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 * enhancement: CONSTANTP is now able to determine constantness of
25 more complex forms, including calls to constant-foldable standardized
26 functions and some special forms beyond QUOTE.
27 * fixed bug: occasional GC crashes on Solaris/x86
28 * optimization: x86-64 supports stack allocation of results of simple
29 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
30 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
31 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
32 (thanks to James Knight)
33 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
34 by the standard (thanks to Stephen Compall)
36 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
37 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
38 be used for bundling the runtime and the core file into one
39 executable binary. This feature is not currently supported on all SBCL
40 platforms. (thanks to James Bielman and NIIMI Satoshi)
41 * new feature: a generational or ephemeral garbage collector is now
42 the default on the PowerPC platform (both Linux and Darwin). The
43 old Cheney (stop and copy) collector is a build-time option.
44 (thanks to Cyrus Harmon, after Raymond Toy)
45 * minor incompatible change: the method by which SBCL finds its
46 contributed modules has changed; it no longer relies on symbolic
47 links from an $SBCL_HOME/systems directory, but searches directly
48 in the subdirectories of $SBCL_HOME.
49 * enhancement: the dynamic heap size on the Linux/PPC platform is
50 markedly larger, even using the older Cheney garbage collector.
51 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
52 certain complicated string types. (reported by Gary King)
53 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
54 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
55 * fixed bug: DOCUMENTATION on structures no longer signals an
56 error if no structure type of the right name exists. (reported by
58 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
59 applicable non-standard primary or :AROUND method on
60 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
61 no longer cause unbound variable errors. (reported by Kevin Reid)
62 * fixed bug: metacircle resolution in cases where methods have slots
63 added before the slots from STANDARD-METHOD. (reported by Jean
65 * fixed bug: the Power PC assembly code for calling into lisp
66 sometimes computed its return address wrongly (depending on the
67 15th bit of the address, fixed at link time). This used to
68 manifest itself in a segmentation violation while building PCL.
69 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
70 many others over the years)
71 * fixed bug: ROOM no longer reports silly numbers for stack usage if
72 the stack pointer (treated as a signed integer) is negative.
73 (thanks to Peter van Eynde)
75 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
76 * new platform: experimental support for the Windows operating
77 system has been added. (thanks to Alastair Bridgewater)
78 * fixed several bugs in and robustified the PPC FFI (including
79 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
80 * bug fix: allow non-simple string symbol names (reported by Paul
82 * bug fix: interrupt handling on NetBSD (thanks to Richard M
84 * bug fix: saving a core corrupted callbacks on x86/x86-64
85 * bug fix: closed a loophole in metacircularity detection and
86 grounding in the PCL implementation of CLOS.
87 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
88 (reported by tomppa on #lisp)
89 * bug fix: clear the direction flag on Lisp -> C transitions, as
90 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
91 (reported by Andrej Grozin and Hendrik Maryns)
92 * optimization: major improvements to GC efficiency on GENCGC platforms
93 * optimization: faster implementation of EQUAL
94 * optimization: emit more efficient opcodes for some common
95 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
96 * optimization: several other minor code-generation improvements on x86-64
97 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
99 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
100 * minor incompatible change: (SETF CLASS-NAME) and (SETF
101 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
102 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
103 by ANSI as a generic function, and for consistency (SETF
104 GENERIC-FUNCTION-NAME) is treated likewise.
105 * fixed bug #233.b: make constraint propagation notice when a variable
106 value is changed after it is referenced but before it is used
107 * fixed bug #296: no more arbitrary behaviour from filenames with
108 odd characters as --load arguments. (reported by Adam Warner)
109 * fixed bug #390: :CHARACTER-SET pathname components now work as
110 expected. (reported by Tim Daly Jr)
111 * fixed bug #391: complicated :TYPE intersections in slot
112 definitions no longer cause an error in PCL internals.
113 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
114 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
115 returning the number of octets which would be written to the
116 file-stream. (thanks to Robert J. Macomber)
117 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
118 arguments to the internal stream functions. (thanks to David
120 * bug fix: stack exhaustion is now handled on the x86/SunOS
121 platform. (thanks to Daisuke Homma)
122 * bug fix: more accurate ROOM results on GENCGC platforms
123 * optimization: improved type inference for arithmetic-for
124 index variables in LOOP
125 * optimization: faster floating-point SQRT on x86-64
126 * fixed some bugs revealed by Paul Dietz' test suite:
127 ** DOCUMENTATION returns NIL instead of "" for method combinations
128 that don't have a docstring
130 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
131 * minor incompatible change: (SETF CLASS-NAME) and (SETF
132 GENERIC-FUNCTION-NAME) are no longer generic functions, and
133 therefore are not customizeable by user code (as seems to be at
134 least permitted and maybe required by AMOP). As a consolation,
135 however, the SBCL implementation of these functions now calls
136 REINITIALIZE-INSTANCE as specified by AMOP.
137 * bug fix: slot-definition documentation is propagated as per ANSI
138 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
139 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
140 argument, as do eslotd initialization methods. (from Pascal
141 Costanza's "Closer" project)
142 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
143 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
145 * bug fix: it is now possible to have more than one subclass of
146 STANDARD-GENERIC-FUNCTION without causing stack overflow.
147 (reported by Bruno Haible, Pascal Costanza and others)
148 * bug fix: the dependent update protocol now works for generic
149 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
151 * bug fix: condition-class instances corresponding to
152 DEFINE-CONDITION forms are now created eagerly. (reported by
153 Kalle Olavi Niemitalo on comp.lang.lisp)
154 * bug fix: floating point printing is more accurate in some
155 circumstances. (thanks to Simon Alexander)
156 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
157 merged with *DEFAULT-PATHNAME-DEFAULTS*.
158 * bug fix: callbacks on OS X now preserve stack-alignment.
159 (thanks to Cyrus Harmon)
160 * enhancement: the x86-64 disassembler is much better at
161 disassembling SSE instructions. (thanks to Lutz Euler)
162 * enhancement: CHECK-TYPE on a variable now allows type-inference
163 (thanks to Vincent Arkesteijn)
164 * optimization: improved performance of EUC-JP external format.
165 (thanks to NIIMI Satoshi)
166 * optimization: performance improvements to IO on file streams of
167 :ELEMENT-TYPE CHARACTER
168 * optimization: much faster memory allocation on x86-64
169 * optimization: faster garbage collections (latency and throughput)
171 * optimization: faster bignum subtraction and division on x86 and x86-64
172 (thanks to Lutz Euler)
173 * optimization: more accurate type-inference for loops that use a
174 floating point index variable or a negative step.
176 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
177 * bug fix: add a workaround to SBCL looping infinitely at startup on
178 Linux kernels with apparently buggy implementations of personality().
179 (thanks to Svein Ove Aas)
180 * bug fix: Unicode symbols are correctly printed in LDB backtraces
181 (thanks to David Lichteblau)
182 * bug fix: local bindings shadow global symbol macros for
183 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
184 * optimization: non-open coded uses of numeric comparison operators
185 (e.g. >) no longer cons when called with more than one parameter
186 on platforms supporting dynamic-extent allocation.
187 * enhancement: saving cores with foreign code loaded is now supported
188 on MIPS/Linux in addition to the previously supported platforms.
189 * bug fix: FIND-RESTART now tests for activity, not applicability when given
190 a restart object as identifier. (reported by Helmut Eller for CMUCL)
191 * bug fix: division by zero in sb-sprof when no samples were collected
192 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
193 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
194 (thanks to Alexander Kjeldaas)
195 * new feature: ignoring the return values of destructive functions
196 like NREVERSE signals a compile-time style-warning.
197 (thanks to Kevin Reid)
198 * new feature: the alignment of alien structure fields can be
199 explicitly specified. (thanks to Cyrus Harmon)
200 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
201 and dump core on SIGQUIT
203 ** incompatible change: threads do not inherit values of specials
204 from their parents (see manual)
205 ** bug fix: threads stacks belonging to dead threads are freed by the
206 next exiting thread, no need to gc to collect thread stacks anymore
207 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
208 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
209 ** bug fix: dynamic variable and thread start related gc lossage
210 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
212 * fixed some bugs revealed by Paul Dietz' test suite:
213 ** SUBTYPEP is slightly more accurate on heinously complicated
214 CONS types where some of the members have uncertain (in the
215 NIL, NIL sense) type relationships to each other.
217 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
218 no longer purified unless :PURIFY T is explicitly specified.
219 ** Non-purified cores are significantly smaller than before
221 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
222 * new feature: timers based on Zach Beane's excellent timer package
223 * added support for the following external formats: koi8-u,
224 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
225 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
226 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
227 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
228 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
229 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
230 * incompatible change: a threaded SBCL will no longer revert to
231 non-threaded mode on non-NPTL systems, but refuse to start entirely.
232 * bug fix: interrupts are disabled until startup is complete; no
233 more sigsegvs when receiving a signal to soon
234 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
236 * bug fix: add a workaround for the memory randomization features in
237 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
238 workaround will only be in effect on systems with the proc filesystem
240 * bug fix: printing objects of type HASH-TABLE signals a
241 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
243 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
244 have been running for over 50 days. (reported by Gilbert Baumann)
245 * bug fix: the logic for getting names of functions gets less
246 confused when confronded with alternate-metaclass
247 funcallable-instances. (reported by Cyrus Harmon)
248 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
249 consistent, even on internal alternate-metaclass objects.
250 * bug fix: flush closure information collected by physical
251 environment analysis prepass before the main pass. (bug reported
253 * bug fix: compiler pack phase does not modify a hashtable, which is
254 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
255 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
256 may be expanded. (reported by Alexander <malishev> on c.l.l)
257 * bug fix: no more style-warnings when compiling code using LAST
258 under high-SPEED low-SPACE settings. (reported by David Wallin
260 * fixed some incompatibilities between SBCL's MOP and the MOP
262 ** the METAOBJECT class is now implemented;
263 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
264 STANDARD-OBJECT, as required;
265 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
266 now compatible (as required by VALIDATE-SUPERCLASS); there
267 remains a consistency requirement over the presence or absence
268 of the FUNCTION class in the superclass list of finalized
269 classes; see the manual for more details;
270 ** the :AROUND method for COMPUTE-SLOTS on
271 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
272 requested slot ordering.
274 ** bug fix: parent thread now can be gc'ed even with a live
276 ** bug fix: binding a special with PROGV to no value is not
277 the same as that symbol not having been bound (thanks to
279 ** bug fix: binding specials is thread safe (thanks to
281 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
282 and SB-SPROF do not die with 'no handler for signal XX in
283 interrupt_handle_now(..)' anymore
284 ** bug fix: WITH-TIMEOUT works with multiple running threads
285 * fixed some bugs revealed by Paul Dietz' test suite:
286 ** ENSURE-GENERIC-FUNCTION should take a method class object for
287 the :method-class keyword argument.
289 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
290 * new port: the Solaris operating system on x86 processors is now
291 mostly supported, though some rough edges in the environment
292 remain. (thanks to Daisuke Homma)
293 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
294 space for its heap. (thanks to Thiemo Seufer)
295 * optimization: SBCL on MIPS platforms now supports dynamic-extent
296 closures. (thanks to Thiemo Seufer)
297 * minor incompatible change: eof selects abort in the debugger.
298 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
299 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
301 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
302 deprecated, and will go away in a future revision of SBCL.
303 * minor incompatible change: GC-ON and GC-OFF are no longer
304 implemented with a counter, it does not matter how many times gc
305 is switched on or off
306 * bug fix: discriminating functions for generic function classes
307 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
308 make invalid assumptions about method precedence order. (reported
310 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
311 platforms. (thanks to Thiemo Seufer)
312 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
313 * bug fix: correct pathname printing with printer escaping is on.
314 (thanks to Kevin Reid)
315 * bug fix: complex VOP definitions in "user-space" no longer trigger
316 package locks. (reported by Zach Beane)
317 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
318 for funcallable-instance functions, and is no different from
320 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
321 characters in the current readtable. (reported by Nicholas Neuss)
322 * bug fix: SUBTYPEP on various CONS types returns more a more
323 accurate acknowledgment of its certainty.
324 * optimizations: REMOVE-DUPLICATES now runs in linear time on
325 lists in some cases. This partially fixes bug 384.
326 * flush all standard streams before prompting in the REPL and the
328 * bug fix: signal handling and triggering gc do not conflict
329 directly anymore, in particular a high frequency sb-sprof does
330 not prevent gc from running
331 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
332 approximation for timezone and DST information between the
333 universal time epoch and the smallest negative 32-bit time_t.
334 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
335 year 1899 so long as the encoded time is non-negative.
336 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
337 the correct answer for non-constant rotate amounts when performing
338 an inline 32-bit rotation.
340 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
341 there is only one thread in the session
342 ** bug fix: memory leak for streams created in one thread and
343 written to in another
344 ** bug fix: lockup when compiled with gcc4
345 ** bug fix: race that allows the gc to be triggered when gc is
347 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
348 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
350 ** bug fix: finished threads are gc'ed properly
351 * fixed some bugs revealed by Paul Dietz' test suite:
352 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
353 has arguments with a different set of applicable methods from
354 the orignal arguments.
355 ** The type error thrown by MAP now has a correct expected-type
357 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
358 name a compiled function.
359 ** Three MISC tests where a large bignum was improperly coerced to
360 a float that couldn't represent that bignum during type
361 derivation were fixed.
362 ** SUBTYPEP can now handle types involving the negation of a
363 list-form FUNCTION type.
364 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
365 as (COMPLEX (AND RATIO (NOT FIXNUM))).
366 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
368 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
369 * New feature: Experimental support for bivalent streams: streams
370 opened with :element-type :default now allow character and binary
371 (unsigned-byte 8) I/O
372 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
373 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
375 * bug fix: no more highly sporadic "couldn't check whether ~S is
376 readable" when reading a stream and an interrupt hits in the middle
377 of a select system call
378 * compiler better recognizes complex arrays (reported by Eduardo
380 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
381 for variables on SBCL built with linkage-tables. (reported by Luis
383 * various error reporting improvements.
384 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
385 (Thanks to Thiemo Seufer)
386 * bug fix: nested reader invokations work correctly
387 * bug fix: it is possible to have simultaneous references to foreign
388 code and foreign data with the same name.
390 ** added x86-64 support
391 ** incompatible change: the threading api now works with thread
392 objects instead of thread ids
393 ** bug fix: threads are protected from signals and interruption when
394 starting up or going down
395 ** bug fix: a race where an exiting thread could lose its stack to gc
396 ** bug fix: don't halt on infinite error in threads if possible
397 ** fixed numerous gc deadlocks introduced in the pthread merge
398 ** bug fix: fixed thread safety issues in read and print
399 ** bug fix: debugger doesn't hang on session lock if interrupted at
400 an inappropriate moment
401 ** bug fix: run-program is now thread safe(r)
402 ** bug fix: inner with-recursive-lock no longer releases the mutex
403 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
404 * fixed some bugs revealed by Paul Dietz' test suite:
405 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
406 EXPECTED-TYPE slots filled.
407 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
409 ** MAKE-CONDITION accepts classes as type-designators.
410 ** COMPILE may never return NIL.
411 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
412 range before calling Unix time functions
414 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
415 * numerous signal handling fixes to increase stability
416 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
417 * minor incompatible change: we now correctly canonize default
418 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
419 as per the MOP, rather than the historical (INITARG INITFUNCTION
420 INITFORM). (reported by Bruno Haible)
421 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
422 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
423 for more information.
424 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
425 pathname is a directory pathname.
426 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
427 * dynamic space size on PPC has been increased to 768Mb. (thanks to
429 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
430 :METACLASS argument in addition to a class name. (reported by
431 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
432 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
433 (reported by Bruno Haible; patch from Gerd Moellmann)
434 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
436 * bug fix: more cleanups to the floating point exception handling on
437 x86-64 (thanks to James Knight)
438 * bug fix: the compiler does not try to constant fold calls of
439 COERCE and BIT-* functions when they return freshly constructed
440 objects. (reported by Nikodemus Siivola and Paul Dietz)
441 * optimization: improved the MIPS versions of generic (in the
442 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
443 * optimization: direct conversion of (unsigned-byte 32) to floats on
444 the PowerPC platform.
445 * optimization: structure instances with raw slots now use less
446 memory, and probably show better memory locality. (thanks to
448 * optimization: DYNAMIC-EXTENT declarations for lists and closures
449 are treated as requests for stack allocation on the x86-64,
450 Alpha-32, PPC and SPARC platforms.
451 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
452 the wrong file descriptor; implementation of SOCKET-OPEN-P.
453 (thanks to Tony Martinez)
455 ** gcing a dead thread can no longer lead to lockups
456 ** threads block signals until they are set up properly
457 ** errno is no longer shared by threads
458 ** interrupt-thread restores the eflags register on x86
459 ** fixed some lockups due to gc/thread interaction
460 * fixed some bugs revealed by Paul Dietz' test suite:
461 ** invalid dotted lists no longer raise a read error when
463 ** signal an error if a symbol that names a declaration is used
464 as the name of a type, or vice versa
465 ** allow using the (declare (typespec var*)) abbreviation for
466 (declare (type typespec var*)) with all type specifiers
467 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
468 ** always return NIL from PPRINT-POP when OBJECT is NIL
469 ** don't signal errors when pretty-printing malformed LABELS,
470 FLET or MACROLET forms
471 ** declarations in a DOLIST body are also in scope for the
473 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
475 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
478 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
479 * fixed cross-compiler leakages that prevented building a 32-bit
480 target with a 64-bit host compiler.
481 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
482 opened with :IF-EXISTS :OVERWRITE.
483 * fixed bug 281: error for an invalid qualifier in a short-form method
484 combination method is not signalled until the faulty method is called.
485 * bug fix: iteration variable type inferrer failed to deal with open
486 intervals. (reported by Alan Shields)
487 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
488 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
489 or more runtime options were provided to the sbcl binary.
