1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
3 * Enhancements for sbcl running on the Windows operating system:
4 ** (user-homedir-pathname) and default initialization file
5 locations now know about the user's "Documents and Settings"
6 directory (thanks to Yaroslav Kavenchuk)
7 ** run-program is implemented (thanks to Mike Thomas)
8 ** sockets support (thanks to Timothy Ritchey)
9 ** better backtrace support (thanks to Alastair Bridgewater)
10 ** sb-grovel supported
11 ** asdf-install and sb-posix work somewhat
12 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
13 * minor incompatible change: The reader no longer ignores errors
14 regarding non-existent packages in #+ and #- feature tests.
15 * new feature: command line options --no-sysinit, --no-userinit to
16 inhibit loading the corresponding init files
17 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
18 for sending data through UDP sockets (thanks to François-René Rideau)
19 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
20 error is signalled instead (thanks to François-René Rideau)
21 * minor incompatible change: Error signalling behaviour of lexical
22 operations violating package locks has changed slightly. Refer to
23 documentation on package locks for details.
24 * enhancement: EVAL can process IF-expressions without resorting to the
26 * bug fix: Some operations involving SIN, COS, or TAN weren't being
27 constant-folded properly due to missing out-of-line functions.
28 (reported by Mika Pihlajamäki)
29 * bug fix: LISTEN sometimes returned T even in cases where no data was
30 immediately available from the stream
31 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
32 were reversed. (reported by Levente Mészáros)
33 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
34 when given a vector displaced to an adjustable array. (reported
36 * bug fix: derivation of float boundaries from numbers outside the
37 appropriate float range (reported by John Wiseman)
38 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
40 * bug fix: precomputing cache entries for generic functions with
41 some subclasses of specializers as yet invalid does not attempt to
42 fill a cache line with a negative offset. (reported by Levente
44 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
45 allow condition class objects as arguments to DOCUMENTATION and
46 (SETF DOCUMENTATION); only find and set documentation for
47 structure names for the STRUCTURE doc-type. (suggested by Gary
49 * improvements to the Solaris/x86 port:
50 ** works on Solaris 11/Solaris Express
51 ** floating-point exception handling support
52 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
53 * fixed some bugs revealed by Paul Dietz' test suite:
54 ** REMOVE-METHOD returns its generic function argument even when
55 the method is not one of the generic functions' methods.
56 ** objects declared as MEMBER types can be admissible arguments to
58 ** printing characters should simply be printed by the FORMAT ~:C
60 ** compiler failure when compiling functions with hairy constant
61 defaults for optional parameters.
62 ** compiler produces wrong code when MAYBE-INLINE-expanding a
63 function, which is already optimized.
65 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
66 * new platform: experimental support for SBCL x86/Darwin, including
67 MacOS X 10.4.5 on Intel.
68 * new feature: Unicode character names are now known to the system
69 (through CHAR-NAME and NAME-CHAR).
70 * new feature: the filesystem location of SBCL's core file is
71 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
72 * minor incompatible change: the contrib modules SB-POSIX and
73 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
74 this change is to make it easier to distribute
75 location-independent binaries.
76 * bug fix: as implied by AMOP, standardized classes no longer have
77 slots named by external symbols of public packages. (reported by
79 * optimization: calling functions via constant symbols -- as in
80 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
81 via the function object as in (FUNCALL #'FOO).
82 * optimization: on x86 and x86-64, the calling convention for the
83 case of dealing with an unknown number of values has been altered
84 to be friendlier to the prediction heuristics implemented,
85 particularly with reference to CALL and RET pairing. (thanks to
87 * optimization: on x86, the code for access to thread-local symbol
88 values and binding/unbinding of thread-local symbols is smaller.
89 (thanks to Alastair Bridgewater)
90 * enhancement: CONSTANTP is now able to determine constantness of
91 more complex forms, including calls to constant-foldable standardized
92 functions and some special forms beyond QUOTE.
93 * fixed bug: occasional GC crashes on Solaris/x86
94 * optimization: x86-64 supports stack allocation of results of simple
95 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
96 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
97 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
98 (thanks to James Knight)
99 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
100 by the standard (thanks to Stephen Compall)
102 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
103 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
104 be used for bundling the runtime and the core file into one
105 executable binary. This feature is not currently supported on all SBCL
106 platforms. (thanks to James Bielman and NIIMI Satoshi)
107 * new feature: a generational or ephemeral garbage collector is now
108 the default on the PowerPC platform (both Linux and Darwin). The
109 old Cheney (stop and copy) collector is a build-time option.
110 (thanks to Cyrus Harmon, after Raymond Toy)
111 * minor incompatible change: the method by which SBCL finds its
112 contributed modules has changed; it no longer relies on symbolic
113 links from an $SBCL_HOME/systems directory, but searches directly
114 in the subdirectories of $SBCL_HOME.
115 * enhancement: the dynamic heap size on the Linux/PPC platform is
116 markedly larger, even using the older Cheney garbage collector.
117 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
118 certain complicated string types. (reported by Gary King)
119 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
120 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
121 * fixed bug: DOCUMENTATION on structures no longer signals an
122 error if no structure type of the right name exists. (reported by
124 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
125 applicable non-standard primary or :AROUND method on
126 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
127 no longer cause unbound variable errors. (reported by Kevin Reid)
128 * fixed bug: metacircle resolution in cases where methods have slots
129 added before the slots from STANDARD-METHOD. (reported by Jean
131 * fixed bug: the Power PC assembly code for calling into lisp
132 sometimes computed its return address wrongly (depending on the
133 15th bit of the address, fixed at link time). This used to
134 manifest itself in a segmentation violation while building PCL.
135 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
136 many others over the years)
137 * fixed bug: ROOM no longer reports silly numbers for stack usage if
138 the stack pointer (treated as a signed integer) is negative.
139 (thanks to Peter van Eynde)
141 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
142 * new platform: experimental support for the Windows operating
143 system has been added. (thanks to Alastair Bridgewater)
144 * fixed several bugs in and robustified the PPC FFI (including
145 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
146 * bug fix: allow non-simple string symbol names (reported by Paul
148 * bug fix: interrupt handling on NetBSD (thanks to Richard M
150 * bug fix: saving a core corrupted callbacks on x86/x86-64
151 * bug fix: closed a loophole in metacircularity detection and
152 grounding in the PCL implementation of CLOS.
153 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
154 (reported by tomppa on #lisp)
155 * bug fix: clear the direction flag on Lisp -> C transitions, as
156 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
157 (reported by Andrej Grozin and Hendrik Maryns)
158 * optimization: major improvements to GC efficiency on GENCGC platforms
159 * optimization: faster implementation of EQUAL
160 * optimization: emit more efficient opcodes for some common
161 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
162 * optimization: several other minor code-generation improvements on x86-64
163 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
165 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
166 * minor incompatible change: (SETF CLASS-NAME) and (SETF
167 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
168 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
169 by ANSI as a generic function, and for consistency (SETF
170 GENERIC-FUNCTION-NAME) is treated likewise.
171 * fixed bug #233.b: make constraint propagation notice when a variable
172 value is changed after it is referenced but before it is used
173 * fixed bug #296: no more arbitrary behaviour from filenames with
174 odd characters as --load arguments. (reported by Adam Warner)
175 * fixed bug #390: :CHARACTER-SET pathname components now work as
176 expected. (reported by Tim Daly Jr)
177 * fixed bug #391: complicated :TYPE intersections in slot
178 definitions no longer cause an error in PCL internals.
