisp
3516 ** code/numbers.lisp -> code/target-numbers.lisp
3517 ** code/early-numbers.lisp -> numbers.lisp
3518 ** early-host-type.lisp -> early-type.lisp
3519 ** late-host-type.lisp -> late-type.lisp
3520 ** host-typep.lisp -> typep.lisp
3521 ** load.lisp -> target-load.lisp
3522 ** host-load.lisp -> load.lisp
3523 ** host-disassem.lisp -> disassem.lisp
3524 ** host-insts.lisp -> insts.lisp
3525 ** byte-comp.lisp -> target-byte-comp.lisp
3526 ** host-byte-comp.lisp -> byte-comp.lisp
3527 ** host-signal.lisp -> signal.lisp
3528 ** host-defstruct.lisp -> defstruct.lisp
3529 ** late-target-type.lisp -> deftypes-for-target.lisp
3530 Furthermore, several other previously target-only files foo.lisp (e.g.
3531 hash-table.lisp and random.lisp) have been split into a target-and-host
3532 foo.lisp file and a target-only target-foo.lisp file, with their key type
3533 definitions in the target-and-host part, so that the cross-compiler will
3534 know more about target types.
3535 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3536 gone away. In their place are various *BACKEND-FOO* variables
3537 corresponding to the slots of the old structure.
3538 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3539 into the target SBCL has been fixed.
3540 * The chill.lisp system for loading cold code into a running SBCL
3542 * Support for the CMU CL "scavenger hook" extension has been removed.
3543 (It was undocumented and unused in the CMU CL sources that SBCL was
3544 derived from, and stale in sbcl-0.6.1.)
3545 * Various errors in the cross-compiler type system were detected
3546 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3547 (enabling various consistency checks). Many of them were fixed,
3548 but some hard problems remain, so the compiler is back to
3549 running without *TYPE-SYSTEM-INITIALIZED* for now.
3550 * As part of the cross-compiler type system cleanup, I implemented
3551 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3552 * I have started adding UNCROSS calls throughout the type system
3553 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3554 on cross-compiler input files will be able to go away, and various
3556 * CONSTANTP now returns true for quoted forms (as explicitly required
3559 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3561 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3562 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3563 (MAKE-LIST -1). In the longer term, it ought to become true
3564 that declarations are assertions even at SAFETY 1. For now, it's not
3565 quite true even at SAFETY 3, but it's at least more nearly true..
3566 (Note that this change seems to increases the size of the system by
3567 O(5%) and to decrease the speed of the compiler by 20% or more.)
3568 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3569 for the problem of printing dozens of lines of distracting information
3570 about low-level system machinery as part of the top stack frame
3571 on entry to the debugger when an undefined function was called.
3572 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3573 is not reset to COMMON-LISP-USER.
3574 * Compilation of stuff related to dyncount.lisp has been made conditional
3575 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3576 smaller. The various dyncount-related symbols have been moved into
3577 a new "SB-DYNCOUNT" package.
3578 * tty-inspect.lisp has been renamed to inspect.lisp.
3579 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3580 feature has gone away. (When we eventually port to other flavors of
3581 libc and/or Unix, we'll try to make the differences between flavors
3582 invisible at the user level.)
3583 * Various other *FEATURES* tags, and/or their associated conditionals,
3584 have been removed if obsolescent, or given better documentation, or
3585 sometimes given more-mnemonic names.
3587 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3589 * tidied up "make.sh" script
3590 * tidied up system directory structure
3591 * better "clean.sh" behavior
3592 * added doc/FOR-CMUCL-DEVELOPERS
3593 * many many small tweaks to output format, e.g. removing possibly-confusing
3594 trailing #\. character in DESCRIBE-INSTANCE
3595 * (EQUALP #\A 'A) no longer signals an error.
3596 * new hashing code, including EQUALP hashing
3597 * tidied up Lisp initialization and toplevel
3598 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3599 * command line argument processing
3600 * added POSIX-GETENV function to deal with Unix-ish environment variables
3601 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3602 terminating SBCL on EOF
3603 * non-verbose GC by default
3604 * There is no more "sbcl" shell script; the sbcl file is now the C
3605 runtime executable (just like CMU CL).
3606 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3608 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3609 symbol and package information as Lisp native symbol and package objects
3610 instead of strings naming symbols and strings naming packages. This way,
3611 whenever packages are renamed (as in warm init), debug information is
3612 transformed along with everything else.
3613 * tweaked the optimization policy declarations which control the building
3614 of SBCL itself. Now, among other things, the system no longer saves
3615 source location debugging information. (This helps two problems at once
3616 by reducing SBCL size and by keeping SBCL from trying to look for its
3617 sources -- which may not exist -- when reporting errors.)
3618 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3619 debugging and testing purposes
3620 * cleaned up printing, making the printer call PRINT-OBJECT for
3621 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3622 methods, giving nearly-ANSI behavior
3623 * converted almost all special variables to use *FOO* naming convention
3624 * deleted PARSE-TIME functionality, since it can be done portably
3625 * moved some files out of cold init into warm init
3626 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3628 * regularized formatting of source files
3629 * added an install.sh script
3630 * fixed ridiculous memory usage of cross-compiler by making
3631 compiler/alloc.lisp not try to do pooling unless it can hook
3632 itself into the GC of the cross-compilation host. Now the system
3633 builds nicely on my old laptop.
3634 * added :SB-ALLOC in target-features.lisp-expr
3635 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3636 was not implemented)
3637 * re-did condition handling and note reporting in the compiler. Notes
3638 are no longer handled by signalling conditions. Style warnings
3639 and warnings are handled more correctly and reported in such a way
3640 that it's easy to find one or the other in your output (so that you
3641 can e.g. figure out which of many problems caused COMPILE-FILE to
3643 * changed the severity of several compiler warnings from full WARNING
3644 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3645 compiler note reporting so that it doesn't use the condition system
3646 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3647 COMPILE-FILE command)
3648 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3649 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3650 no longer supported, since their implementation was deeply intertwingled
3651 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3652 * removed bogus "support" for compiler macros named (SETF FOO), and
3653 removed the compiler macro for SETF INFO (but only after making a fool
3654 of myself on the cmucl-imp mailing list by posting a bogus patch for
3655 DEFINE-COMPILER-MACRO..)
3656 * Compiled files containing forms which have side effects on the Lisp
3657 reader (such as DEFPACKAGE forms) are now handled more correctly.
3658 (Compiler queuing of top level lambdas has been suppressed by setting
3659 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3660 * deleted various currently-unused source files, e.g. gengc.lisp. They
3661 may be added back at some point e.g. when porting to other architectures,
3662 but until they are it's distracting to distribute them and to try to
3664 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3665 there were so many of them they're just distractions, and UNCROSS is
3666 known to be able to handle the current sources
3667 * moved PROFILE functionality into TRACE, so that it will be clear
3668 how the wrapping and unwrapping of functions when you profile them
3669 interacts with the wrapping and unwrapping of functions when you
3670 trace them. (Actually, the functionality isn't there yet, but at least
3671 the interface specification is there. Hopefully, the functionality will
3672 arrive with some future maintenance release.)
3673 * removed host-oops.lisp
3674 * changed signature of QUIT function to allow UNIX-CODE argument
3675 * fixed READ-SEQUENCE bug
3676 * tweaked verbose GC output so that it looks more like the progress
3677 output that ANSI specifies for functions like LOAD
3678 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3679 * added <http://sbcl.sourceforge.com> to the banner information printed by