490 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
491 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
492 * optimization: there's now a fast-path for fixnum arguments in the
493 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
494 * optimization: the code generated on x86-64 is more compact thanks
495 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
496 * bug fix: floating-point exception handling now partly works on
497 x86-64. (Thanks to James Knight)
498 * improvement to the MIPS backend from Thiemo Seufer: C-style
499 64-bit long long arguments and return values to alien functions
501 * fixed some bugs revealed by Paul Dietz' test suite:
502 ** the type-error signalled from WARN has a filled-in DATUM slot.
503 ** the type-error required when a stream is not associated with
504 a file has the stream as its datum.
505 ** type-errors on single-floats on x86-64 no longer have
506 :INVALID-OBJECT as the datum
507 ** the type-errors signalled for invalid function names now have
508 a correct expected type
509 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
510 for structures defined with a :TYPE. Documentation strings for
511 typed structures are no longer immediately discarded
512 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
513 broadcast streams and synonym streams. FILE-LENGTH now also works
514 on broadcast streams.
516 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
517 * incompatible change: the --noprogrammer option, deprecated since
518 version 0.7.5, has been removed. Please use the equivalent
519 --disable-debugger option instead.
520 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
522 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
523 inoperational for a while now) has been completely removed.
524 * null lexical environments are now printed as #<NULL-LEXENV>,
525 significantly reducing the amount of clutter in typical backtraces.
526 * documentation on weak pointers, finalization, and after GC hooks
527 has been added to the manual.
528 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
529 as other specialized array types, is much faster. SUBSEQ and
530 COPY-SEQ on such arrays have also been sped up.
531 * optimization: EQL is now more efficient when at least other argument
532 is known to be of type (OR FIXNUM (NOT NUMBER)).
533 * fixed bug: the runtime is now less vulnerable to changes in the
534 size of the SBCL object on OS X, and virtual memory is reserved for
535 all spaces that need to be at a fixed address.
536 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
537 * fixed bug: finalizers and after GC hooks that cause consing are now
539 * fixed bug: compiler error messages and summaries are now printed to
540 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
541 * fixed inference of the upper bound of an iteration variable.
542 (reported by Rajat Datta).
543 * fixed bug 211e: calling local functions with duplicated constant
544 keyword argument no longer causes a bogus style warning about an
546 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
547 ftype declaration to be disregarded. (reported by Dave Roberts)
548 * fixed bug 373: caused by erronous compilation of references to alien
549 variables in the runtime on ppc/darwin.
550 * fixed bug 376: CONJUGATE type deriver.
551 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
552 (reported by Baughn on #lisp)
553 * a cross-compiler bug on non-x86 platforms has been identified and
554 fixed. (thanks to Bruno Haible)
555 * improvements to the MIPS runtime code for increased stability.
556 (thanks to Thiemo Seufer)
557 * increased the maximimum compact environment size to allow
558 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
559 * partial workaround for bug 135: don't name the function we're
560 calling for hairy cases of EVAL, so as not to accumulate one environment
561 entry per call to EVAL. (reported by Kevin Reid)
562 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
563 * optimization: type testing for non-vector arrays should be faster.
564 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
566 * bug fix: setting 31st element of a bit vector to zero did not work
568 * bug fix: redefining a class definition which failed due to a
569 previous accessor / function clash now works (but see BUGS entry
570 #380 for more problems in this area). (thanks to Zach Beane)
571 * the long form of DEFINE-METHOD-COMBINATION disables method group
572 checking when given a single method group with pattern *. (thanks
574 * on x86 compiler supports stack allocation of results of simple
575 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
576 * support for building and running on Mac OS X 10.4 "Tiger" has been added
577 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
578 * fixed some bugs related to Unicode integration:
579 ** the restarts for recovering from input and output encoding
580 errors only appear when there is in fact such an error to
582 * fixed some bugs revealed by Paul Dietz' test suite:
583 ** MISC.549 and similar: late transformation of unsafe type
584 assertions into derived types caused unexpected code
586 ** SCALE-FLOAT type deriver is less wrong.
587 ** type derivers for EXP, LOG and similar functions compute result
588 types for complex arguments better.
589 ** (MISC.563) CONJUGATE type deriver works for very restricted
591 ** out-of-line type testers for character strings are available.
592 ** EQUAL compiler transform understands specialness of objects
594 ** accessing double-floats stored on the stack now works on x86-64.
595 ** debugger internals could sometimes create invalid lispobjs,
596 resulting in GC crashes.
597 ** MISC.548: type check weakening can convert required type into
599 ** initialization forms for bindings are not in scope of free special
602 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
603 * incompatible change: thread support for non-NPTL systems has
604 been removed - locking is buggy and unreliable. A threaded
605 SBCL build will now warn at startup and refuse to create
606 new threads, unless futex support is detected in the kernel
607 * incompatible change: the top level REPL now has only an ABORT
608 restart associated with it, not TOPLEVEL and ABORT as it used to.
609 TOP and TOPLEVEL are now available as debugger commands for
610 returning to the top level.
611 * incompatible change: forms evaluated in the REPL now use the
612 global optimization policy.
613 * incompatible change: user- and system-initialization files are
614 no longer processed with LOAD, but by READ and EVAL; hence the
615 global optimization policy, startup package, readtable, etc,
617 * The .fasl file format number has been incremented because of
618 various incompatible changes.
619 * internal entry point details and argument counts no longer appear
620 in backtraces unless explicitly requested by setting
621 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
622 * built-in and standard functions no longer have names like "top
623 level local call to FOO".
624 * fixed bug 32: functions defined in non-null lexical environments
625 now have more legible printed representation
626 * fixed bug 33: functions defined in non-null lexical environemnts
627 are now more amenable to inspection by INSPECT.
628 * workaround for bug 354: XEPs no longer appear in backtraces unless
629 explicitly requested.
630 * fixed bug: receiving the signal which results from attempting to
631 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
632 SIGBUS on other *BSDs) on architectures where the C stack is also
633 the Lisp stack (x86 and x86-64) from foreign code no longer leads
634 to debugger confusion or wild instability. (reported by Cheuksan
636 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
637 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
638 (reported by Lutz Euler)
639 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
640 compiled code. (reported by Paul Dietz)
641 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
642 specialized parameter to a method to something that is not TYPEP
643 the specializer is now possible.
644 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
645 face of package deletion.
646 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
647 pathname. (reported by David Tolpin on c.l.l)
648 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
649 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
650 STRING-STREAM and FILE-STREAM, respectively.
651 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
652 than previously; in addition, its test suite is now run on build.
653 (thanks to Luke Gorrie)
654 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
655 * optimization: added a immediate representation for single-floats
657 * fixed some bugs revealed by Paul Dietz' test suite:
658 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
659 ** attempting to create a package with a colliding nickname causes
660 correctable errors to be signalled.
661 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
662 ** several x86-64 backend bugs related to sign-extension of immediate
665 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
666 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
667 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
669 * fixed loading of multiply forward-referenced layouts.
670 (thanks to Cheuksan Wang)
671 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
672 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
673 (see COMPILE-FILE documentation for details of :PRINT option)
674 * fixed bugs 19 and 317: fixed-format floating point printing is
675 more accurate. This also fixes a bug reported by Adam Warner
676 related to the ~@F format directive.
677 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
679 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
680 dispatch table if the from-char is a dispatch macro character.
681 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
682 by the wordsize no longer ignore the last word. (reported by Lutz
684 * fixed bug in type checking of dynamic-extent variables. (reported
686 * optimization: sequence traversal functions use their freedom to
687 coerce function designators to functions.
688 * optimization: code with many calls to CLOS methods specialized on
689 CLOS classes has had redundant type checks removed.
690 * optimization: added declarations to speed up operations that access
691 the internal character database (for example STRING-UPCASE)
692 * optimization: comparison operations between floats and sufficiently small
693 fixnums no longer create extra rationals
694 * fixed some bugs related to Unicode integration:
695 ** portions of multibyte characters at the end of buffers for
696 character-based file input are correctly transferred to the
697 start of the buffer at the next read.
698 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
699 passing it through to OPEN.
700 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
701 argument given to internal calls to OPEN.
702 ** the built-in comment readers (introduced by character sequences
703 ";" and "#|") are more forgiving to encoding errors; they will
704 STYLE-WARN and then attempt to resync the stream at a character
705 boundary. (thanks to Teemu Kalvas)
706 * fixed some bugs in the x86-64 port:
707 ** Negative short int return values from c-calls are sign-extended
709 ** The stack is aligned to 16-bytes for c-calls, as required by
710 the ABI. (thanks to Cheuksan Wang)
711 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
712 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
713 for enums. (thanks to Vincent Arkesteijn)
714 ** Multiple small optimizations and bugfixes for floating point
716 * fixed some bugs revealed by Paul Dietz' test suite:
717 ** Space, Tab, Linefeed, Return and Page have the invalid
718 secondary constituent character trait.
719 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
721 ** WITH-INPUT-FROM-STRING should only update the index place on
723 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
725 ** Bit-array operations (BIT-AND and similar) worked incorrectly
726 with one-dimensional arrays with fill pointers.
727 ** TYPE-OF failed on a complex with an integer realpart and a
729 ** compiler failure during type inference for the code of form
730 (IF (EQL X (THE ...)) ...) (MISC.535).
732 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
733 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
734 platform. The port supports 61-bit fixnums, large memory spaces
735 and reloading shared object files.
736 * enhancement: saving cores with foreign code loaded is now
737 supported on ppc/Darwin in addition to the previously supported
739 * enhancement: the statistical profiler now walks deeper into the
740 call stack for more meaningful call-graphs and accrued time
741 reports (x86/x86-64 only). It also now reports time spent in
743 * enhancement: it is now possible to trace most individual methods
744 of a generic function in addition to tracing the generic function
746 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
748 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
749 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
750 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
751 produces an error. (thanks to Vincent Arkesteijn)
752 * bug fix: NAMESTRING on pathnames with :WILD components in their
753 directories works correctly. (thanks to Artem V. Andreev)
754 * fixed bug 125: compiler preserves identity of closures. (reported
756 * bug fixed: functions with &REST arguments sometimes failed with
757 "Undefined function" when compiled with (DEBUG 3). (reported by
759 * bug fix: overflow during compiling of setting element of a bit
760 vector with constant index and value. (reported by Timmy Douglas)
761 * build fix: fixed the dependence on *LOAD-PATHNAME* and
762 *COMPILE-FILE-PATHNAME* being absolute pathnames.
763 * on x86 compiler partially supports stack allocation of dynamic-extent
765 * GO and RETURN-FROM do not check the extent of their exit points
766 when compiled with SAFETY 0.
767 * fixed some bugs related to Unicode integration:
768 ** encoding and decoding errors are now much more robustly
769 handled; it should now be possible to recover even from invalid
770 input or output to the terminal. (thanks to Teemu Kalvas)
771 ** provided a first cut at implementing STRING-TO-OCTETS and
772 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
773 ** altered the SB-MD5 contributed module slightly, changing the
774 interface just enough for it to be supportable for builds where
775 lisp characters are not eight bits.
776 * fixed some bugs revealed by Paul Dietz' test suite:
777 ** the FORMATTER-generated functions for ~V[ conditionals require
778 the correct number of arguments.
779 ** READ-FROM-STRING returns the mandated second value when applied
780 to displaced strings.
781 ** the #\Rubout and #\Backspace characters are treated as invalid
782 constituent characters by the tokenizer.
784 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
785 * new feature: reloading changed shared object files with
786 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
787 * new feature: references to foreign variables and functions
788 can now be compiled and loaded before the corresponding shared
789 object file is loaded, as long as the foreign definitions are
790 available at runtime.
791 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
792 supported. (thanks to Dan Debertin)
793 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
794 just keywords. (thanks to Vincent Arkesteijn)
795 * fixed bug #331: structure-class instances corresponding to
796 DEFSTRUCT forms are now created eagerly.
797 * fixed bug #345: backtraces from calls to undefined functions work
798 on x86 as well. Related bug #61 is now also partially fixed on x86
799 (backtraces from throws to unknown catch tags.)
800 * bug fix: lambda-list parsing is now stricter vrt. order and number
801 of lambda-list keywords.
802 * bug fix: as specified by AMOP, an error is signalled if a
803 class-option appears multiple times in a DEFCLASS form. (reported
805 * bug fix: RANDOM can be compiled when the compiler derives the type
806 of its numeric argument as a disjoint set of small integers.
807 (reported by Paul Dietz)
808 * bug fix: starting a core saved with shared objects loaded when
809 those objects are not available no longer causes threaded SBCL to
810 hang. (reported by Sean Ross)
811 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
813 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
814 argument, which must always be NIL. (reported by Kalle Niemitalo)
815 * bug fix: printing 1.0d+23 no longer results in an error.
816 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
817 * bug fix: structure slot setters preserve evaluation order. (thanks
819 * fixed some bugs related to Unicode integration:
820 ** RUN-PROGRAM can allow its child to take input from a Lisp
821 stream. (reported by Stefan Scholl)
822 ** ASDF-INSTALL successfully downloads .tar.gz files in all
823 locales. (reported by Ken Causey)
824 * fixed some bugs revealed by Paul Dietz' test suite:
825 ** INCF, DECF and REMF evaluate their place form as specified in
827 ** FORMATTER expands ~{ ~} iteration directives with V or #
828 parameters correctly.
829 ** FORMATTER deals with the ~@[ ~] conditional directive where the
830 consequent uses no arguments correctly.
831 ** the system has a partial understanding of the (COMPLEX RATIO)
833 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
836 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
837 * new feature: a build-time option (controlled by the :SB-UNICODE
838 keyword feature, enabled by default) for building the system with
839 support for the entire 21-bit character space defined by the
841 * new feature: the system now has rudimentary external-format
842 support; the primary user-visible change at this time is that
843 characters with the high bit set (such as lower-case-e-acute) will
844 print correctly to a terminal in a UTF-8 environment.
845 * minor incompatible change: BASE-CHAR no longer names a class;
846 however, CHARACTER continues to do so, as required by ANSI.
847 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
848 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
849 has been moved to the SB-EXT package (temporarily re-exported from
851 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
852 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
854 * fixed bug #351: better error-handlind and reporting for malformed
856 * fixed bug #350: bignum-printing is now more memory-efficient,
857 allowing printing of very large bignums, eg. (expt 2 10000000).
858 (reported by Bruno Haible)
859 * fixed bug #302: better primitive-type selection for intersection
861 * fixed bug #308: non-graphic characters now all have names, as
862 required. (reported by Bruno Haible)
863 * bug fix: redefining a class with different superclasses now correctly
864 removes it from the direct-subclasses of its previous superclasses.
865 (reported by David Morse)
866 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
867 new value now works. (reported by Bruno Haible)
868 * bug fix: correct canonicalization of multiple non-standard slot
869 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
870 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
871 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
872 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
873 now exists, an signals an error.
874 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
875 during macroexpansion for non-iteration variables as well. (reported
876 by Bruno Haible for CMUCL)
877 * bug fix: Cyclic structures and unprintable objects in compiler
878 messages no longer cause errors. (reported by Bruno Haible)
879 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
880 and READ-FROM-STRING all now return a primary value of NIL if
881 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
882 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
883 (reported by Bruno Haible for CMUCL)
884 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
885 do not designate a condition. (reported by Bruno Haible for
887 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
888 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
889 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
890 permitting resolution in favour of any of the conflicting symbols.
891 (reported by Bruno Haible for CMUCL)
892 * bug fix: EQUAL compiler optimizations is less aggressive on
893 strings which can potentially compare true despite having distinct
894 specialized array element types.
895 * bug fix: unit enumerations can be defined without dividing by
896 zero. (thanks to Vincent Arkesteijn)
897 * FORMAT compile-time argument count checking has been enhanced.
898 (report from Bruno Haible for CMUCL)
899 * a partial workaround for the bug 262: the compiler does not try to
900 inline-expand a local function doing RETURN-FROM from a deleted
901 BLOCK. (thanks to Peter Denno for the bug report and to David
902 Wragg for the simple test case)
903 * fixed some bugs revealed by Paul Dietz' test suite:
904 ** RENAME-PACKAGE allows all package designators as new package
906 ** constraint propagation and lambda variable substitution are
907 more cautious in dealing with partially deleted code.
908 ** compiler sometimes lost reoptimization passes.
909 ** CERROR, given a condition as condition designator, passes its
910 remaining arguments to the continue format control without
912 ** Case-altering FORMAT directives work correctly on non-ASCII
914 ** The REAL type specifier handles bounds outside the floating
915 point ranges without signalling FLOATING-POINT-OVERFLOW.
916 ** Functions with IR1-transformations can create intercomponent
917 references to global functions.
918 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
920 ** FORMAT ~R treats a nil value for its first parameter correctly.
922 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
923 * enhancement: saving cores with foreign code loaded is now
924 supported on x86/NetBSD and sparc/Linux in addition to the previously
926 * bug fix: on some platforms repeated installations caused multiple
927 copies of HTML documentation to be installed -- should not happen
928 any more. (reported by Stefan Scholl)
929 * bug fix: parsing self-recursive alien record types multiple times
930 no longer causes infinite recursion. (reported by Thomas F. Burdick,
931 original patch by Helmut Eller for CMUCL)
932 * bug fix: stack-exhaustion detection works now on NetBSD as well.
933 (thanks to Richard Kreuter)
934 * bug fix: defining classes whose accessors are methods on existing
935 generic functions in other (locked) packages no longer signals
936 bogus package lock violations. (reported by François-René Rideau)
937 * bug fix: special variables as DEFMETHOD parameters no longer have
938 associated bogus type declarations. (reported by David Wragg and
940 * bug fix: read-write consistency on streams of element-type
941 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
942 * bug fix: redefiniton of the only method of a generic function with
943 no DEFGENERIC no longer emits a full WARNING. In addition,
944 redefinition of generic functions with no DEFGENERIC to an
945 incompatible lambda list now signals an error. (thanks to Zach
947 * bug fix: DEFGENERIC now works even when there's a function of the
948 same name in an enclosing lexical environment. (thanks to Zach
950 * fixed compiler failure, caused by instrumenting code during
951 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
952 * optimization: added loop analysis and improved register allocation
953 * fixed some bugs revealed by Paul Dietz' test suite:
954 ** POSITION on displaced vectors with non-zero displacement
955 returns the right answer.