179 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
180 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
181 returning the number of octets which would be written to the
182 file-stream. (thanks to Robert J. Macomber)
183 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
184 arguments to the internal stream functions. (thanks to David
186 * bug fix: stack exhaustion is now handled on the x86/SunOS
187 platform. (thanks to Daisuke Homma)
188 * bug fix: more accurate ROOM results on GENCGC platforms
189 * optimization: improved type inference for arithmetic-for
190 index variables in LOOP
191 * optimization: faster floating-point SQRT on x86-64
192 * fixed some bugs revealed by Paul Dietz' test suite:
193 ** DOCUMENTATION returns NIL instead of "" for method combinations
194 that don't have a docstring
196 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
197 * minor incompatible change: (SETF CLASS-NAME) and (SETF
198 GENERIC-FUNCTION-NAME) are no longer generic functions, and
199 therefore are not customizeable by user code (as seems to be at
200 least permitted and maybe required by AMOP). As a consolation,
201 however, the SBCL implementation of these functions now calls
202 REINITIALIZE-INSTANCE as specified by AMOP.
203 * bug fix: slot-definition documentation is propagated as per ANSI
204 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
205 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
206 argument, as do eslotd initialization methods. (from Pascal
207 Costanza's "Closer" project)
208 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
209 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
211 * bug fix: it is now possible to have more than one subclass of
212 STANDARD-GENERIC-FUNCTION without causing stack overflow.
213 (reported by Bruno Haible, Pascal Costanza and others)
214 * bug fix: the dependent update protocol now works for generic
215 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
217 * bug fix: condition-class instances corresponding to
218 DEFINE-CONDITION forms are now created eagerly. (reported by
219 Kalle Olavi Niemitalo on comp.lang.lisp)
220 * bug fix: floating point printing is more accurate in some
221 circumstances. (thanks to Simon Alexander)
222 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
223 merged with *DEFAULT-PATHNAME-DEFAULTS*.
224 * bug fix: callbacks on OS X now preserve stack-alignment.
225 (thanks to Cyrus Harmon)
226 * enhancement: the x86-64 disassembler is much better at
227 disassembling SSE instructions. (thanks to Lutz Euler)
228 * enhancement: CHECK-TYPE on a variable now allows type-inference
229 (thanks to Vincent Arkesteijn)
230 * optimization: improved performance of EUC-JP external format.
231 (thanks to NIIMI Satoshi)
232 * optimization: performance improvements to IO on file streams of
233 :ELEMENT-TYPE CHARACTER
234 * optimization: much faster memory allocation on x86-64
235 * optimization: faster garbage collections (latency and throughput)
237 * optimization: faster bignum subtraction and division on x86 and x86-64
238 (thanks to Lutz Euler)
239 * optimization: more accurate type-inference for loops that use a
240 floating point index variable or a negative step.
242 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
243 * bug fix: add a workaround to SBCL looping infinitely at startup on
244 Linux kernels with apparently buggy implementations of personality().
245 (thanks to Svein Ove Aas)
246 * bug fix: Unicode symbols are correctly printed in LDB backtraces
247 (thanks to David Lichteblau)
248 * bug fix: local bindings shadow global symbol macros for
249 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
250 * optimization: non-open coded uses of numeric comparison operators
251 (e.g. >) no longer cons when called with more than one parameter
252 on platforms supporting dynamic-extent allocation.
253 * enhancement: saving cores with foreign code loaded is now supported
254 on MIPS/Linux in addition to the previously supported platforms.
255 * bug fix: FIND-RESTART now tests for activity, not applicability when given
256 a restart object as identifier. (reported by Helmut Eller for CMUCL)
257 * bug fix: division by zero in sb-sprof when no samples were collected
258 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
259 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
260 (thanks to Alexander Kjeldaas)
261 * new feature: ignoring the return values of destructive functions
262 like NREVERSE signals a compile-time style-warning.
263 (thanks to Kevin Reid)
264 * new feature: the alignment of alien structure fields can be
265 explicitly specified. (thanks to Cyrus Harmon)
266 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
267 and dump core on SIGQUIT
269 ** incompatible change: threads do not inherit values of specials
270 from their parents (see manual)
271 ** bug fix: threads stacks belonging to dead threads are freed by the
272 next exiting thread, no need to gc to collect thread stacks anymore
273 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
274 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
275 ** bug fix: dynamic variable and thread start related gc lossage
276 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
278 * fixed some bugs revealed by Paul Dietz' test suite:
279 ** SUBTYPEP is slightly more accurate on heinously complicated
280 CONS types where some of the members have uncertain (in the
281 NIL, NIL sense) type relationships to each other.
283 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
284 no longer purified unless :PURIFY T is explicitly specified.
285 ** Non-purified cores are significantly smaller than before
287 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
288 * new feature: timers based on Zach Beane's excellent timer package
289 * added support for the following external formats: koi8-u,
290 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
291 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
292 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
293 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
294 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
295 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
296 * incompatible change: a threaded SBCL will no longer revert to
297 non-threaded mode on non-NPTL systems, but refuse to start entirely.
298 * bug fix: interrupts are disabled until startup is complete; no
299 more sigsegvs when receiving a signal to soon
300 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
302 * bug fix: add a workaround for the memory randomization features in
303 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
304 workaround will only be in effect on systems with the proc filesystem
306 * bug fix: printing objects of type HASH-TABLE signals a
307 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
309 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
310 have been running for over 50 days. (reported by Gilbert Baumann)
311 * bug fix: the logic for getting names of functions gets less
312 confused when confronded with alternate-metaclass
313 funcallable-instances. (reported by Cyrus Harmon)
314 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
315 consistent, even on internal alternate-metaclass objects.
316 * bug fix: flush closure information collected by physical
317 environment analysis prepass before the main pass. (bug reported
319 * bug fix: compiler pack phase does not modify a hashtable, which is
320 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
321 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
322 may be expanded. (reported by Alexander <malishev> on c.l.l)
323 * bug fix: no more style-warnings when compiling code using LAST
324 under high-SPEED low-SPACE settings. (reported by David Wallin
326 * fixed some incompatibilities between SBCL's MOP and the MOP
328 ** the METAOBJECT class is now implemented;
329 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
330 STANDARD-OBJECT, as required;
331 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
332 now compatible (as required by VALIDATE-SUPERCLASS); there
333 remains a consistency requirement over the presence or absence
334 of the FUNCTION class in the superclass list of finalized
335 classes; see the manual for more details;
336 ** the :AROUND method for COMPUTE-SLOTS on
337 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
338 requested slot ordering.
340 ** bug fix: parent thread now can be gc'ed even with a live
342 ** bug fix: binding a special with PROGV to no value is not
343 the same as that symbol not having been bound (thanks to
345 ** bug fix: binding specials is thread safe (thanks to
347 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
348 and SB-SPROF do not die with 'no handler for signal XX in
349 interrupt_handle_now(..)' anymore
350 ** bug fix: WITH-TIMEOUT works with multiple running threads
351 * fixed some bugs revealed by Paul Dietz' test suite:
352 ** ENSURE-GENERIC-FUNCTION should take a method class object for
353 the :method-class keyword argument.
355 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
356 * new port: the Solaris operating system on x86 processors is now
357 mostly supported, though some rough edges in the environment
358 remain. (thanks to Daisuke Homma)
359 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
360 space for its heap. (thanks to Thiemo Seufer)
361 * optimization: SBCL on MIPS platforms now supports dynamic-extent
362 closures. (thanks to Thiemo Seufer)
363 * minor incompatible change: eof selects abort in the debugger.
364 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
365 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
367 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
368 deprecated, and will go away in a future revision of SBCL.
369 * minor incompatible change: GC-ON and GC-OFF are no longer
370 implemented with a counter, it does not matter how many times gc
371 is switched on or off
372 * bug fix: discriminating functions for generic function classes
373 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
374 make invalid assumptions about method precedence order. (reported
376 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
377 platforms. (thanks to Thiemo Seufer)
378 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
379 * bug fix: correct pathname printing with printer escaping is on.