956 ** (SIMPLE-STRING) is a valid type specifier for sequence
958 ** *PRINT-LEVEL* handling for slotless structures is pedantically
960 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
961 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
963 ** The justification version of the FORMAT ~< directive treats
964 non-zero minpad parameter correctly.
966 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
967 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
968 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
969 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
970 the supported interface.
971 * new feature: Single-stepping of code compiled with DEBUG 2 or
972 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
973 * new feature: saving cores with foreign code loaded is now
974 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
975 Timothy Moore's work for CMUCL)
976 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
977 parameters to * instead of NIL if no initform is supplied.
978 (reported by Johan Bockgaard)
979 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
980 correctly when FUNCALL appears as the car of the form. Note:
981 despite this FUNCALL forms are not currently subject to
982 compiler-macro expansion. (port of Raymond Toy's fix for the
983 same from CMUCL, reported by Johan Bockgaard)
984 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
985 (thanks for Teemu Kalvas)
986 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
987 case instead of terminating the iteration (thanks for Julian
988 Squires, Sean Champ and Raymond Toy)
989 * bug fix: incorrect expansion of defgeneric that caused a style
990 warning. (thanks for Zach Beane)
991 * bug fix: slot accessor effective method computation works properly
992 for classes with multiple non-standard applicable methods on
993 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
994 * on x86 compiler supports stack allocation of results of LIST and
995 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
996 CMUCL implementation by Gerd Moellmann)
997 * fixed some bugs revealed by Paul Dietz' test suite:
998 ** FORMAT strings with both the ~<~:;~> form of the justification
999 directive and pretty-printing directives cause an error.
1001 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1002 * incompatible change: the internal functions
1003 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1004 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1005 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1006 instead of the old functions.
1007 * new feature: on platforms where "dladdr" is available foreign
1008 function names now appear in backtraces. (based on Helmut Eller's
1010 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1011 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1013 * documentation: ASDF manual is now distributed with SBCL.
1014 * bug fix: SBCL can now load its contributed modules with REQUIRE
1015 even if the system-provided entries have been removed from
1016 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1018 * bug fix: backtraces involving undefined functions or assembly
1019 routines are more informative. (thanks to Brian Downing)
1020 * bug fix: mutually referent alien structures now work correctly.
1021 (reported by Rick Taube)
1022 * bug fix: structures defined by WITH-ALIEN can be referred to
1023 within other definitions in the same WITH-ALIEN.
1024 * bug fix: division operators (MOD, TRUNCATE and the like) with
1025 constant zero divisors and integer dividends no longer generate
1027 * bug fix: provide default methods for INPUT-STREAM-P and
1028 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1029 * bug fix: improve the detection and resolution of MOP metacycles
1030 (where effective method computation depends on the generic function
1031 having its effective method computed). (reported by Bruno Haible)
1032 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1033 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1034 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1035 * bug fix: pathnames with relative directory components can be
1036 represented relative to default pathnames.
1037 * optimization: in taking the GCD of bignums, reduce the two bignums
1038 to approximately the same size (using Euclid's algorithm) before
1039 applying the more sophisticated binary GCD. (thanks to Juho
1041 * optimization: COUNT on bitvectors now operates word-at-a-time.
1042 * optimization: ASH with a positive, but not necessarily constant,
1043 (leftwards) shift, when in a modular context, is compiled to a
1045 * fixed some bugs revealed by Paul Dietz' test suite:
1046 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1047 if the corresponding argument is NIL.
1048 ** FORMAT directives accepting a minpad parameter treat negative
1050 ** literal commas as character parameters to FORMAT directives are
1052 ** literal spaces directly after ~<Newline> directives within a
1053 format-logical-block (~:< ~@:>) do not induce :FILL-style
1054 conditional newlines.
1055 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1056 as specified: it no longer includes conditional newlines).
1057 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1059 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1060 :PER-LINE-PREFIX argument does not evaluate to a string.
1062 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1063 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1064 the manual for details; for now, package locks can be disabled by
1065 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1066 if no major problems are found then it is likely that they will be
1067 compiled in unconditionally.
1068 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1069 unsupported operators on all platforms. To load a shared library
1070 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1071 object file, link it into a shared library outside of SBCL and
1072 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1074 * bug workaround: ROOM T and the GC were not working together
1075 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1076 reporting facility aren't preserved. That reporting has been
1077 disabled, so now until and unless someone figures out how to make it
1078 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1079 * minor incompatible change: as threatened around sbcl-0.8.0, the
1080 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1081 The symbols are also exported from SB-PCL for backwards
1082 compatibility, but more so than before SB-PCL should be treated as
1083 an implementation-internal package.
1084 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1086 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1087 It is known as of this release not to work on the Alpha, however.
1088 * fixed bug #167: errors signalled due to illegal syntax in method
1089 bodies are now more legible.
1090 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1091 designators and can hence be used with TYPEP.
1092 * fixed bug #333: CHECK-TYPE now ensures that the type error
1093 signalled, if any, has the right object to be accessed by
1094 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1095 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1096 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1098 * fixed bug #334: programmatic addition of slots using specialized
1099 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1100 and :ALLOCATION :CLASS slots.
1101 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1102 float exponents. (rereported by Peter Seibel)
1103 * fixed a bug: #\Space (and other whitespace characters) are no
1104 longer considered to be macro characters in standard syntax by
1105 GET-MACRO-CHARACTER.
1106 * fixed bug: initialization of condition class metaobjects no longer
1107 causes an instance of the condition to be created. (reported by
1109 * fixed bug: it is now possible to have slots such that
1110 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1111 description is neither :INSTANCE nor :CLASS.
1112 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1113 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1114 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1115 non-standard SLOT-DEFINITION-ALLOCATION values.
1116 * fixed bugs: various race conditions handling exiting threads.
1117 CL-PPCRE's thread torture test now passes.
1118 * fixed arguably-a-bug: GC time proportional to number of threads in
1119 system even when most of them are idle
1120 * optimization: improved performance of BIT and SBIT on bit-vectors.
1121 * .fasl file incompatibility: The fasl file version number has
1122 been incremented because of changes associated with package locks.
1124 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1125 * minor incompatible change: the system no longer provides
1126 optimization hints (conditions which are TYPEP
1127 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1128 mode; these hints are still emitted when higher SPEED optimization
1130 * new contrib module: a sampling profiler (profiling by statistical
1131 sampling, rather than by instrumenting functions) is available as
1132 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1133 * the behaviour of the standard function ED is now customizeable by
1134 third parties through a hook variable: see ED's documentation
1135 string for information on the protocol.
1136 * the compiler no longer emits efficiency notes for (FUNCALL X)
1137 when the type of X is uncertain under default optimization
1139 * fixed bug 276: mutating a binding of a specialized parameter to a
1140 method to something that is not TYPEP the specializer is now
1142 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1143 exhibit the required behaviour on the x86 platform. (thanks to
1144 Peter van Eynde, Eric Marsden and Bruno Haible)
1145 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1146 even for difficult arguments. (reported by Peter Graves)
1147 * fixed bug 141a: the backquote printer now descends quoted
1149 * fixed another bug in backquote printing: no more destructive
1150 modification of the form's list structure. (reported by Brian
1152 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1153 move between its address being taken and the call to
1154 interrupt_thread, fixing a crashing race condition.
1155 * the SB-POSIX contrib implementation has been adjusted so that it
1156 no longer exhibits ridiculously poor performance when constructing
1157 instances corresponding to C structs.
1159 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1160 * minor incompatible change: the sb-grovel contrib now treats C
1161 structures as alien (in the sense of SB-ALIEN) objects rather than
1162 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1163 has implications for memory management of client code
1164 (sb-grovel-returned objects must now be manually managed) and for
1165 type safety (alien objects now have full types).
1166 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1167 used to control emission of compiler diagnostics, rather than the
1168 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1169 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1170 quality should be considered deprecated.
1171 * install.sh now installs the user manual as well
1172 * (not quite a new documentable feature, but worth considering in
1173 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1174 beginnings of a semantically meaningful condition hierarchy is
1175 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1177 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1178 designator as the defaults argument.
1179 * fixed bug: Displaced arrays whose displaced-to array has become
1180 too small now cause ARRAY-DIMENSION to signal an error, providing
1181 for safer bounds-checking. (reported by Bruno Haible)
1182 * fixed bug: DEFCLASS slot definitions with identical :READER and
1183 :WRITER names now signal a reasonable error. (reported by Thomas
1185 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1187 * fixed bug: Invalid slot specification errors now print correctly.
1188 (thanks to Zach Beane)
1189 * fixed bug 320: Shared to local slot value transfers in class
1190 redefinitions now happen corrently from superclasses as
1191 well. (reported by Bruno Haible)
1192 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1194 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1195 improper lists correctly. (reported by Bruno Haible)
1196 * fixed bug 313: source-transform for <fun-name> was erroneously
1197 applied to a call of a value of a variable with name <fun-name>.
1198 (reported by Antonio Menezes Leitao)
1199 * fixed bug 307: The obsolete instance protocol ensures that
1200 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1201 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1202 remove dead unknown-values globs from the middle of the stack.
1203 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1204 easier to resume long computations after using *BREAK-ON-SIGNALS*
1205 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1206 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1207 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1208 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1209 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1210 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1212 * fixed a (fairly theoretical) bug in string printing: if
1213 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1214 does not have array-element-type equal to the most general string
1216 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1217 function name. (thanks to Nikodemus Siivola)
1218 * fixed bug: compile-time format string checker failed on
1219 non-closed ~{. (reported by Thomas F Burdick)
1220 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1221 don't warn on legal loop code involving more than one aggregate
1222 boolean. (thanks to Nikodemus Siivola)
1223 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1224 in keyword position to MAKE-PACKAGE was not being reported as
1226 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1227 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1229 * fixed bugs 280 and 312: the checking for multiple definitions in a
1230 file is less likely to become confused by uses of inline
1232 * fixed bug: the #S reader macro performs the keyword coercion
1233 specified for slot names. (reported by Kalle Niemitalo)
1234 * fixed bug: lambda lists may contain symbols whose names start with
1235 & but are not lambda-list-keywords; their occurrence triggers a
1237 * fixed bug 321: define-method-combination argument lambda lists do
1238 not cause invalid code to be generated when &optional or &aux
1239 variables have default values. (reported by Bruno Haible)
1240 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1241 have CLOS classes; this appears also to have fixed reported
1242 instability in stack exhaustion detection.
1243 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1244 MUFFLE-WARNING when no associated restart is present is now
1246 * optimization: rearranged the expansion of various defining macros
1247 so that each expands into only one top-level form in a
1248 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1250 * optimization: used a previously little-used slot in symbols to
1251 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1253 * fixed some bugs revealed by Paul Dietz' test suite:
1254 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1256 ** Fixed an optimization bug related to inheritance of initforms
1257 from local to shared slots.
1258 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1259 ** CAST optimizer forgot to flush argument derived type.
1260 ** print/read consistency on floats is now orders of magnitude
1261 more likely. (thanks also to Bruno Haible for a similar report
1263 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1264 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1265 for objects of type REAL. Make it so.
1266 ** SXHASH is specified (infelicitously) to respect similarity,
1267 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1268 it so. (thanks to Markus Ziegler)
1269 ** on the Alpha, the compiler succeeds in compiling functions
1270 returning a known number of arguments greater than 63.
1271 ** fixed handling of invalid NIL arguments in keyword position
1273 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1275 ** provided a readably-printable representation for RANDOM-STATE
1277 ** ensured that pathnames, where they have a namestring, always
1278 print using #P"..." syntax.
1280 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1281 * Support for the forthcoming 2.0 version of the NetBSD kernel
1282 running on x86 hardware has been added. (thanks to Perry
1283 E. Metzger most immediately, and others for their past work)
1284 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1285 current release version; SBCL's OpenBSD support had been broken
1286 since about the time of OpenBSD's switch to ELF binary format.)
1287 * [placeholder for DX summary]
1288 ** user code with &REST lists declared dynamic-extent, under high
1289 speed or space and low safety and debug optimization policy.
1290 * The manual has been converted to Texinfo format and the debugger
1291 chapter from the cmucl manual has been added.
1292 * A facility has been added to extract documentation strings from
1293 sbcl and store them as Texinfo-formatted snippets for inclusion in
1294 the manual (via Texinfo's @include directive)
1295 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1296 the test case to Dave Roberts)
1297 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1298 pointers. (reported by Sean Ross)
1299 * bug fix: PROFILE output is printed nicely even for large numerical
1300 values. (thanks to Zach Beane)
1301 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1302 greater than 32 handle EOF correctly.
1303 * bug fix: on X86 an immediate argument of the IMUL instruction is
1304 correctly printed in disassembly. (thanks to Lutz Euler)
1305 * bug fix: class slots in redefined classes preserve their old
1306 values. (thanks to Bruno Haible and Nikodemus Siivola)
1307 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1308 succeeds. (reported by Marco Baringer)
1309 * fixed some bugs revealed by Paul Dietz' test suite:
1310 ** READ-SEQUENCE now works on ECHO-STREAMs.
1311 ** RATIONALIZE works more according to its specification. (thanks
1314 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1315 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1316 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1317 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1318 should matter to you only if you rebind the printer control
1319 variables and then find you want different bindings in the
1320 debugger than in the ordinary execution of your program.)
1321 * The runtime build system has been tweaked to support building
1322 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1323 assembler and linker. (thanks to Nikodemus Siivola)
1324 * Unbound, undefined, undeclared variables now trigger full
1325 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1326 more often programmer error than deliberate exploitation of undefined
1328 * optimization: the hash algorithm for strings has changed to one
1329 that is less vulnerable to spurious collisions. (thanks to Juho
1331 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1332 needless bounds checking. (thanks to Juho Snellman)
1333 * optimization: implemented multiplication as a modular
1334 (UNSIGNED-BYTE 32) operation on the PPC backend.
1335 * fixed some bugs revealed by Paul Dietz' test suite:
1336 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1338 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1339 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1340 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1342 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1344 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1346 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1347 their output stream on EOF from read.
1348 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1349 have been read to end-of-file.
1350 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1352 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1353 description of determination of which consecutive characters
1355 ** Printing the "Space" character with escaping on now yields "#\\ ",
1356 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1357 ** Reading floating-point numbers with *READ-BASE* set to a number
1358 less than 10 works correctly.
1359 ** Reading floating-point numbers with *READ-BASE* set to a number
1360 more than 10 works correctly.
1361 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1362 the readtable currently in effect.
1364 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1365 * minor incompatible change: parsing of namestrings on a physical
1366 (Unix) host has changed; numbers after the final #\. in a
1367 namestring are no longer interpreted as a version field. This is
1368 intented to be largely invisible to the user, except that the
1369 meaning of the namestring "*.*.*" has changed: it now refers to a
1370 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1371 should usually be replaced by
1372 (make-pathname :name :wild :type :wild :version :wild)
1373 with the added benefit that this is more likely to be portable.
1374 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1375 option to OPEN now signals an error if the file being opened
1376 exists; this may have an impact on existing code.
1377 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1378 signals received, on the PowerPC platforms (both Linux and
1379 Darwin). (thanks to Pierre Mai for pointing out the location of
1381 * several fixes on OS X: The system now builds and runs cleanly on
1382 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1383 (thanks to Brian Mastenbrook)
1384 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1385 second-resolution: integer multiples of 1/3600 between -24 and 24.
1386 (thanks to Vincent Arkesteijn)
1387 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1388 type when called with 1 argument; PEEK-CHAR checked type of
1389 PEEK-TYPE only after having read first character from a
1390 stream. (reported by Peter Graves)
1391 * bug fix: the garbage collector now has much better locality
1392 behaviour, and in particular no longer treats all memory as being
1393 exhausted when many small objects point to each other in a deeply
1395 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1396 recognized as being TYPEP their class.
1397 * bug fix: the PUSHNEW documentation string has been corrected.
1398 (thanks to Vincent Arkesteijn)
1399 * bug fix: defaulting of the value for the last of an atypically
1400 large number of multiple values being bound was not being
1401 performed correctly on the Alpha or PPC platforms
1402 * optimization: implemented multiplication as a modular
1403 (UNSIGNED-BYTE 32) operation on the x86 backend.
1404 * optimization: SEARCH on simple-base-strings can now be open-coded.
1405 (see also contrib/compiler-extras.lisp for inspiration for
1406 teaching the compiler about the Boyer-Moore algorithm).
1407 * value, returned by MAX (and MIN) called with several EQUALP, but
1408 not EQL, arguments now does not depend on compiler settings.
1409 * fixed some bugs revealed by Paul Dietz' test suite:
1410 ** in stack analysis liveness information is propagated from
1411 non-local entry points.
1412 ** pathwise CAST removing failed when the CAST node did not start
1414 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1415 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1417 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1418 host is already defined.
1419 ** RENAME-FILE works on streams instead of signalling an internal
1421 ** PEEK-CHAR uses the current readtable when determining whether
1422 or not a character is whitespace.
1423 ** MERGE-PATHNAMES handles the case when the pathname does not
1424 specify a name while the default-pathname specifies a version.
1425 ** Pathnames now stand a better chance of respecting print/read
1427 ** Attempting to use standardized file system operators with a
1428 pathname with invalid :DIRECTORY components signals a
1430 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1432 ** DIRECTORY on logical pathnames is more correct.
1433 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1434 signal a TYPE-ERROR if their argument is not a stream.
1435 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1436 designator argument does not designate a stream.