380 (thanks to Kevin Reid)
381 * bug fix: complex VOP definitions in "user-space" no longer trigger
382 package locks. (reported by Zach Beane)
383 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
384 for funcallable-instance functions, and is no different from
386 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
387 characters in the current readtable. (reported by Nicholas Neuss)
388 * bug fix: SUBTYPEP on various CONS types returns more a more
389 accurate acknowledgment of its certainty.
390 * optimizations: REMOVE-DUPLICATES now runs in linear time on
391 lists in some cases. This partially fixes bug 384.
392 * flush all standard streams before prompting in the REPL and the
394 * bug fix: signal handling and triggering gc do not conflict
395 directly anymore, in particular a high frequency sb-sprof does
396 not prevent gc from running
397 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
398 approximation for timezone and DST information between the
399 universal time epoch and the smallest negative 32-bit time_t.
400 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
401 year 1899 so long as the encoded time is non-negative.
402 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
403 the correct answer for non-constant rotate amounts when performing
404 an inline 32-bit rotation.
406 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
407 there is only one thread in the session
408 ** bug fix: memory leak for streams created in one thread and
409 written to in another
410 ** bug fix: lockup when compiled with gcc4
411 ** bug fix: race that allows the gc to be triggered when gc is
413 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
414 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
416 ** bug fix: finished threads are gc'ed properly
417 * fixed some bugs revealed by Paul Dietz' test suite:
418 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
419 has arguments with a different set of applicable methods from
420 the orignal arguments.
421 ** The type error thrown by MAP now has a correct expected-type
423 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
424 name a compiled function.
425 ** Three MISC tests where a large bignum was improperly coerced to
426 a float that couldn't represent that bignum during type
427 derivation were fixed.
428 ** SUBTYPEP can now handle types involving the negation of a
429 list-form FUNCTION type.
430 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
431 as (COMPLEX (AND RATIO (NOT FIXNUM))).
432 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
434 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
435 * New feature: Experimental support for bivalent streams: streams
436 opened with :element-type :default now allow character and binary
437 (unsigned-byte 8) I/O
438 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
439 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
441 * bug fix: no more highly sporadic "couldn't check whether ~S is
442 readable" when reading a stream and an interrupt hits in the middle
443 of a select system call
444 * compiler better recognizes complex arrays (reported by Eduardo
446 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
447 for variables on SBCL built with linkage-tables. (reported by Luis
449 * various error reporting improvements.
450 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
451 (Thanks to Thiemo Seufer)
452 * bug fix: nested reader invokations work correctly
453 * bug fix: it is possible to have simultaneous references to foreign
454 code and foreign data with the same name.
456 ** added x86-64 support
457 ** incompatible change: the threading api now works with thread
458 objects instead of thread ids
459 ** bug fix: threads are protected from signals and interruption when
460 starting up or going down
461 ** bug fix: a race where an exiting thread could lose its stack to gc
462 ** bug fix: don't halt on infinite error in threads if possible
463 ** fixed numerous gc deadlocks introduced in the pthread merge
464 ** bug fix: fixed thread safety issues in read and print
465 ** bug fix: debugger doesn't hang on session lock if interrupted at
466 an inappropriate moment
467 ** bug fix: run-program is now thread safe(r)
468 ** bug fix: inner with-recursive-lock no longer releases the mutex
469 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
470 * fixed some bugs revealed by Paul Dietz' test suite:
471 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
472 EXPECTED-TYPE slots filled.
473 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
475 ** MAKE-CONDITION accepts classes as type-designators.
476 ** COMPILE may never return NIL.
477 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
478 range before calling Unix time functions
480 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
481 * numerous signal handling fixes to increase stability
482 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
483 * minor incompatible change: we now correctly canonize default
484 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
485 as per the MOP, rather than the historical (INITARG INITFUNCTION
486 INITFORM). (reported by Bruno Haible)
487 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
488 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
489 for more information.
490 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
491 pathname is a directory pathname.
492 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
493 * dynamic space size on PPC has been increased to 768Mb. (thanks to
495 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
496 :METACLASS argument in addition to a class name. (reported by
497 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
498 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
499 (reported by Bruno Haible; patch from Gerd Moellmann)
500 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
502 * bug fix: more cleanups to the floating point exception handling on
503 x86-64 (thanks to James Knight)
504 * bug fix: the compiler does not try to constant fold calls of
505 COERCE and BIT-* functions when they return freshly constructed
506 objects. (reported by Nikodemus Siivola and Paul Dietz)
507 * optimization: improved the MIPS versions of generic (in the
508 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
509 * optimization: direct conversion of (unsigned-byte 32) to floats on
510 the PowerPC platform.
511 * optimization: structure instances with raw slots now use less
512 memory, and probably show better memory locality. (thanks to
514 * optimization: DYNAMIC-EXTENT declarations for lists and closures
515 are treated as requests for stack allocation on the x86-64,
516 Alpha-32, PPC and SPARC platforms.
517 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
518 the wrong file descriptor; implementation of SOCKET-OPEN-P.
519 (thanks to Tony Martinez)
521 ** gcing a dead thread can no longer lead to lockups
522 ** threads block signals until they are set up properly
523 ** errno is no longer shared by threads
524 ** interrupt-thread restores the eflags register on x86
525 ** fixed some lockups due to gc/thread interaction
526 * fixed some bugs revealed by Paul Dietz' test suite:
527 ** invalid dotted lists no longer raise a read error when
529 ** signal an error if a symbol that names a declaration is used
530 as the name of a type, or vice versa
531 ** allow using the (declare (typespec var*)) abbreviation for
532 (declare (type typespec var*)) with all type specifiers
533 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
534 ** always return NIL from PPRINT-POP when OBJECT is NIL
535 ** don't signal errors when pretty-printing malformed LABELS,
536 FLET or MACROLET forms
537 ** declarations in a DOLIST body are also in scope for the
539 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
541 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
544 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
545 * fixed cross-compiler leakages that prevented building a 32-bit
546 target with a 64-bit host compiler.
547 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
548 opened with :IF-EXISTS :OVERWRITE.
549 * fixed bug 281: error for an invalid qualifier in a short-form method
550 combination method is not signalled until the faulty method is called.
551 * bug fix: iteration variable type inferrer failed to deal with open
552 intervals. (reported by Alan Shields)
553 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
554 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
555 or more runtime options were provided to the sbcl binary.
556 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
557 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
558 * optimization: there's now a fast-path for fixnum arguments in the
559 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
560 * optimization: the code generated on x86-64 is more compact thanks
561 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
562 * bug fix: floating-point exception handling now partly works on
563 x86-64. (Thanks to James Knight)
564 * improvement to the MIPS backend from Thiemo Seufer: C-style
565 64-bit long long arguments and return values to alien functions
567 * fixed some bugs revealed by Paul Dietz' test suite:
568 ** the type-error signalled from WARN has a filled-in DATUM slot.
569 ** the type-error required when a stream is not associated with
570 a file has the stream as its datum.
571 ** type-errors on single-floats on x86-64 no longer have
572 :INVALID-OBJECT as the datum
573 ** the type-errors signalled for invalid function names now have
574 a correct expected type
575 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
576 for structures defined with a :TYPE. Documentation strings for
577 typed structures are no longer immediately discarded
578 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
579 broadcast streams and synonym streams. FILE-LENGTH now also works
580 on broadcast streams.
582 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
583 * incompatible change: the --noprogrammer option, deprecated since
584 version 0.7.5, has been removed. Please use the equivalent
585 --disable-debugger option instead.
586 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
588 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
589 inoperational for a while now) has been completely removed.
590 * null lexical environments are now printed as #<NULL-LEXENV>,
591 significantly reducing the amount of clutter in typical backtraces.
592 * documentation on weak pointers, finalization, and after GC hooks
593 has been added to the manual.