1437 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1438 examining the synonym.
1439 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1441 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1442 element-type larger than ([UN]SIGNED-BYTE 32).
1444 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1445 * When built with the :SB-FUTEX feature, threaded builds now take
1446 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1447 for faster/more reliable mutex and condition variable support.
1448 * Incompatible change (but one you probably shouldn't have been using
1449 anyway): the interface and code for arbitrating between multiple
1450 threads in the same user session has been redesigned.
1451 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1452 error when called without an explicit environment argument.
1453 (thanks to Nikodemus Siivola)
1454 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1455 confused about their position. (thanks to Adam Warner and Gerd
1457 * bug fix: Pretty printing backquoted forms with unquotations in the
1458 argument list position of various code constructs such as LAMBDA
1459 now works correctly. (reported by Paul Dietz)
1460 * bug fix: Pretty printing unquotations no longer loses all
1461 stream position information.
1462 * optimization: performance of string output streams is now less
1463 poor for multiple small sequence writes.
1464 * optimization: performance of CSUBTYPEP in the presence of complex
1465 expressions involving CONS and NOT many times has been improved.
1466 (reported by Paul Dietz)
1467 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1469 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1470 streams for the REPL, for the debugger, and so on. (thanks to
1472 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1473 * fixed some bugs revealed by Paul Dietz' test suite:
1474 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1475 is ignored if it is too small, rather than propagating through
1476 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1477 ** extremely complex negations of CONS types were not being
1478 sufficiently canonicalized, leading to inconsistencies in
1480 ** VALUES tranformer lost derived type.
1482 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1483 * fixed a bootstrapping bug: the build process no longer assumes
1484 that the various BOOLE-related constants have the same value in
1485 host and target lisps. (noted by Paul Dietz' test suite on an
1486 SBCL binary built from CLISP)
1487 * The system can now be dynamically linked on the MIPS platform,
1488 which enables dynamic loading of foreign code from Lisp. (thanks
1489 to Ralf Baechle for discussions on the MIPS ABI)
1490 * The system now records debugging information for its own source
1491 files in a filesystem-position-independent manner, relative to
1492 the "SYS" logical host.
1493 * fixed a compiler bug: MV-LET convertion did not check references
1494 to the "max args" entry point. (reported by Brian Downing)
1495 * tweaked disassembly notes to be less confident about proclaiming
1496 some instruction as an LRA. (thanks to Brian Downing)
1497 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1498 now each have their own history, command character, and other
1499 characteristics. (thanks to David Lichteblau)
1500 * fixed some bugs revealed by Paul Dietz' test suite:
1501 ** compiler failure in compiling LOGAND expressions including a
1503 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1504 shift greater than 32.
1505 ** FLUSH-DEST did not mark blocks for type check regeneration.
1506 ** HANDLER-CASE failed to accept declarations in handler clauses
1507 in some circumstances.
1509 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1510 * New code in contrib/sb-introspect (still probably not entirely
1511 stable yet) provides some support for smart Lisp development
1512 environments like SLIME.
1513 * The conditions signalled for errors occurring when loading .fasl
1514 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1515 in a way which should help ASDF recover gracefully.
1516 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1517 stuff has been cleaned up. If you code contrib/ stuff, this might
1518 affect you, and you can look at contrib/README, contrib/STANDARDS,
1519 and/or the 0.8.4.27 diff to check.
1520 * In full calls the compiler now does not generate checks for declared
1521 argument types for all arguments.
1522 * various threading fixes
1523 ** and some experimental patches which didn't make it into
1524 the main tree for this release, but which are shipped in
1525 contrib/experimental-thread.patch as a possible fix for some
1526 failures (deadlock, spinning...) in GC-intensive multithreaded
1528 * fixed PPC build problem (source code incompatibility of different
1529 library versions): added offsetof() hackery which attempts to divine
1530 where glibc maintainers put uc_mcontext today
1531 * fixed bug 282: compiler does not trust type assertions while passing
1532 arguments to a full call.
1533 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1534 &OPTIONAL VALUES type parameter.
1535 * fix bug 214: algorithm for noting rejected templates is now more
1536 similar to that of template seletion. (also reported by rydis on
1538 * fixed bug 141b: printing backquoted information readably and prettily
1539 inserts a space where necessary.
1540 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1541 (CHARACTER 10) are now reported as errors, rather than propagated
1542 as unknown types. (reported by piso on #lisp)
1543 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1544 with values NIL and :ERROR. (thanks to Milan Zamazal)
1545 * fixed bug 191c: CLOS now does proper keyword argument checking as
1546 described in CLHS 7.6.5 and 7.6.5.1.
1547 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1548 longer raise an error; further, using a list as a for-as-arithmetic
1549 counter now raises a meaningful error.
1550 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1551 now understood by sequence creation functions such as MAKE-SEQUENCE
1553 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1554 a STRING-INPUT-STREAM.
1555 * compiler enhancement: SIGNUM is now better able to derive the type
1557 * type declarations inside WITH-SLOTS are checked. (reported by
1559 * fixed some bugs revealed by Paul Dietz' test suite:
1560 ** incorrect optimization of TRUNCATE for a positive first
1561 argument and negative second.
1562 ** compiler failure in let-convertion during flushing dead code.
1563 ** compiler failure while deriving type of TRUNCATE on an
1564 interval, containing 0.
1565 ** ASH of a negative bignum by a negative bignum count now returns
1567 ** intersection of CONS types now canonicalizes properly, fixing
1568 inconsistencies in SUBTYPEP.
1570 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1571 * incompatible change: The --disable-debugger command line
1572 option now clobbers the debugger at a more fundamental
1573 level, by redefining #'INVOKE-DEBUGGER instead of by
1574 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1575 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1576 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1577 enter the debugger REPL and then suffer recursive errors
1578 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1579 while under the new system, BREAK in --disable-debugger mode
1580 terminates the system just as an unhandled error would.
1581 * fixed compiler performance when processing loops with a step >1;
1582 * bug fix: DOCUMENTATION now retrieves generic function
1583 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1584 support has been systematized, and now supports the methods
1585 specified by ANSI, along with a default method and a method for
1586 slot documentation. (reported by Nathan Froyd)
1587 * bug fix: effective methods associated with a generic function are
1588 no longer cached over a change of that generic function's method
1589 combination. (reported by Andreas Fuchs)
1590 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1591 names a non-existent directory. (thanks to Andreas Fuchs)
1592 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1593 the PPC platform, lead to stack corruption; this has been fixed.
1594 (reported by Rainer Joswig)
1595 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1596 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1597 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1598 arithmetic in general, is now much more reliable on the PPC
1600 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1601 platform now returns the right answer.
1602 * optimization: restored some effective method precomputation in
1603 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1604 precomputation is now tunable.
1605 * optimization: compiler-internal data structure use has been
1606 reviewed, and changes have been made that should improve the
1607 performance of the compiler by about 20%.
1608 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1609 simple-base-strings and simple-bit-vectors is improved.
1610 * optimization: the optimization of 32-bit logical and arithmetic
1611 functions introduced in version 0.8.3 on the x86 has been
1612 implemented on the mips, ppc and sparc platforms; an
1613 implementation of the same facility, but for 64-bit arithmetic,
1614 has been added for the alpha.
1615 * microoptimization: the compiler is better able to make use of the
1616 x86 LEA instruction for multiplication by constants.
1617 * bug fix: in some situations compiler did not report usage of
1618 generic arithmetic in (SPEED 3) policy.
1619 * bug 145b fix: compiler used wrong type specifier while converting
1620 MEMBER-types to numeric.
1621 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1623 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1624 index argument now works properly on the Alpha platform.
1625 * bug fix: floating point exception treatment on the Alpha platform
1627 * bug fix: FILE-POSITION works much better on string input and
1628 output streams. (thanks to Nikodemus Siivola)
1629 * bug fix: many threading/garbage collection symptoms sorted.
1630 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1631 might be pseudo-atomic.
1632 * internal change: Stopping for GC is now done with signals not
1633 ptrace. GC is now done in whichever thread wanted it, instead of
1635 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1637 * bug fix: result form in DO is not contained in the implicit
1639 * incompatible change: ICR structure is changed; the value part of
1640 CONTINUATION is now called LVAR; corresponding functions are
1641 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1642 * added type deriver for ISQRT (thanks to Robert E. Brown).
1643 * bug fix: better support for loading from the command line when an
1644 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1646 * fixed some bugs revealed by Paul Dietz' test suite:
1647 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1648 ** ROUND and FROUND now give the right answer when given very
1649 small float arguments.
1650 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1652 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1653 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1654 ** LCM should return a non-negative integer.
1655 ** PARSE-INTEGER returned the index of a terminator instead of the
1656 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1657 ** PARSE-INTEGER returned an incorrect index being applied to a
1659 ** LCM with two arguments of 0 returns 0 rather than signalling
1661 ** unsigned addition of a 32-bit constant with the high bit set no
1662 longer causes an internal compiler error on the x86.
1663 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1664 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1665 with negative last argument.
1666 ** byte specifiers with zero size and position no longer cause
1667 an error during type derivation.
1668 ** bignum multiplication on the Alpha platform now returns the
1670 * porting: The system now builds on SuSE AMD64, although it still
1671 generates a 32-bit binary.
1672 * .fasl file incompatibility: The fasl file version number has
1673 been incremented (because of the changes to internal compiler
1674 data structures referred to above).
1676 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1677 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1678 more accurately, on the Darwin kernel running on PowerPC hardware.
1679 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1680 * Compiler code deletion notes now signal a condition of type
1681 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1682 an associated MUFFLE-WARNING restart.
1683 * The compiler now performs limited argument count validation of
1684 constant format strings in FORMAT, and where appropriate in ERROR,
1685 CERROR and WARN. (thanks to Gerd Moellmann)
1686 * New ASDF-INSTALL contrib can be used for automatic download and
1687 installation of third-party Lisp code from CCLAN or other sites
1689 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1690 which forces another thread to execute a function supplied by the
1692 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1693 now accept and act upon their :ELEMENT-TYPE keyword argument.
1694 (reported by Martin Atzmueller, Edi Weitz)
1695 * bug fix: FILE-POSITION now accepts position designators up to
1696 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1697 is the greater. (thanks to Patrik Nordebo)
1698 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1699 Kalle Olavi Niemitalo)
1700 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1701 printable. (reported by Eric Marsden)
1702 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1703 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1704 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1705 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1706 (thanks to Stig Sandoe)
1707 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1708 UNREAD-CHAR now correctly includes the unread character in the
1709 target sequence. (thanks to Gerd Moellmann)
1710 * bug fix in threaded builds: the system can now be suspended and
1711 resumed by shell job control with minimal disruption.
1712 * bug fixes in times and timezones >2038AD
1713 * better handling of "where is GNU make?" problem in build scripts
1714 (thanks to Nikodemus Siivola)
1715 * new optimization: inside a named function any reference to a
1716 function with the same name is considered to be a self-reference;
1717 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1718 optimization quality.
1719 * new optimization on x86: logical functions and + now have
1720 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1721 used when the result is truncated to 32 bits.
1722 * VALUES declaration is partially enabled.
1723 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1724 * fixed some bugs revealed by Paul Dietz' test suite:
1725 ** The system now obeys the constraint imposed by
1726 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1727 types form a lattice under type intersection.
1728 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1729 ** ASSOC now ignores NIL elements in an alist.
1730 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1731 and (1+ MOST-POSITIVE-FIXNUM) answers.
1732 ** The addition of a method with invalid qualifiers to a generic
1733 function does not cause an error to be signalled immediately;
1734 a warning is signalled, and the error is generated only on
1735 calling the generic function.
1736 * changed .fasl file version number, in part to add type codes for
1737 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1738 obscure ANSI requirements
1740 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1741 * fixed bug 148: failure to inline-expand a local function left
1742 garbage, confusing the compiler.
1743 * fixed bugs 3cd: structure slot readers perform type check if the
1744 slot can have an invalid value (i.e. it is either not initialized
1745 or can be written with a less specific slot writer).
1746 * bug fix: the compiler now traps array references to elements off
1747 the end of an array; previously, the bounds checking in some
1748 circumstances could go off-by-one.
1749 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1751 * type declarations for array element types now obey the description
1752 on the CLHS page "Declaration TYPE", as per discussions on
1753 sbcl-help around 2003-05-08. This means that a declaration
1754 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1755 declaration, all references to BAR will be asserted or assumed
1756 (with THE, so dependent on compiler policy) to involve objects of
1757 type FOO. Note that no such declaration is implied in
1758 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1759 * declared types of functions from the "Conditions"
1760 chapter. (reported by Paul Dietz)
1761 * bug fix: CERROR accepts a function as its first argument.
1762 * bug fix: NTH an NTHCDR accept a bignum as index
1763 arguments. (reported by Adam Warner)
1764 * optimization: character compare routines now optimize comparing
1765 against a constant character. (reported by Gilbert Baumann)
1766 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1767 argument now works properly on the MIPS platform.
1768 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1769 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1770 declared to have a functional type, does not cause an error
1772 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1773 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1774 * fixed bug: if last continuation of a deleted block has a
1775 destination, this destination should be deleted too. (reported by
1777 * fixed a bug in the bootstrap process: the host compiler's values
1778 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1779 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1780 test case from Patrik Nordebo)
1781 * improved the ability of the disassembler on the PPC platform to
1782 provide helpful disassembly notes.
1783 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1784 the class in more cases than previously.
1785 * bug fix: FILE-POSITION now understands :START and :END for
1786 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1787 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1788 8) data. (thanks to David Lichteblau)
1789 * bug fix: it is possible to add a method to a generic function
1790 without lambda list.
1791 * bug fix: reader failed to signal END-OF-FILE inside an
1792 object representation. (reported by Nikodemus Siivola)
1793 * fixed some bugs revealed by Paul Dietz' test suite:
1794 ** LAST and [N]BUTLAST should accept a bignum.
1795 ** condition slot accessors are methods.
1796 ** (VECTOR NIL) is a subtype of STRING.
1798 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1799 * minor incompatible change: some nonsensical specialized lambda
1800 lists (used in DEFMETHOD) which were previously ignored now signal
1802 * minor incompatible change: the system is now aware of the types of
1803 variables in the COMMON-LISP package, and will signal errors for
1804 most violations of these type constraints (where previously they
1805 were silently accepted).
1806 * minor incompatible change: COMPILE-FILE now uses the freedom
1807 afforded (ANSI 3.2.2.3) to use derived function types for
1808 functions defined in the same file. This also permits the system
1809 to warn on static type mismatches and function
1810 redefinition. (Currently it does not work with high DEBUG level.)
1811 * minor incompatible change: VALUES declaration is disabled.
1812 * When issuing notes, the compiler now signals a condition of type
1813 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1814 restart for use in user handlers. It is expected that the
1815 COMPILER-NOTE condition will eventually become a condition
1816 supertype to a hierarchy of note types, which will then be
1817 handleable in a similar fashion. However, at the moment, no such
1818 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1819 but it's an implementation detail, not a classification for the
1821 * Changes in type checking closed the following bugs:
1822 ** type checking of unused values (192b, 194d, 203);
1823 ** template selection based on unsafe type assertions (192c, 236);
1824 ** type checking in branches (194bc).
1825 * A short form of VALUES type specifier has ANSI meaning (it has
1826 increased the number of situations when SBCL cannot perform type
1828 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1829 keywords or constants is permissible.
1830 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1831 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1832 classes. (thanks to Antonio Martinez)
1833 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1834 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1835 argument to a pretty-print stream. Instead, it leaves any such
1836 operations to DESCRIBE-OBJECT methods.
1837 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1839 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1840 lambda lists. (thanks to David Lichteblau)
1841 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1842 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1843 (thanks to Gerd Moellmann)
1844 * fixed bug 246: increased compilation speed of long
1845 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1847 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1849 * DEFCONSTANT now throws a condition of type
1850 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1851 constant to a non-EQL value; CONTINUE and ABORT restarts
1852 respectively change and preserve the value.
1853 * fixed bug 63: The code walker, part of the implementation of CLOS,
1854 is now better at handling symbol macros.
1855 * bug fix: There is no longer an internal implementation type named
1856 CL:LENGTH. (reported by Raymond Toy)
1857 * bug fix: In macro-like defining macros/special operators the
1858 implicit block does not enclose the lambda list.
1859 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1860 enclosed in parentheses) are not suitable as type specifiers, and
1861 their use properly signals an error now.
1862 * bug fix: An argument count mismatch for a type specifier in code
1863 being compiled no longer causes an unhandled error at compile
1864 time, but signals a compile-time warning.
1865 * fixed simple vector readable printing
1866 * bug fix: DESCRIBE takes more care over whether the class
1867 precedence list slot of a class is bound before accessing it.
1868 (reported by Markus Krummenacker)
1869 * bug fix: FORMATTER can successfully compile pretty-printer format
1870 strings which use variants of the ~* directive inside.
1871 * bug fix: SEARCH now applies its TEST predicate to the elements of
1872 the arguments in the correct order. (thanks to Wolfhard Buss)
1873 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1874 in type inference. (thanks to Robert E. Brown)
1875 * bug fix: Reading in symbols with an explicit package name of ""
1876 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1877 * fixed some bugs revealed by Paul Dietz' test suite:
1878 ** NIL is now allowed as a structure slot name.
1879 ** Arbitrary numbers, not just REALs, are allowed in certain
1880 circumstances in LOOP for-as-arithmetic clauses.
1881 ** Multiple class redefinitions before slot access no longer
1882 causes a type error.
1883 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1884 association between the name and a class.
1885 ** Generic functions with non-standard method-combination and over
1886 five methods all of which return constants no longer return NIL
1887 after the first few invocations. (thanks to Gerd Moellmann)
1888 ** CALL-NEXT-METHOD with no arguments now passes the original
1889 values of the arguments, even in the presence of assignment.
1890 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1892 ** DISASSEMBLE works with closures and funcallable instances.
1893 ** ADD-METHOD now returns the generic function, not the new method.
1894 ** FIND-METHOD signals an error if the lengths of the specializers
1895 is incompatible with the generic function, even if the ERRORP
1897 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1898 which its argument is a member.
1899 ** DEFCLASS only redefines the class named by its class-name
1900 argument if that name is the proper name of the class;
1901 otherwise, it creates a new class.
1902 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1903 of the UNBOUND-SLOT condition to the name of the slot.
1904 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1905 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1906 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1907 SLOT-MAKUNBOUND in the specified fashion.
1909 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1910 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1911 cross-compilation host. As a consequence, we can now bootstrap our
1912 way up to SBCL starting with a bare gcc toolchain and human-readable
1913 source code (first the source to CLISP, then the source to SBCL).
1914 * A contributed module containing a partial implementation of the
1915 simple-streams interface has been included. (thanks to Rudi
1917 * A contributed module implementing the RFC1321 Message Digest
1918 Algorithm, known as MD5, has been included.
1919 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1920 no longer has any effect, as the code controlled by this feature
1921 has been deleted. (As far as we know, no-one has ever built using
1922 this feature, and its semantics were confused in any case).
1923 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1924 work on conditions (as required by the ANSI specification),
1925 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1926 expected behaviour on conditions. Users should note, however,
1927 that such behaviour is not required by the ANSI specification,
1928 and so use of this behaviour may render their code unportable.
1929 * fixed some bugs revealed by Paul Dietz' test suite:
1930 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1932 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1933 ** SXHASH on condition objects no longer returns NIL.
1934 ** :ALLOCATION :CLASS slots are better treated; their values are
1935 updated on class redefinition, and initforms inherited from
1936 superclasses are applied.
1937 ** REMOVE-METHOD returns its generic function argument even when
1938 no method was removed.
1939 ** SHARED-INITIALIZE now initializes the values of the requested
1940 slots, including those with :ALLOCATION :CLASS.
1941 ** ALLOCATE-INSTANCE now works on structure classes defined via
1942 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1944 ** SLOT-EXISTS-P now works on conditions, as well as structures
1946 ** MAKE-LOAD-FORM now has the required methods on
1947 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1948 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1949 keyword, and now implements the SLOT-NAMES argument.
1950 ** methods with &OPTIONAL arguments no longer allow too many
1951 arguments to be passed in the call without error.
1952 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1953 option is consistent with the required arguments of the generic
1954 function lambda list.
1955 * bug fix: REQUIRE accepts a string designator. (Thanks to
1957 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1958 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1959 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1960 * bug fix: The compiler now checks for duplicated variables in macro
1962 * bug fix: SETQ on globals returns the correct value.
1963 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1964 not a COMPILER-ERROR (followed by some other strange error on
1965 choosing the CONTINUE restart).
1966 * bug fix: make.sh and friends are now more consistent in the way that
1967 they look for GNU "make".
1969 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1970 * experimental native threads support (on x86 Linux >=2.4 only).
1971 This is not compiled in by default: you need to add :SB-THREAD to
1972 the target features. See the "Beyond ANSI" chapter of the manual
1974 * fix for longstanding nonANSIism: The old distinction between
1975 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1976 The return value from CL:FIND-CLASS is now a CLOS class, and
1977 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1978 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1979 * An interface to the MetaObject Protocol, as described in Kiczales,
1980 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1981 MIT Press, 1991, is available from the SB-MOP package.
1982 * incompatible change: the SB-PCL package should now be considered
1983 a private implementation detail, and no longer a semi-private MOP
1985 * minor incompatible change: due to rearrangement for threads, the
1986 control stack and binding stack are now allocated at arbitrary
1987 addresses instead of being hardcoded per-port. Users affected by
1988 this probably have to be doing advanced things with shared
1989 libraries, and will know who they are.
1990 * minor incompatible change: Previously, all --eval forms used were
1991 processed with READ before any of them were processed with EVAL.
1992 Now each --eval form is processed with both READ and EVAL before
1993 the next --eval form is processed. (Thus package operations like
1994 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1995 work as the user might reasonably expect.)
1996 * minor incompatible change: *STANDARD-INPUT* is now only an
1997 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1999 * minor incompatible change: Y-OR-N-P is now character-oriented,
2000 not line oriented. Also, YES-OR-NO-P now works without errors.
2001 (thanks to Antonio Martinez)
2002 * sb-aclrepl module improvements: an integrated inspector, added
2003 repl features, and a bug fix to :trace command.
2004 * Known functions, which cannot be open coded by the backend, are
2005 considered to be able to check types of their arguments. (fixing
2006 a bug report by Nathan J. Froyd)
2007 * fixed a bug in computing method discriminating functions: It is
2008 now possible to define methods specialized on classes which have
2009 forward-referenced superclasses. (thanks to Gerd Moellmann)
2010 * fixed evaluation order in optional entries (reported by Gilbert
2012 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2013 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2014 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2015 required-by-AMOP NAME argument, as well as CLASS and
2016 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2017 * fixed bug 20: DEFMETHOD can define methods using names that are
2018 not the proper names of classes to designate class specializers.
2019 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2020 Unix file descriptors, instead of blowing up. (thanks to Antonio
2022 * Garbage collection refactoring: user-visible change is that a
2023 call to the GC function during WITHOUT-GCING will not do garbage
2024 collection until the end of the WITHOUT-GCING. If you were doing
2025 this you were probably losing anyway.
2026 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2027 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2028 TYPEP the latter but not the former.
2029 * The compiler issues a full WARNING for calls to undefined functions
2030 with names from the CL package.
2031 * MAP-INTO for a vector destination is open coded. (reported by
2032 Brian Downing on c.l.l)
2033 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2034 documentation string.
2035 * fixed some bugs revealed by Paul Dietz' test suite:
2036 ** COPY-ALIST now signals an error if its argument is a dotted
2038 ** Condition slots are now accessed more correctly in the presence
2039 of multiple initargs for a given slot.
2040 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2041 exclude restarts of the same name associated with a different
2043 ** DEFCLASS of forward-referenced classes with another
2044 forward-referenced class in the superclasses list no longer
2046 ** Condition slots are now initialized once each, not multiple
2047 times. (thanks to Gerd Moellmann)
2048 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2050 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2051 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2052 arguments contain duplicated elements.
2053 ** RESTART-CASE understands local macros.
2054 ** RESTART-CASE associates exactly its own restarts with a condition.
2055 ** ENDP in safe mode checks its argument to be of type LIST.
2056 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2057 in question is unbound.
2058 ** Optimized MAKE-INSTANCE functions no longer cause internal
2059 assertion failures in the presence of duplicate initargs.
2060 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2061 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2062 propagated even in the ctor optimized implementation of
2064 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2066 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2067 * a better implementation of SXHASH on (simple) bit vectors,
2068 measured both in execution speed and in distribution of results
2069 over the positive fixnums, has been installed. Likewise, a better
2070 implementation of EQUAL for simple bit vectors is now available.
2071 * fixed CEILING optimization for a divisor of form 2^k.
2072 * fixed bug 240 (emitting extra style warnings "using the lexical
2073 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2074 by Antonio Martinez)
2075 * fixed SXHASH, giving different results for NIL depending on type
2076 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2077 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2078 variable. (found by Rolf Wester)
2079 * SBCL does not ignore type declarations for special
2080 variables. (reported by rif on c.l.l 2003-03-05)
2081 * some bug fixes in contrib/sb-aclrepl/
2082 * fixed some bugs revealed by Paul Dietz' test suite:
2083 ** a bug in the CONS type specifier, whereby the CAR and CDR
2084 types got intertwined, has been fixed;
2085 ** the type system is now able to reason about the interaction
2086 between INTEGER and RATIO types more completely;
2087 ** APPEND, [N]REVERSE and NRECONC check that those their
2088 arguments, which must be proper lists, are really so;
2089 ** An array specialized to be unable to hold elements has been
2090 implemented, as required -- yes, really -- by ANSI;
2091 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2092 on malformed property lists;
2094 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2095 * incompatible packaging change: in line with Unix convention,
2096 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2097 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2098 * REQUIRE and PROVIDE are now optionally capable of doing something
2099 useful. See the documentation string for REQUIRE.
2100 * infrastructure for a managed SBCL contrib system: contributed
2101 modules in this release include:
2102 ** the ASDF system definition facility;
2103 ** an interface to the BSD Sockets API;
2104 ** an ACL-like convenience interface to the repl;
2105 (thanks to Kevin Rosenberg)
2106 ** an implementation of ROTATE-BYTE, with efficient implementation
2108 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2109 gives a closer approximation to the right answer than previously.
2110 (thanks to Raymond Toy)
2111 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2112 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2113 optional environment arguments, as required by ANSI.
2114 * fixed bugs in other functions taking environment objects, allowing
2115 calls with an explicit NIL environment argument to be compiled
2117 * fixed bug 228: primary return values from
2118 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2119 COMPILE or FUNCTION.
2120 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2121 :TYPE will no longer signal errors on innocuous objects.
2122 * fixed bug 231b: SETQ is better at respecting type declarations in
2123 the lexical environment.
2124 * fixed a bug in DEFCLASS: classes named by symbols with no or
2125 unprintable packages can now be defined.
2126 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2127 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2128 * fixed compiler failure related to checking types of functions.
2129 (reported by Robert E. Brown)
2130 * the compiler is now much more consistent in its error-checking
2131 treatment of bounding index arguments to sequence functions: in
2132 (SAFETY 3) code, errors will be signalled in almost all cases if
2133 invalid sequence bounding indices are passed to functions defined
2134 by ANSI to operate on sequences.
2135 * fixed a bug in the build procedure: documentation of SBCL-specific
2136 packages is now preserved and available in the final Lisp image.
2137 * lifted FDEFINITION lookup out of loops in the implementation of
2138 many list operations. (thanks to Robert E. Brown)
2139 * fixed a bug in the reader: the #n# reader macro now works for
2140 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2141 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2142 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2143 E. Brown and Tony Martinez respectively)
2144 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2145 SLOT-BOUNDP: in particular, it is now able to optimize them much
2146 better, and is now not vulnerable to having packages renamed.
2147 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2148 always called when a slot is not present in an instance. (thanks
2150 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2151 longer signal an error on unreading a character following EOF on
2152 the previous constituent stream. (thanks to Tony Martinez)
2153 * fixed some bugs revealed by Paul Dietz' test suite:
2154 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2155 not just nonnegative fixnums;
2156 ** the logical bit-array operators such as BIT-AND now accept an
2157 explicit NIL for their "opt-arg" argument (to indicate a
2158 freshly-consed result bit-array);
2159 ** ELT now signals an error on an invalid sequence index in safe
2161 ** the type system is now cleverer about negations of numeric
2162 types, and consequently understands the BIGNUM and RATIO types
2164 ** the type system is now cleverer about the interaction between
2165 INTEGER and RATIO types: while bugs still remain, many more
2166 cases are accurately computed;
2167 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2168 if it is in the last clause;
2169 ** CONSTANTLY now correctly returns a side-effect-free function in
2171 ** DECLARE is no longer treated as a special-operator; in
2172 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2173 * incremented fasl file version number due to the change in the
2174 DEFSTRUCT-SLOT-DESCRIPTION structure.
2176 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2177 * minor incompatible change: code processed by the "interpreter" or
2178 EVAL now has a compilation optimization policy of (DEBUG 2)
2179 (changed from (DEBUG 1)) to improve debuggability of interactive
2180 development, and to allow the use of the debug RETURN command in
2182 * an experimental implementation of the RETURN command for the
2183 debugger has been included. (thanks to Frederik Kuivinen)
2184 * fixed bug 62: constraints were not propagated into a loop.
2185 * fixed bug in embedded calls of SORT (reported and investigated by
2187 * fixed some bugs revealed by Paul Dietz' test suite:
2188 ** printing and reading of arrays with some dimensions having
2189 length 0 (thanks to Gerd Moellmann);
2190 ** BOA constructor with &AUX argument without a default value does
2191 not cause a type error;
2192 ** CONSTANTP now returns true for all self-evaluating objects.
2194 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2195 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2196 accessors that are related by inheritance, as specified in the
2197 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2198 Valtteri Vuorikoski)
2199 * The compiler is now able to inline functions that were defined in
2200 a complex lexical environment (e.g. inside a MACROLET).
2201 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2203 * Support for the upcoming FreeBSD-5.0 release has been included.
2204 (thanks to Dag-Erling Smorgrav)
2205 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2206 effect when it is not in a toplevel context.
2207 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2208 stand a better chance of being correct. (thanks to Gerd
2210 * fixed bug in COERCE, which now signals an error on coercing a
2211 rational to a bounded real type which excludes the expected
2213 * The compiler is now able to derive types more accurately from the
2214 COERCE and COMPILE functions.
2215 * fixed bug 223: functional binding is considered to be constant
2216 only for symbols in the CL package.
2217 * fixed bug 231: SETQ did not check the type of a variable being set
2218 (reported by Robert E. Brown)
2219 * A new optimization for MAKE-INSTANCE has been included, fixing
2220 various bugs (including relating to :ALLOCATION :CLASS slots and
2221 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2223 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2224 ** As required by ANSI, LOOP now disallows anonymous collection
2225 clauses such as COLLECT I in conjunction with aggregate boolean
2226 clauses such as THEREIS (= I 1);
2227 ** LOOP now signals an error when any variable is reused in the
2228 same loop (including the potentially useful construct analogous
2229 to WITH A = 1 WITH A = (1+ A);
2230 ** IT is only a special loop symbol within the first clause of a
2231 conditional loop clause;
2232 ** LOOP with a typed iteration variable over a hashtable now
2233 signals a type error iff it should.
2234 * fixed some other bugs revealed by Paul Dietz' test suite:
2235 ** FILE-STREAM now names the class previously known as FD-STREAM;
2236 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2237 argument) no longer signals an error;
2238 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2239 of the slot symbol, rather than using the current package
2240 ((:CONC-NAME "") continues to intern the slot's name in the
2242 * incremented fasl file version number, because of the incompatible
2243 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2244 of the new implementation of DEFINE-COMPILER-MACRO.
2246 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2247 * Support for building SBCL for MIPS platforms running in
2248 little-endian mode has now been checked in, and basic
2249 functionality on said platforms verified.
2250 * minor incompatible change: PCL now records the pathname of a file
2251 in which methods and the like are defined, rather than its
2253 * minor incompatible change: TRUENAME now considers the truename of
2254 a file naming a directory to be the pathname with :DIRECTORY
2255 component indicating that directory.
2256 * minor incompatible change: a NAMED clause in the extended form of
2257 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2258 reason for the previous behaviour is unclear.
2259 * more systematization and improvement of CLOS and MOP conformance
2260 in PCL (thanks to Gerd Moellman and Pierre Mai):
2261 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2263 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2264 primary methods with no specializers;
2265 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2267 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2268 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2269 FORWARD-REFERENCED-CLASSes; error reporting on
2270 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2272 ** SXHASH on CLOS instances now uses a slot internal to the
2273 instance to return different numbers on distinct instances,
2274 while preserving the same return value through invocations of
2276 ** DEFMETHOD signals errors when methods with longer incongruent
2277 lambda lists are added to generic functions;
2278 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2279 CLASS, as specified in AMOP;
2280 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2281 on the order returned by the primary method for classes of
2282 class STANDARD-CLASS;
2283 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2284 * fixed some bugs shown by Paul Dietz' test suite:
2285 ** DOLIST puts its body in TAGBODY;
2286 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2288 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2289 value producing form;
2290 ** if more variables are given to PROGV than values, extra
2291 variables are bound and made to have no value;
2292 ** NSUBSTITUTE on list arguments gets the right answer with
2294 ** ELT signals an error of type TYPE-ERROR when the index argument
2295 is not a valid sequence index;
2296 ** LOOP signals (at macroexpansion time) an error of type
2297 PROGRAM-ERROR when duplicate variable names are found;
2298 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2299 ** FUNCALL of special-operators now cause an error of type
2301 ** PSETQ now works as required in the presence of side-effecting
2302 symbol-macro places;
2303 ** NCONC accepts any object as its last argument;
2304 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2306 ** loop-for-as-package does not require a package to be explicitely
2308 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2310 * fixed bug 166: compiler preserves "there is a way to go"
2311 invariant when deleting code.
2312 * fixed bug 172: macro lambda lists with required arguments after
2313 &REST arguments now cause an error to be signalled. (thanks to
2315 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2316 bug. (thanks to Matthew Danish)
2317 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2319 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2320 arguments are lexically rebound. (thanks to Gerd Moellmann and
2322 * fixed bug 194: error messages are now more informative when there
2323 is no primary method applicable in a call to a generic
2324 function. (thanks to Gerd Moellmann)
2325 * fixed bug in command line argument checking (thanks to Julian
2327 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2328 (thanks to Matthew Danish)
2329 * incremented fasl file version number, because of the
2330 SXHASH-related changes in the layout of CLOS data structures
2332 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2333 * minor incompatible change: The runtime (the Unix executable named
2334 "sbcl") is now much pickier about the .core files it will load.
2335 Essentially it now requires .core files to descend from the same
2336 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2337 as the runtime does. (The intent is to prevent the crashes which
2338 can occur, and which can even be reported as mysterious failures,
2339 when people patch the sources or change the build parameters
2340 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2341 sbcl and .core files.)
2342 * fixed bug: VALUES-LIST is no longer optimized away.
2343 * fixed bug 142: The FFI conversion of C string values to Lisp
2344 string values no longer conses excessively. (thanks to Nathan
2345 Froyd porting Raymond Toy's fix to CMU CL)
2346 * began to systematize and improve MOP conformance in PCL (thanks to
2347 Nathan Froyd, Gerd Moellman and Pierre Mai):
2348 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2350 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2351 ** FINALIZE-INHERITANCE is now called on class finalization;
2352 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2353 argument precedence order.