594 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
595 as other specialized array types, is much faster. SUBSEQ and
596 COPY-SEQ on such arrays have also been sped up.
597 * optimization: EQL is now more efficient when at least other argument
598 is known to be of type (OR FIXNUM (NOT NUMBER)).
599 * fixed bug: the runtime is now less vulnerable to changes in the
600 size of the SBCL object on OS X, and virtual memory is reserved for
601 all spaces that need to be at a fixed address.
602 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
603 * fixed bug: finalizers and after GC hooks that cause consing are now
605 * fixed bug: compiler error messages and summaries are now printed to
606 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
607 * fixed inference of the upper bound of an iteration variable.
608 (reported by Rajat Datta).
609 * fixed bug 211e: calling local functions with duplicated constant
610 keyword argument no longer causes a bogus style warning about an
612 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
613 ftype declaration to be disregarded. (reported by Dave Roberts)
614 * fixed bug 373: caused by erronous compilation of references to alien
615 variables in the runtime on ppc/darwin.
616 * fixed bug 376: CONJUGATE type deriver.
617 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
618 (reported by Baughn on #lisp)
619 * a cross-compiler bug on non-x86 platforms has been identified and
620 fixed. (thanks to Bruno Haible)
621 * improvements to the MIPS runtime code for increased stability.
622 (thanks to Thiemo Seufer)
623 * increased the maximimum compact environment size to allow
624 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
625 * partial workaround for bug 135: don't name the function we're
626 calling for hairy cases of EVAL, so as not to accumulate one environment
627 entry per call to EVAL. (reported by Kevin Reid)
628 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
629 * optimization: type testing for non-vector arrays should be faster.
630 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
632 * bug fix: setting 31st element of a bit vector to zero did not work
634 * bug fix: redefining a class definition which failed due to a
635 previous accessor / function clash now works (but see BUGS entry
636 #380 for more problems in this area). (thanks to Zach Beane)
637 * the long form of DEFINE-METHOD-COMBINATION disables method group
638 checking when given a single method group with pattern *. (thanks
640 * on x86 compiler supports stack allocation of results of simple
641 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
642 * support for building and running on Mac OS X 10.4 "Tiger" has been added
643 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
644 * fixed some bugs related to Unicode integration:
645 ** the restarts for recovering from input and output encoding
646 errors only appear when there is in fact such an error to
648 * fixed some bugs revealed by Paul Dietz' test suite:
649 ** MISC.549 and similar: late transformation of unsafe type
650 assertions into derived types caused unexpected code
652 ** SCALE-FLOAT type deriver is less wrong.
653 ** type derivers for EXP, LOG and similar functions compute result
654 types for complex arguments better.
655 ** (MISC.563) CONJUGATE type deriver works for very restricted
657 ** out-of-line type testers for character strings are available.
658 ** EQUAL compiler transform understands specialness of objects
660 ** accessing double-floats stored on the stack now works on x86-64.
661 ** debugger internals could sometimes create invalid lispobjs,
662 resulting in GC crashes.
663 ** MISC.548: type check weakening can convert required type into
665 ** initialization forms for bindings are not in scope of free special
668 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
669 * incompatible change: thread support for non-NPTL systems has
670 been removed - locking is buggy and unreliable. A threaded
671 SBCL build will now warn at startup and refuse to create
672 new threads, unless futex support is detected in the kernel
673 * incompatible change: the top level REPL now has only an ABORT
674 restart associated with it, not TOPLEVEL and ABORT as it used to.
675 TOP and TOPLEVEL are now available as debugger commands for
676 returning to the top level.
677 * incompatible change: forms evaluated in the REPL now use the
678 global optimization policy.
679 * incompatible change: user- and system-initialization files are
680 no longer processed with LOAD, but by READ and EVAL; hence the
681 global optimization policy, startup package, readtable, etc,
683 * The .fasl file format number has been incremented because of
684 various incompatible changes.
685 * internal entry point details and argument counts no longer appear
686 in backtraces unless explicitly requested by setting
687 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
688 * built-in and standard functions no longer have names like "top
689 level local call to FOO".
690 * fixed bug 32: functions defined in non-null lexical environments
691 now have more legible printed representation
692 * fixed bug 33: functions defined in non-null lexical environemnts
693 are now more amenable to inspection by INSPECT.
694 * workaround for bug 354: XEPs no longer appear in backtraces unless
695 explicitly requested.
696 * fixed bug: receiving the signal which results from attempting to
697 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
698 SIGBUS on other *BSDs) on architectures where the C stack is also
699 the Lisp stack (x86 and x86-64) from foreign code no longer leads
700 to debugger confusion or wild instability. (reported by Cheuksan
702 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
703 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
704 (reported by Lutz Euler)
705 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
706 compiled code. (reported by Paul Dietz)
707 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
708 specialized parameter to a method to something that is not TYPEP
709 the specializer is now possible.
710 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
711 face of package deletion.
712 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
713 pathname. (reported by David Tolpin on c.l.l)
714 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
715 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
716 STRING-STREAM and FILE-STREAM, respectively.
717 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
718 than previously; in addition, its test suite is now run on build.
719 (thanks to Luke Gorrie)
720 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
721 * optimization: added a immediate representation for single-floats
723 * fixed some bugs revealed by Paul Dietz' test suite:
724 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
725 ** attempting to create a package with a colliding nickname causes
726 correctable errors to be signalled.
727 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
728 ** several x86-64 backend bugs related to sign-extension of immediate
731 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
732 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
733 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
735 * fixed loading of multiply forward-referenced layouts.
736 (thanks to Cheuksan Wang)
737 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
738 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
739 (see COMPILE-FILE documentation for details of :PRINT option)
740 * fixed bugs 19 and 317: fixed-format floating point printing is
741 more accurate. This also fixes a bug reported by Adam Warner
742 related to the ~@F format directive.
743 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
745 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
746 dispatch table if the from-char is a dispatch macro character.
747 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
748 by the wordsize no longer ignore the last word. (reported by Lutz
750 * fixed bug in type checking of dynamic-extent variables. (reported
752 * optimization: sequence traversal functions use their freedom to
753 coerce function designators to functions.
754 * optimization: code with many calls to CLOS methods specialized on
755 CLOS classes has had redundant type checks removed.
756 * optimization: added declarations to speed up operations that access
757 the internal character database (for example STRING-UPCASE)
758 * optimization: comparison operations between floats and sufficiently small
759 fixnums no longer create extra rationals
760 * fixed some bugs related to Unicode integration:
761 ** portions of multibyte characters at the end of buffers for
762 character-based file input are correctly transferred to the
763 start of the buffer at the next read.
764 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
765 passing it through to OPEN.
766 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
767 argument given to internal calls to OPEN.
768 ** the built-in comment readers (introduced by character sequences
769 ";" and "#|") are more forgiving to encoding errors; they will
770 STYLE-WARN and then attempt to resync the stream at a character
771 boundary. (thanks to Teemu Kalvas)
772 * fixed some bugs in the x86-64 port:
773 ** Negative short int return values from c-calls are sign-extended
775 ** The stack is aligned to 16-bytes for c-calls, as required by
776 the ABI. (thanks to Cheuksan Wang)
777 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
778 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
779 for enums. (thanks to Vincent Arkesteijn)
780 ** Multiple small optimizations and bugfixes for floating point
782 * fixed some bugs revealed by Paul Dietz' test suite:
783 ** Space, Tab, Linefeed, Return and Page have the invalid
784 secondary constituent character trait.
785 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
787 ** WITH-INPUT-FROM-STRING should only update the index place on
789 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
791 ** Bit-array operations (BIT-AND and similar) worked incorrectly
792 with one-dimensional arrays with fill pointers.