2354 * fixed bug 202: The compiler no longer fails on functions whose
2355 derived types contradict their declared type.
2356 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2357 so it can be non-toplevel.
2358 * The fasl file version number has changed (because of the new
2359 implementation of DEFMACRO).
2360 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2361 safe code, that any length requirement by their type-specifier
2362 argument is valid. The exceptions to this are described in bug
2364 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2365 functions now check the types of their inputs as required by ANSI.
2366 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2367 introduced symbol is DECLAREd to be SPECIAL.
2368 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2369 * fixed bug: :COUNT argument to sequence functions may be negative
2370 * fixed bug: body of DO-SYMBOLS may contain declarations
2371 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2372 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2373 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2374 symbol macro only once
2375 * fixed printing of call frame when argument list is unavailable
2376 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2377 * compiler no longer signals WARNING on unknown keyword
2380 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2381 * A beta-quality port to the mips architecture running Linux,
2382 based on the old CMUCL backend, has been made. It has been tested
2383 on a big-endian kernel, and works sufficiently well to be able to
2384 rebuild itself; it has not been tested in little-endian mode.
2385 * fixed an inconsistency between gencgc.c and purify.c which made
2386 dumping/loading .core files unreliable
2387 * fixed bug 120a: The compiler now deals correctly with IFs where
2388 the consequent is the same as the alternative, instead of
2389 misderiving the return type. (thanks to Alexey Dejneka)
2390 * fixed bug 113: Logical pathnames are now dumpable (the logical
2391 host is resolved at load-time, throwing an error if it is not
2393 * fixed bug 174: FORMAT's error message is slightly clearer when a
2394 non-printing character is used in a format directive.
2395 * fixed several bugs in compiler checking of type declarations, i.e.
2396 violations of the Python "declarations are assertions" principle
2397 (thanks to Alexey Dejneka)
2398 * fixed several bugs in PCL's error checking (thanks to Gerd
2400 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2402 * fixed bug in compilation of functions as first class values
2403 (thanks to Antonio Martinez-Shotton)
2404 * The compiler's handling TYPE-ERRORs which it can prove will
2405 inevitably happen at runtime has been cleaned up and corrected
2406 in several ways. (thanks to Alexey Dejneka)
2407 * improved argument type checking for various basic arithmetic
2408 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2409 lavished on them in the past that they can be compiled in many
2410 ways in different special cases
2411 * fixed bug 181: compiler checks validity of user supplied type
2413 * cleaned up code flushing in optimization: Function calls which
2414 should signal errors for safety purposes (e.g. which ANSI says
2415 should signal errors when their arguments are of incorrect type)
2416 are no longer optimized away.
2417 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2418 * incremented fasl file version number, because changes in the
2419 implementation of sequence functions like COERCE caused
2420 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2421 in old inline expansions) to become undefined. (Actually these
2422 changes were later undone, so we might very well be binary
2423 compatible with 0.7.7 after all, but leaving the version number
2424 incremented seemed like the simplest and most conservative
2427 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2428 * An alpha-quality port to the parisc architecture running Linux,
2429 based on the old CMUCL backend, has been made. This, even more so
2430 than the other backends, should be considered still a work in
2431 progress; known problems include that the Linux kernel in 64-bit
2432 mode does not propagate the correct sigcontext structure to
2433 userspace, and consequently SBCL on a parisc64 kernel will not
2435 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2436 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2437 suppressed.) Also now that the compiler is looking at declarations
2438 in the environment, it checks optimization declarations as well,
2439 and suppresses inlining when (> DEBUG SPEED).
2440 * More fixes have been made to treatment of floating point exception
2441 treatment and other Unix signals. In particular, floating point
2442 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2443 * The detection and handling of control stack exhaustion (infinite
2444 or very deeply nested recursion) has changed. Stack exhaustion
2445 detection is now done by write-protecting pages at the OS level
2446 and applies at all optimization settings; when found, a
2447 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2448 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2449 be caught using IGNORE-ERRORS.
2450 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2451 correctly when transforming recursive LABELS functions to LETs.
2452 (thanks to Alexey Dejneka)
2453 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2454 that are names of constants or global variables.
2455 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2456 alien routines with docstrings.
2457 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2458 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2460 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2461 (thanks to Eric Marsden)
2462 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2463 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2464 * Bug fix: The SPARC backend can now compile functions involving
2465 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2466 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2467 to a functions expecting a PATHNAME-DESIGNATOR.
2468 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2469 lambda lists. (thanks to Alexey Dejneka)
2470 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2471 functions so that EXPORTing the name of the class doesn't cause
2472 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2473 to Antonio Martinez for reporting this)
2474 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2475 constant-folded without failing an assertion. (thanks to Einar
2476 Floystad Dorum for reporting this)
2477 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2478 types are now dealt with more correctly by the compiler.
2479 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2480 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2481 behaviour analogous to RENAME-FILE. This puts its behaviour more
2482 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2484 * The fasl file version number has changed again. (because of the
2485 bug fix involving the names of PCL MAKE-INSTANCE functions)
2487 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2488 * bug fix: Floating point exceptions are treated much more
2489 consistently on the x86/Linux and PPC/Linux platforms.
2490 * Array initialization with :INITIAL-ELEMENT is now much faster for
2491 cases when the compiler cannot open code the array creation, but
2492 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2493 array accesses have also seen a speed increase.
2494 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2495 specified. (This was at the root of some bad interactions between
2496 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2498 * bug fix: Internal error arguments for undefined functions are now
2499 computed correctly on the PPC/Linux platform.
2500 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2501 Raymond Toy's patch for CMU CL)
2502 * Support for the Solaris 9 operating environment has been included
2503 (thanks to Daniel Merritt)
2504 * A very ugly but hopefully complete draft of the missing FFI chapter
2505 of the manual has been created by reformatting the corresponding
2506 CMU CL manual chapter into (currently very ugly and incoherent)
2507 DocBook and bringing it up to date for SBCL behavior. Thus, the
2508 manual is now essentially complete, at least by my extreme
2509 once-and-only-once standards, whereby it's acceptable to refer to
2510 the doc strings of SB-EXT functions as the primary documentation.
2511 * The fasl file version number has changed again, due to cleanup of
2512 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2513 is no longer a static symbol.)
2515 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2516 * SBCL now builds with OpenMCL (version 0.12) as the
2517 cross-compilation host; also, more progress has been made toward
2518 bootstrapping under CLISP.
2519 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2521 * bug 158 fixed: The compiler can now deal with integer loop
2522 increments different from 1; fixing this turned out also to fix
2524 * bug 169 fixed: no more bogus warnings about using lexical bindings
2525 despite the presence of perfectly good SPECIAL declarations (thanks
2526 to David Lichteblau)
2527 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2528 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2529 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2531 * bug 180 fixed: Method combination specifications no longer ignore
2532 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2533 * bug fix: Structure type predicate functions now check their argument
2534 count as they should.
2535 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2536 correctly. (thanks to Pierre Mai)
2537 * minor incompatible change: The --noprogrammer option is deprecated
2538 in favor of the new --disable-debugger option, which is very similar.
2539 (The major difference is that it takes effect at a slightly different
2540 time at startup, causing handling of errors in --sysinit and
2541 --userinit files will be affected differently.) The
2542 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2543 been added to allow this functionality to be controlled from ordinary
2544 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2545 maintainers, who might want to run non-interactive scripts to
2546 build SBCL cores which will later be used interactively.)
2547 * minor incompatible change: The LOAD function no longer, when given
2548 a wild pathname to load, loads all files matching that pathname.
2549 Instead, an error of type FILE-ERROR is signalled.
2551 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2552 * bug 147 fixed: The compiler preserves its block link/count
2553 invariants more correctly now so that it doesn't crash. (thanks
2555 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2557 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2558 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2559 no-value-for-this-character as NIL (as specified by ANSI).
2560 * HOST-NAMESTRING on physical pathnames now returns a string that is
2561 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2562 (thanks to Christophe Rhodes)
2563 * The Alpha port handles icache flushing more correctly. (thanks to
2565 * More progress has been made toward bootstrapping under CLISP. (thanks
2566 to Christophe Rhodes)
2567 * The fasl file format has changed again, because dynamic loading
2568 on OpenBSD (which has non-ELF object files) motivated some cleanups
2569 in the way that foreign symbols are transformed and passed around.
2570 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2571 is no longer treated as whitespace by the reader, but instead as
2572 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2573 |A
\7fB|, instead of A as it used to.
2575 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2576 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2577 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2578 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2579 this as of 0.7.1.45, but was left out of the previous news section
2580 (thanks to Dan Barlow)
2581 * SBCL now runs on the Solaris operating system on SPARC architectures
2582 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2583 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2584 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2585 * SPARC backend cleanups, allowing builds of cores optimized for V8
2586 and V9 SPARCS, and also emission of code targeted to a particular
2587 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2589 * SBCL is closer to bootstrapping under CLISP, thanks to various
2590 fixes by Christophe Rhodes.
2591 * The fasl file format has changed again, to allow the compiler's
2592 INFO database to support symbol macros.
2593 * The user manual (in doc/) is formatted into HTML more nicely.
2594 (thanks to coreythomas)
2595 * The system is smarter about SUBTYPEP relationships, especially
2596 those involving NOT types (including types such as ATOM which are
2597 represented internally using NOT types). Thus SUBTYPEP is less
2598 likely to return (VALUES NIL NIL) in general, and in particular
2599 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2601 * The fasl file format has changed again, because the internal
2602 representation of types now includes a new slot to support the new
2603 SUBTYPEP-of-NOT-types logic.
2604 * (not a change in the main branch of SBCL, but a related prototype
2605 which can hopefully be merged into the main branch of SBCL in the
2606 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2607 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2608 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2609 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2610 dispatch macro character combinations. (thanks to Alexey Dejneka)
2611 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2612 that superficially look like logical namestrings correctly.
2613 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2615 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2616 default supplied-p) for &optional and &key arguments. (thanks to
2619 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2620 * incompatible change: The compiler is now less aggressive about
2621 tail call optimization, doing it only when (> SPACE DEBUG) or
2622 (> SPEED DEBUG). (This is an incompatible change because there are
2623 programs which relied on the old CMU-CL-style behavior to optimize
2624 away their unbounded recursion which will now die of stack overflow.)
2625 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2626 for non-GENCGC systems has been increased to 20M (since that
2627 seems much closer to the likely performance optimum for modern
2628 systems than the old 4M value was)
2629 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2630 and *DEBUG-PRINT-LEVEL*
2631 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2632 of CMU CL's support for SPARC, and various endianness and other
2633 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2634 * new syntactic sugar for the Unix command line: --load foo.bar is now
2635 an alternate notation for --eval '(load "foo.bar")'.
2637 ** The system now detects stack overflow and handles it gracefully,
2638 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2639 optimization settings. (This is a good thing in general, and
2640 its introduction in this version should be particularly timely
2641 for anyone whose code fails because of suppression of tail
2643 ** The system now hunts for the C variable "environ" in a more
2644 devious way, to avoid segfaults when the C library version differs
2645 between compile time and run time. (thanks to Christophe Rhodes)
2646 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2647 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2648 ** The compiler no longer issues bogus style warnings for undefined
2649 classes in the same source file as the DEFCLASSes which defined
2650 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2652 ** fixes in CONDITION class precedence list for undefined function
2653 errors (thanks to Alexei Dejneka)
2654 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2655 correctly. (thanks to Dan Barlow)
2656 ** portability fixes aiming at bootstrapping under CLISP (thanks
2657 to Dave McDonald and Christophe Rhodes)
2658 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2659 ** fixes in type translation and and type inference (thanks to
2661 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2662 ** various fixes in the new ports (thanks to Dan Barlow)
2663 * several changes related to debugging:
2664 ** suppression of tail recursion, as noted above
2665 ** stack overflow detection, as noted above
2666 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2667 is now the default. (For some time encapsulation has been more
2668 reliable than the breakpoint-based :ENCAPSULATE NIL
2669 implementation, at least on X86 systems; and I just noticed that
2670 encapsulation also seems closer to the spirit of the ANSI
2673 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2675 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2676 up properly again. (There was a packaging bug in 0.7.0 which
2677 left their definitions in SB-SYS::LOAD-FOREIGN and
2678 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2679 vital for most things which interface to C-level interfaces,
2680 like extensions working with sockets or databases or
2681 Perl-compatible regexes or whatever, and the need to fix
2682 this bug is the main reason that 0.7.1 was released so
2684 ** DEFGENERIC is now choosier about the methods it redefines, so that
2685 reLOADing a previously-LOADed file containing DEFGENERICs does
2686 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2687 cycle now works as it should. (thanks to Alexey Dejneka)
2688 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2689 fixed by Christophe Rhodes. (He actually submitted this patch
2690 months ago, and I delayed until after 0.7.0.)
2691 ** Bug 111 (internal compiler confusion about runtime checks on
2692 FUNCTION types) was fixed by Alexey Dejneka.
2693 * Some internal cleanups (getting rid of variables which aren't
2694 needed now that the byte interpreter is gone) caused the fasl
2695 file format number to change again.
2697 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2698 * major incompatible change: The default fasl file extension, i.e. the
2699 default extension for files produced by COMPILE-FILE, has changed
2700 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2702 ** There are many changes in the implementation of the compiler.
2703 SBCL is now essentially a compiler-only implementation of ANSI
2704 Common Lisp. EVAL still "interprets" a few special cases, but
2705 almost all the interesting cases are handled by creating
2706 a LAMBDA expression, calling COMPILE on it, then calling
2707 FUNCALL on the result.
2708 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2709 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2710 Since the code is newer, there might still be some new bugs
2711 (though not as many as before Martin Atzmueller's fixes:-). But
2712 the new code is substantially simpler and clearer, and hopefully
2713 any remaining bugs will be simpler, less fundamental, and more
2714 fixable then the bugs in the old code.
2715 ** The revised compiler is still a little unsteady on its feet.
2717 *** The debugging information it produces (particularly the names
2718 of FUNCTION objects) is sometimes much less useful than what
2719 the old compiler produced.
2720 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2721 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2722 within a MACROLET) has been temporarily weakened.
2723 ** There are new compiler optimizations for various functions:
2724 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2725 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2726 *** the math functions TRUNCATE, FLOOR, and CEILING
2727 *** the function-of-all-trades COERCE
2728 Mostly these should be transparent, but there's one
2729 potentially-annoying problem (bug 117): when the compiler
2730 inline-expands a function and does type analysis on the result,
2731 it can create control paths which have type mismatches, and
2732 when it can't prove that those control paths aren't taken,
2733 it will issue WARNINGs about the type mismatches. This is
2734 a particular problem in practice for the new sequence functions.
2735 It's not clear how this should be fixed, and for now, a
2736 workaround is given in the entry for 117 in the BUGS file.
2737 ** (Because of the interaction between the two previous items --
2738 occasional inlining problems and new inline expansions -- some
2739 of the new sequence function optimizations won't really kick in
2740 completely until debugging information, and then inlining, are
2741 straightened out in some future version.)
2742 * minor incompatible changes:
2743 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2744 DIRECTORY no longer implicitly promotes NIL slots of its
2745 pathname argument to :WILD. In particular, when you ask for the
2746 contents of a directory (which you used to be able to do without
2747 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2748 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2749 ** changes in behavior that ANSI explicitly defines to be
2750 implementation dependent:
2751 *** The new compiler-only implementation still conforms with ANSI,
2752 but acts a little different than before. Besides the obvious
2753 changes in performance tradeoffs (that the cost per form passed
2754 to EVAL has gone up, and the cost per form executed by EVAL
2755 has gone down), the behavior of the system changes a little
2756 because there are no longer any interpreted function objects.
2757 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2758 e.g. doing COMPILE on the output of interactive DEFUN is
2760 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2762 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2763 has changed from (:CL) to NIL.
2764 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2765 e.g. #\Newline and #\Tab, don't have names specified in the
2766 ANSI Common Lisp standard, is now based on their ASCII symbolic
2767 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2768 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2769 are no longer used for output.
2770 ** changes in internal implementation constants:
2771 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2772 4 million. (If your application spends a lot of time GCing and
2773 you have a lot of RAM, you might want to experiment with
2774 increasing it even more.)
2775 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2776 However, almost all old code should still continue to work without
2777 immediate update, as SB-C-CALL is now a (deprecated) nickname
2779 ** Old operator names in the style DEF-FOO are now deprecated in
2780 favor of new corresponding names DEFINE-FOO, for consistency with
2781 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2782 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2783 internal symbols, but a few supported extensions like
2784 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2785 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2786 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2787 etc. as you get deeper into recursive calls to the debugger
2788 command loop, instead of the old "5]", "5]]", "5]]]"
2789 sequence. (I was motivated to do this when squabbles between
2790 ILISP and SBCL left me very deeply nested in the debugger. In the
2791 short term, this change will probably provoke more ILISP/SBCL
2792 squabbles, but hopefully it will be an improvement in the long run.)
2793 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2794 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2795 function, since declaiming return types involving aliens is
2796 (1) annoyingly messy to do by hand and (2) vital to efficient
2797 compilation of code which calls such functions.
2798 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2799 longer reexported by the SB-EXT package. They're solely useful
2800 for alien code, so it seems more logical that you should get
2801 them from the SB-ALIEN package, not in SB-EXT.
2802 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2803 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2804 features. Instead, the code that they used to control is always
2805 built into the system.
2806 * many other bug fixes
2807 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2808 advantage of the new EVAL-WHEN stuff and to clean them up in
2809 general, and they are now more ANSI-compliant in a number of
2810 ways. Martin Atzmueller is responsible for a lot of this.