793 ** TYPE-OF failed on a complex with an integer realpart and a
795 ** compiler failure during type inference for the code of form
796 (IF (EQL X (THE ...)) ...) (MISC.535).
798 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
799 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
800 platform. The port supports 61-bit fixnums, large memory spaces
801 and reloading shared object files.
802 * enhancement: saving cores with foreign code loaded is now
803 supported on ppc/Darwin in addition to the previously supported
805 * enhancement: the statistical profiler now walks deeper into the
806 call stack for more meaningful call-graphs and accrued time
807 reports (x86/x86-64 only). It also now reports time spent in
809 * enhancement: it is now possible to trace most individual methods
810 of a generic function in addition to tracing the generic function
812 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
814 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
815 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
816 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
817 produces an error. (thanks to Vincent Arkesteijn)
818 * bug fix: NAMESTRING on pathnames with :WILD components in their
819 directories works correctly. (thanks to Artem V. Andreev)
820 * fixed bug 125: compiler preserves identity of closures. (reported
822 * bug fixed: functions with &REST arguments sometimes failed with
823 "Undefined function" when compiled with (DEBUG 3). (reported by
825 * bug fix: overflow during compiling of setting element of a bit
826 vector with constant index and value. (reported by Timmy Douglas)
827 * build fix: fixed the dependence on *LOAD-PATHNAME* and
828 *COMPILE-FILE-PATHNAME* being absolute pathnames.
829 * on x86 compiler partially supports stack allocation of dynamic-extent
831 * GO and RETURN-FROM do not check the extent of their exit points
832 when compiled with SAFETY 0.
833 * fixed some bugs related to Unicode integration:
834 ** encoding and decoding errors are now much more robustly
835 handled; it should now be possible to recover even from invalid
836 input or output to the terminal. (thanks to Teemu Kalvas)
837 ** provided a first cut at implementing STRING-TO-OCTETS and
838 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
839 ** altered the SB-MD5 contributed module slightly, changing the
840 interface just enough for it to be supportable for builds where
841 lisp characters are not eight bits.
842 * fixed some bugs revealed by Paul Dietz' test suite:
843 ** the FORMATTER-generated functions for ~V[ conditionals require
844 the correct number of arguments.
845 ** READ-FROM-STRING returns the mandated second value when applied
846 to displaced strings.
847 ** the #\Rubout and #\Backspace characters are treated as invalid
848 constituent characters by the tokenizer.
850 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
851 * new feature: reloading changed shared object files with
852 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
853 * new feature: references to foreign variables and functions
854 can now be compiled and loaded before the corresponding shared
855 object file is loaded, as long as the foreign definitions are
856 available at runtime.
857 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
858 supported. (thanks to Dan Debertin)
859 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
860 just keywords. (thanks to Vincent Arkesteijn)
861 * fixed bug #331: structure-class instances corresponding to
862 DEFSTRUCT forms are now created eagerly.
863 * fixed bug #345: backtraces from calls to undefined functions work
864 on x86 as well. Related bug #61 is now also partially fixed on x86
865 (backtraces from throws to unknown catch tags.)
866 * bug fix: lambda-list parsing is now stricter vrt. order and number
867 of lambda-list keywords.
868 * bug fix: as specified by AMOP, an error is signalled if a
869 class-option appears multiple times in a DEFCLASS form. (reported
871 * bug fix: RANDOM can be compiled when the compiler derives the type
872 of its numeric argument as a disjoint set of small integers.
873 (reported by Paul Dietz)
874 * bug fix: starting a core saved with shared objects loaded when
875 those objects are not available no longer causes threaded SBCL to
876 hang. (reported by Sean Ross)
877 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
879 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
880 argument, which must always be NIL. (reported by Kalle Niemitalo)
881 * bug fix: printing 1.0d+23 no longer results in an error.
882 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
883 * bug fix: structure slot setters preserve evaluation order. (thanks
885 * fixed some bugs related to Unicode integration:
886 ** RUN-PROGRAM can allow its child to take input from a Lisp
887 stream. (reported by Stefan Scholl)
888 ** ASDF-INSTALL successfully downloads .tar.gz files in all
889 locales. (reported by Ken Causey)
890 * fixed some bugs revealed by Paul Dietz' test suite:
891 ** INCF, DECF and REMF evaluate their place form as specified in
893 ** FORMATTER expands ~{ ~} iteration directives with V or #
894 parameters correctly.
895 ** FORMATTER deals with the ~@[ ~] conditional directive where the
896 consequent uses no arguments correctly.
897 ** the system has a partial understanding of the (COMPLEX RATIO)
899 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
902 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
903 * new feature: a build-time option (controlled by the :SB-UNICODE
904 keyword feature, enabled by default) for building the system with
905 support for the entire 21-bit character space defined by the
907 * new feature: the system now has rudimentary external-format
908 support; the primary user-visible change at this time is that
909 characters with the high bit set (such as lower-case-e-acute) will
910 print correctly to a terminal in a UTF-8 environment.
911 * minor incompatible change: BASE-CHAR no longer names a class;
912 however, CHARACTER continues to do so, as required by ANSI.
913 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
914 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
915 has been moved to the SB-EXT package (temporarily re-exported from
917 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
918 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
920 * fixed bug #351: better error-handlind and reporting for malformed
922 * fixed bug #350: bignum-printing is now more memory-efficient,
923 allowing printing of very large bignums, eg. (expt 2 10000000).
924 (reported by Bruno Haible)
925 * fixed bug #302: better primitive-type selection for intersection
927 * fixed bug #308: non-graphic characters now all have names, as
928 required. (reported by Bruno Haible)
929 * bug fix: redefining a class with different superclasses now correctly
930 removes it from the direct-subclasses of its previous superclasses.
931 (reported by David Morse)
932 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
933 new value now works. (reported by Bruno Haible)
934 * bug fix: correct canonicalization of multiple non-standard slot
935 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
936 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
937 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
938 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
939 now exists, an signals an error.
940 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
941 during macroexpansion for non-iteration variables as well. (reported
942 by Bruno Haible for CMUCL)
943 * bug fix: Cyclic structures and unprintable objects in compiler
944 messages no longer cause errors. (reported by Bruno Haible)
945 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
946 and READ-FROM-STRING all now return a primary value of NIL if
947 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
948 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
949 (reported by Bruno Haible for CMUCL)
950 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
951 do not designate a condition. (reported by Bruno Haible for
953 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
954 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
955 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
956 permitting resolution in favour of any of the conflicting symbols.
957 (reported by Bruno Haible for CMUCL)
958 * bug fix: EQUAL compiler optimizations is less aggressive on
959 strings which can potentially compare true despite having distinct
960 specialized array element types.
961 * bug fix: unit enumerations can be defined without dividing by
962 zero. (thanks to Vincent Arkesteijn)
963 * FORMAT compile-time argument count checking has been enhanced.
964 (report from Bruno Haible for CMUCL)
965 * a partial workaround for the bug 262: the compiler does not try to
966 inline-expand a local function doing RETURN-FROM from a deleted
967 BLOCK. (thanks to Peter Denno for the bug report and to David
968 Wragg for the simple test case)
969 * fixed some bugs revealed by Paul Dietz' test suite:
970 ** RENAME-PACKAGE allows all package designators as new package
972 ** constraint propagation and lambda variable substitution are
973 more cautious in dealing with partially deleted code.
974 ** compiler sometimes lost reoptimization passes.
975 ** CERROR, given a condition as condition designator, passes its
976 remaining arguments to the continue format control without
978 ** Case-altering FORMAT directives work correctly on non-ASCII
980 ** The REAL type specifier handles bounds outside the floating
981 point ranges without signalling FLOATING-POINT-OVERFLOW.
982 ** Functions with IR1-transformations can create intercomponent
983 references to global functions.
984 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
986 ** FORMAT ~R treats a nil value for its first parameter correctly.
988 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
989 * enhancement: saving cores with foreign code loaded is now
990 supported on x86/NetBSD and sparc/Linux in addition to the previously
992 * bug fix: on some platforms repeated installations caused multiple
993 copies of HTML documentation to be installed -- should not happen
994 any more. (reported by Stefan Scholl)
995 * bug fix: parsing self-recursive alien record types multiple times
996 no longer causes infinite recursion. (reported by Thomas F. Burdick,
997 original patch by Helmut Eller for CMUCL)
998 * bug fix: stack-exhaustion detection works now on NetBSD as well.
999 (thanks to Richard Kreuter)
1000 * bug fix: defining classes whose accessors are methods on existing
1001 generic functions in other (locked) packages no longer signals
1002 bogus package lock violations. (reported by François-René Rideau)
1003 * bug fix: special variables as DEFMETHOD parameters no longer have
1004 associated bogus type declarations. (reported by David Wragg and
1006 * bug fix: read-write consistency on streams of element-type
1007 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1008 * bug fix: redefiniton of the only method of a generic function with
1009 no DEFGENERIC no longer emits a full WARNING. In addition,
1010 redefinition of generic functions with no DEFGENERIC to an
1011 incompatible lambda list now signals an error. (thanks to Zach
1013 * bug fix: DEFGENERIC now works even when there's a function of the
1014 same name in an enclosing lexical environment. (thanks to Zach
1016 * fixed compiler failure, caused by instrumenting code during
1017 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1018 * optimization: added loop analysis and improved register allocation
1019 * fixed some bugs revealed by Paul Dietz' test suite:
1020 ** POSITION on displaced vectors with non-zero displacement
1021 returns the right answer.
1022 ** (SIMPLE-STRING) is a valid type specifier for sequence
1024 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1026 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1027 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1029 ** The justification version of the FORMAT ~< directive treats
1030 non-zero minpad parameter correctly.
1032 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1033 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1034 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1035 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1036 the supported interface.
1037 * new feature: Single-stepping of code compiled with DEBUG 2 or
1038 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1039 * new feature: saving cores with foreign code loaded is now
1040 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1041 Timothy Moore's work for CMUCL)
1042 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1043 parameters to * instead of NIL if no initform is supplied.
1044 (reported by Johan Bockgaard)
1045 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1046 correctly when FUNCALL appears as the car of the form. Note:
1047 despite this FUNCALL forms are not currently subject to
1048 compiler-macro expansion. (port of Raymond Toy's fix for the
1049 same from CMUCL, reported by Johan Bockgaard)
1050 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1051 (thanks for Teemu Kalvas)
1052 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1053 case instead of terminating the iteration (thanks for Julian
1054 Squires, Sean Champ and Raymond Toy)
1055 * bug fix: incorrect expansion of defgeneric that caused a style
1056 warning. (thanks for Zach Beane)
1057 * bug fix: slot accessor effective method computation works properly
1058 for classes with multiple non-standard applicable methods on
1059 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1060 * on x86 compiler supports stack allocation of results of LIST and
1061 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1062 CMUCL implementation by Gerd Moellmann)
1063 * fixed some bugs revealed by Paul Dietz' test suite:
1064 ** FORMAT strings with both the ~<~:;~> form of the justification
1065 directive and pretty-printing directives cause an error.
1067 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1068 * incompatible change: the internal functions
1069 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1070 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1071 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1072 instead of the old functions.
1073 * new feature: on platforms where "dladdr" is available foreign
1074 function names now appear in backtraces. (based on Helmut Eller's
1076 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1077 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1079 * documentation: ASDF manual is now distributed with SBCL.
1080 * bug fix: SBCL can now load its contributed modules with REQUIRE
1081 even if the system-provided entries have been removed from
1082 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1084 * bug fix: backtraces involving undefined functions or assembly
1085 routines are more informative. (thanks to Brian Downing)
1086 * bug fix: mutually referent alien structures now work correctly.
1087 (reported by Rick Taube)
1088 * bug fix: structures defined by WITH-ALIEN can be referred to
1089 within other definitions in the same WITH-ALIEN.
1090 * bug fix: division operators (MOD, TRUNCATE and the like) with
1091 constant zero divisors and integer dividends no longer generate
1093 * bug fix: provide default methods for INPUT-STREAM-P and
1094 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1095 * bug fix: improve the detection and resolution of MOP metacycles
1096 (where effective method computation depends on the generic function
1097 having its effective method computed). (reported by Bruno Haible)
1098 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1099 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1100 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1101 * bug fix: pathnames with relative directory components can be
1102 represented relative to default pathnames.
1103 * optimization: in taking the GCD of bignums, reduce the two bignums
1104 to approximately the same size (using Euclid's algorithm) before
1105 applying the more sophisticated binary GCD. (thanks to Juho
1107 * optimization: COUNT on bitvectors now operates word-at-a-time.
1108 * optimization: ASH with a positive, but not necessarily constant,
1109 (leftwards) shift, when in a modular context, is compiled to a
1111 * fixed some bugs revealed by Paul Dietz' test suite:
1112 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1113 if the corresponding argument is NIL.
1114 ** FORMAT directives accepting a minpad parameter treat negative
1116 ** literal commas as character parameters to FORMAT directives are
1118 ** literal spaces directly after ~<Newline> directives within a
1119 format-logical-block (~:< ~@:>) do not induce :FILL-style
1120 conditional newlines.
1121 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1122 as specified: it no longer includes conditional newlines).
1123 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1125 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1126 :PER-LINE-PREFIX argument does not evaluate to a string.
1128 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1129 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1130 the manual for details; for now, package locks can be disabled by
1131 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1132 if no major problems are found then it is likely that they will be
1133 compiled in unconditionally.
1134 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1135 unsupported operators on all platforms. To load a shared library
1136 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1137 object file, link it into a shared library outside of SBCL and
1138 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1140 * bug workaround: ROOM T and the GC were not working together
1141 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1142 reporting facility aren't preserved. That reporting has been
1143 disabled, so now until and unless someone figures out how to make it
1144 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1145 * minor incompatible change: as threatened around sbcl-0.8.0, the
1146 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1147 The symbols are also exported from SB-PCL for backwards
1148 compatibility, but more so than before SB-PCL should be treated as
1149 an implementation-internal package.
1150 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1152 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1153 It is known as of this release not to work on the Alpha, however.
1154 * fixed bug #167: errors signalled due to illegal syntax in method
1155 bodies are now more legible.
1156 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1157 designators and can hence be used with TYPEP.
1158 * fixed bug #333: CHECK-TYPE now ensures that the type error
1159 signalled, if any, has the right object to be accessed by
1160 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1161 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1162 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1164 * fixed bug #334: programmatic addition of slots using specialized
1165 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1166 and :ALLOCATION :CLASS slots.
1167 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1168 float exponents. (rereported by Peter Seibel)
1169 * fixed a bug: #\Space (and other whitespace characters) are no
1170 longer considered to be macro characters in standard syntax by
1171 GET-MACRO-CHARACTER.
1172 * fixed bug: initialization of condition class metaobjects no longer
1173 causes an instance of the condition to be created. (reported by
1175 * fixed bug: it is now possible to have slots such that
1176 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1177 description is neither :INSTANCE nor :CLASS.
1178 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1179 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1180 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1181 non-standard SLOT-DEFINITION-ALLOCATION values.
1182 * fixed bugs: various race conditions handling exiting threads.
1183 CL-PPCRE's thread torture test now passes.
1184 * fixed arguably-a-bug: GC time proportional to number of threads in
1185 system even when most of them are idle
1186 * optimization: improved performance of BIT and SBIT on bit-vectors.
1187 * .fasl file incompatibility: The fasl file version number has
1188 been incremented because of changes associated with package locks.