2811 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2813 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2814 *** correct ERROR type for various file operations
2815 *** some fixes for Lisp streams
2816 *** DEFMETHOD syntax checking
2817 *** changing old weird representation of debug information as
2818 strings (which, among their other deficiencies, don't transform
2819 correctly when you rename packages, and don't change their
2820 print representation when you change things like *PACKAGE*
2821 and *PRINT-LENGTH*) to symbols and lists of symbols
2822 He also made several improvements and fixed several bugs in DESCRIBE.
2823 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2825 *** misbehavior of WRITE-STRING/WRITE-LINE
2826 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2827 and several other LOOP problems as well
2828 *** DIRECTORY when similar filenames are present
2829 *** DEFGENERIC with :METHOD options
2830 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2831 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2832 *** argument ordering in FIND with :TEST option
2833 *** mishandled package designator argument in APROPOS-LIST
2834 *** various problems in the backquote readmacro
2835 *** a bug in APROPOS
2836 *** probably some others that I'm not describing very well here,
2837 since the CVS log documents them by reference to sbcl-devel
2838 messages, and the SourceForge archives aren't working well.:-(
2839 ** Dan Barlow improved the Alpha port (and is making progress on the
2840 PPC port, for those of you who think different).
2841 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2842 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2843 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2844 continues to work on the SPARC port (for those of you in a position
2845 to look down upon our little PC-compatible boxes from a great height).
2846 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2847 and uses that copy, rather than the current dynamic value, when
2848 it's trying to decide whether to truncate output. Thus e.g.
2849 (let ((*print-lines* 50))
2850 (pprint-logical-block (stream nil)
2852 (let ((*print-lines* 8))
2853 (print (aref possiblybigthings i) stream)))))
2854 should now truncate the logical block only at 50 lines, instead of
2855 often truncating it at 8 lines, as it did before.
2856 * The doc/cmucl/ directory, containing old CMU CL documentation
2857 from the time of the fork, is no longer part of the base system.
2858 SourceForge has shut down its anonymous FTP service, and with it
2859 my original plan for distributing the old CMU CL documentation
2860 there. For now, if you need these files you can download an old
2861 SBCL source release and extract them from it.
2862 * The fasl file version number changed again, for dozens of reasons,
2863 some of which are apparent above.
2865 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2866 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2867 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2868 patch, so that DISASSEMBLE output is much nicer.
2869 * The code in the SB-PROFILE package now seems reasonably stable.
2870 I still haven't decided what the final interface should look like
2871 (I'd like PROFILE to interact cleanly with TRACE, since both
2872 facilities use function encapsulation) but if you have a need
2873 for profiling now, you can probably use it successfully with
2874 the current CMU-CL-style interface.
2875 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2876 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2877 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2878 link now returns the dangling link itself, and for similar
2879 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2880 link itself. (In these cases the old code signalled an error and
2881 looped endlessly, respectively.) Thus, DIRECTORY now works even
2882 in the presence of dangling and cyclic symbolic links.
2883 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2884 is now a supported extension again, since the consensus on
2885 sbcl-devel was that it can be useful for ordinary development
2886 work, not just for debugging SBCL itself.
2887 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2888 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2889 that currently-defined functions might be redefined later with
2890 different return types.
2891 * Hash tables can be printed readably, as inspired by CMU CL code
2892 of Eric Marsden and SBCL code of Martin Atzmueller.
2893 * better error handling in CLOS method combination, thanks to
2894 Martin Atzmueller porting Pierre Mai's CMU CL patches
2895 * more overflow fixes for >16Mbyte I/O buffers
2896 * A bug in READ has been fixed, so that now a single Ctrl-D
2897 character suffices to cause end-of-file on character streams.
2898 In particular, now you only need one Ctrl-D at the command
2899 line (not two) to exit SBCL.
2900 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2902 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2903 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2904 does the right thing.
2905 * The implementation of some type tests, especially for CONDITION
2906 types, is now tidier and maybe faster, due to CMU CL code
2907 originally by Douglas Crosher, ported by Martin Atzmueller.
2908 * Some math functions have been fixed, and there are new
2909 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2910 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2911 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2912 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2913 likely to be built into sbcl-0.7.0.)
2914 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2915 CL, ported to SBCL by Martin Atzmueller)
2916 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2917 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2918 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2919 * There's a new slam.sh hack to shorten the edit/compile/debug
2920 cycle for low-level changes to SBCL itself, and a new
2921 :SB-AFTER-XC-CORE target feature to control the generation of
2922 the after-xc.core file needed by slam.sh.
2923 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2924 COMPILE-FILE is no longer supported, so that now every function
2925 gets an entry point, so that block compilation looks a little
2926 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2927 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2928 returns the number of bytes consed since the system started,
2929 rather than the number consed since the first time the function
2930 was called. (The new definition parallels ANSI functions like
2931 CL:GET-INTERNAL-RUN-TIME.)
2932 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2933 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2934 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2935 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2936 * Fasl file version numbers are now independent of the target CPU,
2937 since historically most system changes which required version
2938 number changes have affected all CPUs equally. Similarly,
2939 the byte fasl file version is now equal to the ordinary
2942 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2943 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2944 is no longer recognized. I apologize for this, because it was
2945 listed in SB-EXT as a supported extension, but I found that
2946 its existing behavior was poorly specified, as well as incorrectly
2947 specified, and it looked like too much of a mess to straighten it
2948 out. I have enough on my hands trying to get ANSI stuff to work..
2949 * many patches ported from CMU CL by Martin Atzmueller, with
2950 half a dozen bug fixes in pretty-printing and the debugger, and
2951 half a dozen others elsewhere
2952 * fixed bug 13: Floating point infinities are now supported again.
2953 They might still be a little bit flaky, but thanks to bug reports
2954 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2955 as flaky as they were.
2956 * The --noprogrammer command line option is now supported. (Its
2957 behavior is slightly different in detail from what the old man
2958 page claimed it would do, but it's still appropriate under the
2959 same circumstances that the man page talks about.)
2960 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2961 are now supported, and enabled by default. Thus, the compiler can
2962 handle many floating point and complex operations much less
2963 inefficiently. (Thus e.g. you can implement a complex FFT
2965 * The compiler now detects type mismatches between DECLAIM FTYPE
2966 and DEFUN better, and implements CHECK-TYPE more correctly, and
2967 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2968 thanks to patches from Martin Atzmueller.
2969 * various fixes to make the cross-compiler more portable to
2970 ANSI-conforming-but-different cross-compilation hosts (notably
2971 Lispworks for Windows, following bug reports from Arthur Lemmens)
2972 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2973 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2974 fixed thanks to Pierre Mai's CMU CL patches.
2975 * fixes to keep the system from overflowing internal counters when
2976 it tries to use i/o buffers larger than 16M bytes
2977 * fixed bug 45a: Various internal functions required to support
2978 complex special functions have been merged from CMU CL sources.
2979 (When I was first setting up SBCL, I misunderstood a compile-time
2980 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2981 * improved support for type intersection and union, fixing bug 12
2982 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2983 more obscure bugs as well
2984 * some steps toward byte-compiling non-performance-critical
2985 parts of the system, courtesy of patches from Martin Atzmueller
2986 * Christophe Rhodes has made some debian packages of sbcl at
2987 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2988 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2989 stable, but are nonetheless usable. When he's ready, I'd be happy
2990 to add them to the SourceForge "File Releases" section. (And if
2991 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2992 * new fasl file format version number (because of changes in
2993 internal representation of (OR ..) types to accommodate the new
2994 support for (AND ..) types, among other things)
2996 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2997 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2999 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3001 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3002 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3003 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3004 * DESCRIBE now works on structure objects again.
3005 * Most function call argument type mismatches are now handled as
3006 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3007 know whether the function will be redefined before the call is
3008 executed. (The compiler could flag local calls with full WARNINGs,
3009 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3010 it doesn't keep track of enough information to know whether calls
3011 are local in this sense.)
3012 * Compiler output is now more verbose, with messages truncated
3013 later than before. (There should be some supported way for users
3014 to override the default verbosity, but I haven't decided how to
3015 provide it yet, so this behavior is still controlled by the internal
3016 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3017 src/compiler/ir1util.lisp.)
3018 * Fasl file format version numbers have increased again, because
3019 support for the Gray streams extension changes the layout of the
3020 system's STREAM objects.
3021 * The Gray subclassable streams extension now works, thanks to a
3022 patch from Martin Atzmueller.
3023 * The full LOAD-FOREIGN extension (not just the primitive
3024 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3025 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3026 but like most other programs, it defaults to copying the Unix
3027 environment from the original process instead of starting the
3028 new process in an empty environment.
3029 * Extensions which manipulate the Unix environment now support
3030 an :ENVIRONMENT keyword option which doesn't smash case or
3031 do other bad things. The CMU-CL-style :ENV option is retained
3032 for porting convenience.
3033 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3034 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3036 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3038 * A patch from Martin Atzmueller seems to have solved the SIGINT
3039 problem, and as far as we know, signal-handling now works cleanly.
3040 (If you find any new bugs, please report them!)
3041 * The system no longer defaults Lisp source file names to types
3042 ".l", ".cl", or ".lsp", but only to ".lisp".
3043 * The compiler no longer uses special default file extensions for
3044 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3045 seems to expect a single default extension for all compiled code,
3046 and there's no compelling reason to try to stretch the standard
3047 to allow two different extensions.) Instead, byte-compiled files
3048 default to the same extension as native-compiled files.
3049 * Fasl file format version numbers have increased again, because
3050 a rearrangement of internal implementation packages made some
3051 dumped symbols in old fasl files unreadable in new cores.
3052 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3053 many fewer weird special cases.
3054 * Bug #17 (differing COMPILE-FILE behavior between logical and
3055 physical pathnames) has been fixed, and some related misbehavior too,
3056 thanks to a patch from Martin Atzmueller.
3057 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3058 by Tim Moore, ported to SBCL by Martin Atzmueller.
3059 * Martin Atzmueller fixed several filesystem-related problems,
3060 including bug #36, in part by porting CMU CL patches, which were
3061 written in part by Paul Werkowski.
3062 * More compiler warnings in src/runtime/ are gone, thanks to
3063 more patches from Martin Atzmueller.
3064 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3067 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3069 * DESCRIBE now works on CONDITION objects.
3070 * The debugger now handles errors which arise when trying to print
3071 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3073 * The build system now uses an additional file, customize-target-features.lisp,
3074 to allow local modifications to the target *FEATURES* list. (The point of
3075 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3076 debugging features enabled, without having to worry about propagating it
3077 into everyone's system when I do a "cvs update".) When no
3078 customize-target-features.lisp file exists, the target *FEATURES* list
3079 should be constructed the same way as before.
3080 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3081 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3082 warn about behavior which is undefined under the ANSI spec. Note: This
3083 is specified by ANSI, but it's not very popular with programmers.
3084 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3085 macro in the SBCL sources for an example of a workaround which you
3086 might use to make portable ANSI-standard code which does what you want.
3087 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3088 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3089 when it's used as a non-toplevel form. (This is required in order
3090 to implement the DEFCONSTANT-EQX macro.)
3091 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3092 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3093 funny things, due to bugs in the implementation of EVAL-WHEN
3094 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3095 non-toplevel DEFCONSTANTs are uncommon.)
3096 * The core file version number and fasl file version number have been
3097 incremented, because the old noncompliant DEFCONSTANT behavior involved
3098 calling functions which no longer exist, and because I also took the
3099 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3100 * fixed bug 1 (error handling before read-eval-print loop starts), and
3101 redid debugger restarts and related debugger commands somewhat while
3103 ** The QUIT debugger command is gone, since it did something
3104 rather different than the SB-EXT:QUIT command, and since it never
3105 worked properly outside the main toplevel read/eval/print loop.
3106 Invoking the new TOPLEVEL restart provides the same functionality.
3107 ** The GO debugger command is also gone, since you can just invoke
3108 the CONTINUE restart directly instead.
3109 ** The TOP debugger command is also gone, since it's redundant with the
3110 FRAME 0 command, and since it interfered with abbreviations for the
3112 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3114 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3115 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3116 some porting work by Martin Atzmueller)
3117 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3118 with a boatload of patches which he ported to SBCL. Now that those
3120 ** The system tries to make sure that its low-priority messages
3121 are prefixed by semicolons, to help people who like to use
3122 syntax highlighting in their ILISP buffer. (This patch
3123 was originally due to Raymond Toy.)
3124 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3125 patches originally written by Raymond Toy.
3126 ** The compiler understands coercion between single-value and
3127 multiple-VALUES type expressions better, getting rid of some very
3128 weird behavior, thanks to patches originally by Robert MacLachlan
3129 and Douglas Crosher.
3130 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3131 lambda lists, thanks to a patch originally by Pierre Mai.
3132 ** The system no longer bogusly warns about "abbreviated type
3134 ** The compiler gets less confused by inlining and RETURN-FROM,
3135 thanks to some patches originally by Tim Moore.
3136 ** The system no longer hangs when dumping circular lists to fasl
3137 files, thanks to a patch originally from Douglas Crosher.
3138 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3139 undefined function error.
3140 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3141 in structure slot initforms) for now, documented workaround instead:-|
3142 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3143 * fixed bug 5: added stubs for various Gray stream functions called
3144 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3145 installed, at least appropriate type errors are generated
3146 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3147 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3148 consistently in DEFMETHOD forms.
3149 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3150 it doesn't seem to affect SBCL after all
3151 * The C runtime system now builds with better optimization and many
3152 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3154 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3156 * The system is now under CVS at SourceForge (instead of the
3157 CVS repository on my home machine).
3158 * The new signal handling code has been tweaked to treat register
3159 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3160 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3161 downstream, e.g. in debug-int.lisp, has implicit dependencies
3162 on the unsignedness of integer representation of machine words,
3163 and that caused the system to bomb out with infinite regress
3164 when trying to recover from type errors involving signed values,
3165 e.g. (BUTLAST '(1 2 3) -1).)
3166 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3167 (This was one of the bugs Peter Van Eynde reported back in July.)
3168 * The system now uses code inspired by Colin Walters' O(N)
3169 implementation of MAP (from the cmucl-imp@cons.org mailing
3170 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3171 inline the MAP operation, and there is more than one
3172 sequence argument to the MAP call (so that it can't just
3173 do ETYPECASE once and for all based on the type of the
3174 single sequence argument). (The old non-inline implementation
3175 of the general M-argument sequence-of-length-N case required
3176 O(M*N*N) time when any of the sequence arguments were LISTs.)
3177 * The QUIT :UNIX-CODE keyword argument has been renamed to
3178 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3179 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3181 * Raymond Wiker's patches to port dynamic loading from Linux to
3182 FreeBSD have been added.
3183 * The BUGS file is now more nearly up to date, thanks in large part
3184 to Martin Atzmueller's review of it.
3185 * The debugger now flushes standard output streams before it begins
3186 its output ("debugger invoked" and so forth).
3187 * The core version number and fasl file version number have both
3188 been incremented, because of incompatible changes in the layout
3190 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3191 used to not be called for a saved Lisp image.)
3192 * Martin Atzmueller's version of a patch to fix a compiler crash,
3193 as posted on sbcl-devel 13 September 2000, has been installed.
3194 * Instead of installing Martin Atzmueller's patch for the
3195 compiler transform for SUBSEQ, I deleted the compiler transform,
3196 and transforms for some similar consing operations.
3197 * A bug in signal handling which kept TRACE from working on OpenBSD
3199 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3200 away by constant folding
3201 * The system now defines its address space constants in one place
3202 (in the Lisp sources), and propagates them automatically elsewhere
3203 (through GENESIS and the sbcl.h file). Therefore, patching the
3204 address map is less unnecessarily tedious and error-prone. The
3205 Lisp names of address space constants have also been systematized.
3206 * CVS tags like dollar-Header-dollar have been removed from
3207 the sources, because they have never saved me trouble and
3208 they've been source of trouble working with patches and other
3209 diff-related operations.
3210 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3211 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3213 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3215 * The system has been ported to OpenBSD.
3216 * The system now compiles with a simple "sh make.sh" on the systems
3217 that it's supported on. I.e., now you no longer need to tweak
3218 text in the target-features.lisp-expr and symlinks in src/runtime/
3219 by hand, the make.sh takes care of it for you.
3220 * The system is no longer so grossly inefficient when compiling code
3221 involving vectors implemented as general (not simple) vectors (VECTOR T),
3222 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3223 which dares to use the various sequence functions on non-simple
3224 vectors, takes less of a performance hit.
3225 * There is now a primitive type predicate VECTOR-T-P
3226 to test for the (VECTOR T) type, so that e.g.
3227 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3228 can now be compiled with some semblance of efficiency. (The old code
3229 turned the type declaration into a full call to %TYPEP at runtime!)
3230 * AREF on (VECTOR T) is still not fast, since it's still compiled
3231 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3232 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3233 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3234 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3235 without making full calls to SUBTYPEP at runtime.
3236 (Some analogous efficiency issues for non-simple vectors specialized to
3237 element types other than T, or for non-simple multidimensional arrays,
3238 have not been addressed. They could almost certainly be handled the
3239 same way if anyone is motivated to do so.)
3240 * The changes in array handling break binary compatibility, so
3241 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3242 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3243 instead of (VALUES T T).
3244 * By following the instructions that Dan Barlow posted to sbcl-devel
3245 on 2 July 2000, I was able to enable primitive dynamic object
3246 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3247 functionality is not implemented (since it calls ld to resolve
3248 library references automatically, requiring RUN-PROGRAM for its
3249 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3250 try to resolve library references) is now supported.
3251 * The system now flushes the standard output streams when it terminates,
3252 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3253 them at several other probably-convenient times, e.g. in each pass of
3254 the toplevel read-eval-print loop, and after evaluating a form given
3255 as an "--eval" command-line option. (These changes were motivated by a
3256 discussion of stream flushing issues on cmucl-imp in August 2000.)
3257 * The source transform for TYPEP of array types no longer assumes
3258 that an array whose element type is a not-yet-defined type
3259 is implemented as an array of T, but instead punts, so that the
3260 type will be interpreted at runtime.
3261 * There is now some support for cross-compiling in make.sh: each of
3262 the phases of make.sh has its own script. (This should be transparent
3263 to people doing ordinary, non-cross-compile builds.)