1190 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1191 * minor incompatible change: the system no longer provides
1192 optimization hints (conditions which are TYPEP
1193 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1194 mode; these hints are still emitted when higher SPEED optimization
1196 * new contrib module: a sampling profiler (profiling by statistical
1197 sampling, rather than by instrumenting functions) is available as
1198 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1199 * the behaviour of the standard function ED is now customizeable by
1200 third parties through a hook variable: see ED's documentation
1201 string for information on the protocol.
1202 * the compiler no longer emits efficiency notes for (FUNCALL X)
1203 when the type of X is uncertain under default optimization
1205 * fixed bug 276: mutating a binding of a specialized parameter to a
1206 method to something that is not TYPEP the specializer is now
1208 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1209 exhibit the required behaviour on the x86 platform. (thanks to
1210 Peter van Eynde, Eric Marsden and Bruno Haible)
1211 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1212 even for difficult arguments. (reported by Peter Graves)
1213 * fixed bug 141a: the backquote printer now descends quoted
1215 * fixed another bug in backquote printing: no more destructive
1216 modification of the form's list structure. (reported by Brian
1218 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1219 move between its address being taken and the call to
1220 interrupt_thread, fixing a crashing race condition.
1221 * the SB-POSIX contrib implementation has been adjusted so that it
1222 no longer exhibits ridiculously poor performance when constructing
1223 instances corresponding to C structs.
1225 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1226 * minor incompatible change: the sb-grovel contrib now treats C
1227 structures as alien (in the sense of SB-ALIEN) objects rather than
1228 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1229 has implications for memory management of client code
1230 (sb-grovel-returned objects must now be manually managed) and for
1231 type safety (alien objects now have full types).
1232 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1233 used to control emission of compiler diagnostics, rather than the
1234 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1235 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1236 quality should be considered deprecated.
1237 * install.sh now installs the user manual as well
1238 * (not quite a new documentable feature, but worth considering in
1239 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1240 beginnings of a semantically meaningful condition hierarchy is
1241 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1243 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1244 designator as the defaults argument.
1245 * fixed bug: Displaced arrays whose displaced-to array has become
1246 too small now cause ARRAY-DIMENSION to signal an error, providing
1247 for safer bounds-checking. (reported by Bruno Haible)
1248 * fixed bug: DEFCLASS slot definitions with identical :READER and
1249 :WRITER names now signal a reasonable error. (reported by Thomas
1251 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1253 * fixed bug: Invalid slot specification errors now print correctly.
1254 (thanks to Zach Beane)
1255 * fixed bug 320: Shared to local slot value transfers in class
1256 redefinitions now happen corrently from superclasses as
1257 well. (reported by Bruno Haible)
1258 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1260 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1261 improper lists correctly. (reported by Bruno Haible)
1262 * fixed bug 313: source-transform for <fun-name> was erroneously
1263 applied to a call of a value of a variable with name <fun-name>.
1264 (reported by Antonio Menezes Leitao)
1265 * fixed bug 307: The obsolete instance protocol ensures that
1266 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1267 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1268 remove dead unknown-values globs from the middle of the stack.
1269 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1270 easier to resume long computations after using *BREAK-ON-SIGNALS*
1271 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1272 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1273 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1274 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1275 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1276 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1278 * fixed a (fairly theoretical) bug in string printing: if
1279 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1280 does not have array-element-type equal to the most general string
1282 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1283 function name. (thanks to Nikodemus Siivola)
1284 * fixed bug: compile-time format string checker failed on
1285 non-closed ~{. (reported by Thomas F Burdick)
1286 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1287 don't warn on legal loop code involving more than one aggregate
1288 boolean. (thanks to Nikodemus Siivola)
1289 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1290 in keyword position to MAKE-PACKAGE was not being reported as
1292 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1293 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1295 * fixed bugs 280 and 312: the checking for multiple definitions in a
1296 file is less likely to become confused by uses of inline
1298 * fixed bug: the #S reader macro performs the keyword coercion
1299 specified for slot names. (reported by Kalle Niemitalo)
1300 * fixed bug: lambda lists may contain symbols whose names start with
1301 & but are not lambda-list-keywords; their occurrence triggers a
1303 * fixed bug 321: define-method-combination argument lambda lists do
1304 not cause invalid code to be generated when &optional or &aux
1305 variables have default values. (reported by Bruno Haible)
1306 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1307 have CLOS classes; this appears also to have fixed reported
1308 instability in stack exhaustion detection.
1309 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1310 MUFFLE-WARNING when no associated restart is present is now
1312 * optimization: rearranged the expansion of various defining macros
1313 so that each expands into only one top-level form in a
1314 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1316 * optimization: used a previously little-used slot in symbols to
1317 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1319 * fixed some bugs revealed by Paul Dietz' test suite:
1320 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1322 ** Fixed an optimization bug related to inheritance of initforms
1323 from local to shared slots.
1324 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1325 ** CAST optimizer forgot to flush argument derived type.
1326 ** print/read consistency on floats is now orders of magnitude
1327 more likely. (thanks also to Bruno Haible for a similar report
1329 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1330 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1331 for objects of type REAL. Make it so.
1332 ** SXHASH is specified (infelicitously) to respect similarity,
1333 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1334 it so. (thanks to Markus Ziegler)
1335 ** on the Alpha, the compiler succeeds in compiling functions
1336 returning a known number of arguments greater than 63.
1337 ** fixed handling of invalid NIL arguments in keyword position
1339 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1341 ** provided a readably-printable representation for RANDOM-STATE
1343 ** ensured that pathnames, where they have a namestring, always
1344 print using #P"..." syntax.
1346 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1347 * Support for the forthcoming 2.0 version of the NetBSD kernel
1348 running on x86 hardware has been added. (thanks to Perry
1349 E. Metzger most immediately, and others for their past work)
1350 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1351 current release version; SBCL's OpenBSD support had been broken
1352 since about the time of OpenBSD's switch to ELF binary format.)
1353 * [placeholder for DX summary]
1354 ** user code with &REST lists declared dynamic-extent, under high
1355 speed or space and low safety and debug optimization policy.
1356 * The manual has been converted to Texinfo format and the debugger
1357 chapter from the cmucl manual has been added.
1358 * A facility has been added to extract documentation strings from
1359 sbcl and store them as Texinfo-formatted snippets for inclusion in
1360 the manual (via Texinfo's @include directive)
1361 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1362 the test case to Dave Roberts)
1363 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1364 pointers. (reported by Sean Ross)
1365 * bug fix: PROFILE output is printed nicely even for large numerical
1366 values. (thanks to Zach Beane)
1367 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1368 greater than 32 handle EOF correctly.
1369 * bug fix: on X86 an immediate argument of the IMUL instruction is
1370 correctly printed in disassembly. (thanks to Lutz Euler)
1371 * bug fix: class slots in redefined classes preserve their old
1372 values. (thanks to Bruno Haible and Nikodemus Siivola)
1373 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1374 succeeds. (reported by Marco Baringer)
1375 * fixed some bugs revealed by Paul Dietz' test suite:
1376 ** READ-SEQUENCE now works on ECHO-STREAMs.
1377 ** RATIONALIZE works more according to its specification. (thanks
1380 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1381 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1382 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1383 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1384 should matter to you only if you rebind the printer control
1385 variables and then find you want different bindings in the
1386 debugger than in the ordinary execution of your program.)
1387 * The runtime build system has been tweaked to support building
1388 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1389 assembler and linker. (thanks to Nikodemus Siivola)
1390 * Unbound, undefined, undeclared variables now trigger full
1391 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1392 more often programmer error than deliberate exploitation of undefined
1394 * optimization: the hash algorithm for strings has changed to one
1395 that is less vulnerable to spurious collisions. (thanks to Juho
1397 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1398 needless bounds checking. (thanks to Juho Snellman)
1399 * optimization: implemented multiplication as a modular
1400 (UNSIGNED-BYTE 32) operation on the PPC backend.