3264 * Since my laptop doesn't have hundreds of megabytes of memory like
3265 my desktop machine, I became more motivated to do some items on
3266 my to-do list in order to reduce the size of the system a little:
3267 ** Arrange for various needed-only-at-cold-init things to be
3268 uninterned after cold init. To support this, those things have
3269 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3270 symbols with such names are now uninterned after cold init).
3271 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3272 fasl files for cold load.
3273 ** Remove the old compiler structure pooling code (which used to
3274 be conditional on the target feature :SB-ALLOC) completely.
3275 ** Redo the representation of some data in cold init to be more compact.
3276 (I also looked into supporting byte compiled code at bootstrap time,
3277 which would probably reduce the size of the system a lot, but that
3278 looked too complicated, so I punted for now.)
3279 * The maximum signal nesting depth in the src/runtime/ support code has
3280 been reduced from 4096 to 256. (I don't know any reason for the very
3281 large old value. If the new smaller value turns out to break something,
3282 I'll probably just bump it back up.)
3283 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3285 * Many, many bugs reported by Peter Van Eynde have been added to
3286 the BUGS list; some have even been fixed.
3287 * While enabling dynamic object file loading, I tried to make the
3288 code easier to understand, renaming various functions and variables
3289 with less ambiguous names, and changing some function calling
3290 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3291 * While trying to figure out how to do the OpenBSD port, I tried to
3292 clean up some of the code in src/runtime/. In particular, I dropped
3293 support for non-POSIX signal handling, added various comments,
3294 tweaked the code to reduce the number of compilation warnings, and
3295 renamed some files to increase consistency.
3296 * To support the new automatic configuration functionality in make.sh,
3297 the source file target-features.lisp-expr has been replaced with the
3298 source file base-target-features.lisp-expr and the machine-generated
3299 file local-target-features.lisp-expr.
3300 * fixed a stupid quoting error in make.sh so that using CMU CL
3301 "lisp -batch" as cross-compilation host works again
3303 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3305 * DESCRIBE no longer tries to call itself recursively to describe
3306 bound/fbound values, so that it no longer fails on symbols which are
3307 bound to themselves (like keywords, T, and NIL).
3308 * DESCRIBE now works on generic functions.
3309 * The printer now prints less-screwed-up representations of closures
3310 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3311 * A private symbol is used instead of the :EMPTY keyword previously
3312 used to mark empty slots in hash tables. Thus
3313 (DEFVAR *HT* (MAKE-HASH-TABLE))
3314 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3315 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3316 now does what ANSI says that it should. (You can still get
3317 similar noncompliant behavior if bang on the hash table
3318 implementation with all the symbols you get back from
3319 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3320 This breaks binary compatibility, since tests for equality to
3321 :EMPTY are wired into things like the macroexpansion of
3322 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3324 * There's now a minimal placeholder implementation for CL:STEP,
3325 as required by ANSI.
3326 * An obscure bug in the interaction of the normal compiler, the byte
3327 compiler, inlining, and structure predicates has been patched
3328 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3329 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3331 * Missing ordinary arguments in a macro call are now detected even
3332 when the macro lambda list contains &KEY or &REST.
3333 * The debugger no longer complains about encountering the top of the
3334 stack when you type "FRAME 0" to explicitly instruct it to go to
3335 the top of the stack. And it now prints the frame you request even
3336 if it's the current frame (instead of saying "You are here.").
3337 * As specified by ANSI, the system now always prints keywords
3338 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3340 * The default initial SIZE of HASH-TABLEs is now smaller.
3341 * Type information from CLOS class dispatch is now propagated
3342 into DEFMETHOD bodies, so that e.g.
3343 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3345 is now basically equivalent to
3346 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3347 (DECLARE (TYPE SINGLE-FLOAT X))
3349 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3350 operation, without having to do run-time type dispatch.
3351 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3352 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3353 * Since I don't seem to be making big file reorganizations very often
3354 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3355 is overflowing my ability to conveniently back them up), I've finally
3356 checked the system into CVS. (The CVS repository is on my home system,
3357 not at SourceForge -- putting it on SourceForge might come later.)
3358 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3359 high-level GC-NOTIFY-FOO functions send their output. (There's
3360 still very little control of where low-level verbose GC functions
3361 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3362 less than it used to -- the GC-NOTIFY-FOO functions are now under
3363 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3364 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3365 in only one place in the source code, and propagates it automatically
3366 everywhere that it's needed. Thus e.g. when I bump the version from
3367 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3368 * The C source files now include boilerplate legalese and documentation
3369 at the head of each file (just as the Lisp source files already did).
3370 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3371 to his page will be replaced with a link to his new CLiki service.
3373 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3375 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3376 * The build process now looks for GNU make under the default name "gmake",
3377 instead of "make" as it used to. If GNU make is not available as "gmake"
3378 on your system, you can change this default behavior by setting the
3379 GNUMAKE environment variable.
3380 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3381 can build without error under CMU CL.
3383 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3385 * There is now a partial SBCL user manual (with some new text and some
3386 text cribbed from the CMU CL manual).
3387 * The beginnings of a profiler have been added (starting with the
3388 CMU CL profiler and simplifying and cleaning up). Eventually the
3389 main interface should be through the TRACE macro, but for now,
3390 it's still accessed through vaguely CMU-CL-style functions and macros
3391 exported from the package SB-PROFILE.
3392 * Some problems left over from porting CMU CL to the new
3393 cross-compilation bootstrap process have been cleaned up:
3394 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3395 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3396 part when getting rid of the compiler *BACKEND* stuff.)
3397 ** The value of *NULL-TYPE* was screwed up, because it was
3398 being initialized before the type system knew the final
3399 definition of the 'NULL type. This screwed up several key
3400 optimizations in the compiler, causing inefficiency in all sorts
3401 of places. (I found it because I wanted to understand why
3402 GET-INTERNAL-RUN-TIME was consing.)
3403 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3404 PROCLAIM FTYPE information. Unfortunately this broke binary
3405 compatibility again, since now the forms output by DEFGENERIC
3406 to refer to functions which didn't exist in 0.6.3.
3407 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3408 can use the new (as of 0.6.3) transform for SOME into MAP into
3410 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3411 values to (INTEGER 0 1000000), so that the time code will no longer
3412 occasionally get blown up by Linux returning 1000000 microseconds
3413 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3414 its output conform to the ANSI spec. (Alas, this makes its output
3415 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3416 * A full call to MAP NIL with a single sequence argument no longer conses.
3417 * fixes to problems pointed out by Martin Atzmueller:
3418 * The manual page no longer talks about multiprocessing as though
3419 it were currently supported.
3420 * The ILISP support patches have been removed from the distribution,
3421 because as of version 5.10.1, ILISP now supports SBCL without us
3422 having to maintain patches.
3423 * added a modified version of Raymond Toy's recent CMU CL patch for
3424 EQUALP comparison of HASH-TABLE
3426 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3428 * The system still can't cross-compile itself with
3429 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3430 entails), but at least it can compile more of itself that way
3431 than it used to be able to, and various buglets which were uncovered
3432 by trying to cross-compile itself that way have now been fixed.
3433 * This release breaks binary compatibility again. This time
3434 at least I've incremented the FASL file format version to 2, so that the
3435 problem can be detected reliably instead of just causing weird errors.
3436 * various new style warnings:
3437 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3438 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3439 ** using the lexical binding of a variable named in the *FOO* style
3440 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3441 as specified by ANSI.
3442 * *RANDOM-STATE* is no longer automatically initialized from
3443 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3444 default behavior of the system is to repeat its behavior every time
3445 it's run. If you'd like to change this behavior, you can always
3446 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3447 old convention there was no comparably easy way to get the system to
3448 repeat its behavior every time it was run.
3449 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3450 FTYPE declarations has been removed, in favor of their ANSI
3451 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3452 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3453 the types of their sequence arguments can be determined at compile time.
3454 This is done through a new open code expansion for MAP which eliminates
3455 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3456 argument types can be determined at compile time.
3457 * The optimizer now transforms COERCE into an identity operation when it
3458 can prove that the coerced object is already of the correct type. (This
3459 can be a win for machine generated code, including the output of other
3460 optimization transforms, such as the MAP transform above.)
3461 * Credit information has been moved from source file headers into CREDITS.
3462 * Source file headers have been made more standard.
3463 * The CASE macro now compiles without complaining even when it has
3466 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3468 * (Note that the way that the PCL macroexpansions were rewritten
3469 to accommodate the change in DEFGENERIC below breaks binary
3470 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3471 not run under sbcl-0.6.2. Once we get out of alpha releases,
3472 i.e. hit release 1.0.0, we'll probably try to maintain binary
3473 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3474 and sbcl-1.4.4. Until then, however, it might be fairly common
3475 for maintenance releases to break binary compatibility.)
3476 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3478 * The reporting of unhandled signals has been changed to print some
3479 explanatory text as well as the report form. (Previously only
3480 the report form was printed.)
3481 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3482 it defines, so that the compiler no longer issues undefined function
3483 warnings for compiled-but-not-yet-loaded generic functions.
3484 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3485 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3486 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3487 section of the ANSI spec are supported.
3488 * The "" nickname for the "KEYWORD" package has been removed.
3489 The reader still handles symbol tokens which begin with a package marker
3490 as keywords, but it doesn't expose its mechanism for doing so in the
3491 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3492 * The system now issues STYLE-WARNINGs for contradictory TYPE
3493 proclamations. (Warnings for contradictory FTYPE proclamations would
3494 be nice too, but those can't be done usefully unless the type system
3495 is made smarter about FUNCTION types.)
3496 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3497 systematized, so that "*target-*.lisp is supposed to exist only on the
3498 target and imply that there's a related file which exists on the
3499 host, and *host-*.lisp is supposed to exist only on the host and imply
3500 that there's a related file which exists on the target. This involves a
3501 lot of renaming. Hopefully the acute confusion caused by the renaming
3502 will be justified by the reduction in chronic confusion..
3503 ** runtime-type.lisp -> early-target-type.lisp
3504 ** target-type.lisp -> late-target-type.lisp
3505 ** early-host-format.lisp -> early-format.lisp
3506 ** late-host-format.lisp -> late-format.lisp
3507 ** host-error.lisp -> misc-error.lisp
3508 ** early-error.lisp -> early-target-error.lisp
3509 ** late-error.lisp -> late-target-error.lisp
3510 ** host-defboot.lisp -> early-defboot.lisp
3511 ** code/misc.lisp -> code/target-misc.lisp
3512 ** code/host-misc.lisp -> code/misc.lisp
3513 ** code/numbers.lisp -> code/target-numbers.lisp
3514 ** code/early-numbers.lisp -> numbers.lisp
3515 ** early-host-type.lisp -> early-type.lisp
3516 ** late-host-type.lisp -> late-type.lisp
3517 ** host-typep.lisp -> typep.lisp
3518 ** load.lisp -> target-load.lisp
3519 ** host-load.lisp -> load.lisp
3520 ** host-disassem.lisp -> disassem.lisp
3521 ** host-insts.lisp -> insts.lisp
3522 ** byte-comp.lisp -> target-byte-comp.lisp
3523 ** host-byte-comp.lisp -> byte-comp.lisp
3524 ** host-signal.lisp -> signal.lisp
3525 ** host-defstruct.lisp -> defstruct.lisp
3526 ** late-target-type.lisp -> deftypes-for-target.lisp
3527 Furthermore, several other previously target-only files foo.lisp (e.g.
3528 hash-table.lisp and random.lisp) have been split into a target-and-host
3529 foo.lisp file and a target-only target-foo.lisp file, with their key type
3530 definitions in the target-and-host part, so that the cross-compiler will
3531 know more about target types.
3532 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3533 gone away. In their place are various *BACKEND-FOO* variables
3534 corresponding to the slots of the old structure.
3535 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3536 into the target SBCL has been fixed.
3537 * The chill.lisp system for loading cold code into a running SBCL
3539 * Support for the CMU CL "scavenger hook" extension has been removed.
3540 (It was undocumented and unused in the CMU CL sources that SBCL was
3541 derived from, and stale in sbcl-0.6.1.)
3542 * Various errors in the cross-compiler type system were detected
3543 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3544 (enabling various consistency checks). Many of them were fixed,
3545 but some hard problems remain, so the compiler is back to
3546 running without *TYPE-SYSTEM-INITIALIZED* for now.
3547 * As part of the cross-compiler type system cleanup, I implemented
3548 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3549 * I have started adding UNCROSS calls throughout the type system
3550 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3551 on cross-compiler input files will be able to go away, and various
3553 * CONSTANTP now returns true for quoted forms (as explicitly required
3556 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3558 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3559 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3560 (MAKE-LIST -1). In the longer term, it ought to become true
3561 that declarations are assertions even at SAFETY 1. For now, it's not
3562 quite true even at SAFETY 3, but it's at least more nearly true..
3563 (Note that this change seems to increases the size of the system by
3564 O(5%) and to decrease the speed of the compiler by 20% or more.)
3565 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3566 for the problem of printing dozens of lines of distracting information
3567 about low-level system machinery as part of the top stack frame
3568 on entry to the debugger when an undefined function was called.
3569 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3570 is not reset to COMMON-LISP-USER.
3571 * Compilation of stuff related to dyncount.lisp has been made conditional
3572 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3573 smaller. The various dyncount-related symbols have been moved into
3574 a new "SB-DYNCOUNT" package.
3575 * tty-inspect.lisp has been renamed to inspect.lisp.
3576 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3577 feature has gone away. (When we eventually port to other flavors of
3578 libc and/or Unix, we'll try to make the differences between flavors
3579 invisible at the user level.)
3580 * Various other *FEATURES* tags, and/or their associated conditionals,
3581 have been removed if obsolescent, or given better documentation, or
3582 sometimes given more-mnemonic names.
3584 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3586 * tidied up "make.sh" script
3587 * tidied up system directory structure
3588 * better "clean.sh" behavior
3589 * added doc/FOR-CMUCL-DEVELOPERS
3590 * many many small tweaks to output format, e.g. removing possibly-confusing
3591 trailing #\. character in DESCRIBE-INSTANCE
3592 * (EQUALP #\A 'A) no longer signals an error.
3593 * new hashing code, including EQUALP hashing
3594 * tidied up Lisp initialization and toplevel
3595 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3596 * command line argument processing
3597 * added POSIX-GETENV function to deal with Unix-ish environment variables
3598 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3599 terminating SBCL on EOF
3600 * non-verbose GC by default
3601 * There is no more "sbcl" shell script; the sbcl file is now the C
3602 runtime executable (just like CMU CL).
3603 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3605 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3606 symbol and package information as Lisp native symbol and package objects
3607 instead of strings naming symbols and strings naming packages. This way,
3608 whenever packages are renamed (as in warm init), debug information is
3609 transformed along with everything else.
3610 * tweaked the optimization policy declarations which control the building
3611 of SBCL itself. Now, among other things, the system no longer saves
3612 source location debugging information. (This helps two problems at once
3613 by reducing SBCL size and by keeping SBCL from trying to look for its
3614 sources -- which may not exist -- when reporting errors.)
3615 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3616 debugging and testing purposes
3617 * cleaned up printing, making the printer call PRINT-OBJECT for
3618 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3619 methods, giving nearly-ANSI behavior
3620 * converted almost all special variables to use *FOO* naming convention
3621 * deleted PARSE-TIME functionality, since it can be done portably
3622 * moved some files out of cold init into warm init
3623 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3625 * regularized formatting of source files
3626 * added an install.sh script
3627 * fixed ridiculous memory usage of cross-compiler by making
3628 compiler/alloc.lisp not try to do pooling unless it can hook
3629 itself into the GC of the cross-compilation host. Now the system
3630 builds nicely on my old laptop.
3631 * added :SB-ALLOC in target-features.lisp-expr
3632 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3633 was not implemented)
3634 * re-did condition handling and note reporting in the compiler. Notes
3635 are no longer handled by signalling conditions. Style warnings
3636 and warnings are handled more correctly and reported in such a way
3637 that it's easy to find one or the other in your output (so that you
3638 can e.g. figure out which of many problems caused COMPILE-FILE to
3640 * changed the severity of several compiler warnings from full WARNING
3641 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3642 compiler note reporting so that it doesn't use the condition system
3643 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3644 COMPILE-FILE command)
3645 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3646 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3647 no longer supported, since their implementation was deeply intertwingled
3648 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3649 * removed bogus "support" for compiler macros named (SETF FOO), and
3650 removed the compiler macro for SETF INFO (but only after making a fool
3651 of myself on the cmucl-imp mailing list by posting a bogus patch for
3652 DEFINE-COMPILER-MACRO..)
3653 * Compiled files containing forms which have side effects on the Lisp
3654 reader (such as DEFPACKAGE forms) are now handled more correctly.
3655 (Compiler queuing of top level lambdas has been suppressed by setting
3656 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3657 * deleted various currently-unused source files, e.g. gengc.lisp. They
3658 may be added back at some point e.g. when porting to other architectures,
3659 but until they are it's distracting to distribute them and to try to
3661 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3662 there were so many of them they're just distractions, and UNCROSS is
3663 known to be able to handle the current sources
3664 * moved PROFILE functionality into TRACE, so that it will be clear
3665 how the wrapping and unwrapping of functions when you profile them
3666 interacts with the wrapping and unwrapping of functions when you
3667 trace them. (Actually, the functionality isn't there yet, but at least
3668 the interface specification is there. Hopefully, the functionality will
3669 arrive with some future maintenance release.)
3670 * removed host-oops.lisp
3671 * changed signature of QUIT function to allow UNIX-CODE argument
3672 * fixed READ-SEQUENCE bug
3673 * tweaked verbose GC output so that it looks more like the progress
3674 output that ANSI specifies for functions like LOAD
3675 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3676 * added <http://sbcl.sourceforge.com> to the banner information printed by