1401 * fixed some bugs revealed by Paul Dietz' test suite:
1402 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1404 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1405 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1406 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1408 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1410 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1412 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1413 their output stream on EOF from read.
1414 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1415 have been read to end-of-file.
1416 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1418 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1419 description of determination of which consecutive characters
1421 ** Printing the "Space" character with escaping on now yields "#\\ ",
1422 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1423 ** Reading floating-point numbers with *READ-BASE* set to a number
1424 less than 10 works correctly.
1425 ** Reading floating-point numbers with *READ-BASE* set to a number
1426 more than 10 works correctly.
1427 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1428 the readtable currently in effect.
1430 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1431 * minor incompatible change: parsing of namestrings on a physical
1432 (Unix) host has changed; numbers after the final #\. in a
1433 namestring are no longer interpreted as a version field. This is
1434 intented to be largely invisible to the user, except that the
1435 meaning of the namestring "*.*.*" has changed: it now refers to a
1436 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1437 should usually be replaced by
1438 (make-pathname :name :wild :type :wild :version :wild)
1439 with the added benefit that this is more likely to be portable.
1440 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1441 option to OPEN now signals an error if the file being opened
1442 exists; this may have an impact on existing code.
1443 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1444 signals received, on the PowerPC platforms (both Linux and
1445 Darwin). (thanks to Pierre Mai for pointing out the location of
1447 * several fixes on OS X: The system now builds and runs cleanly on
1448 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1449 (thanks to Brian Mastenbrook)
1450 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1451 second-resolution: integer multiples of 1/3600 between -24 and 24.
1452 (thanks to Vincent Arkesteijn)
1453 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1454 type when called with 1 argument; PEEK-CHAR checked type of
1455 PEEK-TYPE only after having read first character from a
1456 stream. (reported by Peter Graves)
1457 * bug fix: the garbage collector now has much better locality
1458 behaviour, and in particular no longer treats all memory as being
1459 exhausted when many small objects point to each other in a deeply
1461 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1462 recognized as being TYPEP their class.
1463 * bug fix: the PUSHNEW documentation string has been corrected.
1464 (thanks to Vincent Arkesteijn)
1465 * bug fix: defaulting of the value for the last of an atypically
1466 large number of multiple values being bound was not being
1467 performed correctly on the Alpha or PPC platforms
1468 * optimization: implemented multiplication as a modular
1469 (UNSIGNED-BYTE 32) operation on the x86 backend.
1470 * optimization: SEARCH on simple-base-strings can now be open-coded.
1471 (see also contrib/compiler-extras.lisp for inspiration for
1472 teaching the compiler about the Boyer-Moore algorithm).
1473 * value, returned by MAX (and MIN) called with several EQUALP, but
1474 not EQL, arguments now does not depend on compiler settings.
1475 * fixed some bugs revealed by Paul Dietz' test suite:
1476 ** in stack analysis liveness information is propagated from
1477 non-local entry points.
1478 ** pathwise CAST removing failed when the CAST node did not start
1480 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1481 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1483 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1484 host is already defined.
1485 ** RENAME-FILE works on streams instead of signalling an internal
1487 ** PEEK-CHAR uses the current readtable when determining whether
1488 or not a character is whitespace.
1489 ** MERGE-PATHNAMES handles the case when the pathname does not
1490 specify a name while the default-pathname specifies a version.
1491 ** Pathnames now stand a better chance of respecting print/read
1493 ** Attempting to use standardized file system operators with a
1494 pathname with invalid :DIRECTORY components signals a
1496 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1498 ** DIRECTORY on logical pathnames is more correct.
1499 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1500 signal a TYPE-ERROR if their argument is not a stream.
1501 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1502 designator argument does not designate a stream.
1503 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1504 examining the synonym.
1505 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1507 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1508 element-type larger than ([UN]SIGNED-BYTE 32).
1510 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1511 * When built with the :SB-FUTEX feature, threaded builds now take
1512 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1513 for faster/more reliable mutex and condition variable support.
1514 * Incompatible change (but one you probably shouldn't have been using
1515 anyway): the interface and code for arbitrating between multiple
1516 threads in the same user session has been redesigned.
1517 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1518 error when called without an explicit environment argument.
1519 (thanks to Nikodemus Siivola)
1520 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1521 confused about their position. (thanks to Adam Warner and Gerd
1523 * bug fix: Pretty printing backquoted forms with unquotations in the
1524 argument list position of various code constructs such as LAMBDA
1525 now works correctly. (reported by Paul Dietz)
1526 * bug fix: Pretty printing unquotations no longer loses all
1527 stream position information.
1528 * optimization: performance of string output streams is now less
1529 poor for multiple small sequence writes.
1530 * optimization: performance of CSUBTYPEP in the presence of complex
1531 expressions involving CONS and NOT many times has been improved.
1532 (reported by Paul Dietz)
1533 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1535 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1536 streams for the REPL, for the debugger, and so on. (thanks to
1538 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1539 * fixed some bugs revealed by Paul Dietz' test suite:
1540 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1541 is ignored if it is too small, rather than propagating through
1542 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1543 ** extremely complex negations of CONS types were not being
1544 sufficiently canonicalized, leading to inconsistencies in
1546 ** VALUES tranformer lost derived type.
1548 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1549 * fixed a bootstrapping bug: the build process no longer assumes
1550 that the various BOOLE-related constants have the same value in
1551 host and target lisps. (noted by Paul Dietz' test suite on an
1552 SBCL binary built from CLISP)
1553 * The system can now be dynamically linked on the MIPS platform,
1554 which enables dynamic loading of foreign code from Lisp. (thanks
1555 to Ralf Baechle for discussions on the MIPS ABI)
1556 * The system now records debugging information for its own source
1557 files in a filesystem-position-independent manner, relative to
1558 the "SYS" logical host.
1559 * fixed a compiler bug: MV-LET convertion did not check references
1560 to the "max args" entry point. (reported by Brian Downing)
1561 * tweaked disassembly notes to be less confident about proclaiming
1562 some instruction as an LRA. (thanks to Brian Downing)
1563 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1564 now each have their own history, command character, and other
1565 characteristics. (thanks to David Lichteblau)
1566 * fixed some bugs revealed by Paul Dietz' test suite:
1567 ** compiler failure in compiling LOGAND expressions including a
1569 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1570 shift greater than 32.
1571 ** FLUSH-DEST did not mark blocks for type check regeneration.
1572 ** HANDLER-CASE failed to accept declarations in handler clauses
1573 in some circumstances.
1575 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1576 * New code in contrib/sb-introspect (still probably not entirely
1577 stable yet) provides some support for smart Lisp development
1578 environments like SLIME.
1579 * The conditions signalled for errors occurring when loading .fasl
1580 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1581 in a way which should help ASDF recover gracefully.
1582 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1583 stuff has been cleaned up. If you code contrib/ stuff, this might
1584 affect you, and you can look at contrib/README, contrib/STANDARDS,
1585 and/or the 0.8.4.27 diff to check.
1586 * In full calls the compiler now does not generate checks for declared
1587 argument types for all arguments.
1588 * various threading fixes
1589 ** and some experimental patches which didn't make it into
1590 the main tree for this release, but which are shipped in
1591 contrib/experimental-thread.patch as a possible fix for some
1592 failures (deadlock, spinning...) in GC-intensive multithreaded
1594 * fixed PPC build problem (source code incompatibility of different
1595 library versions): added offsetof() hackery which attempts to divine
1596 where glibc maintainers put uc_mcontext today
1597 * fixed bug 282: compiler does not trust type assertions while passing
1598 arguments to a full call.
1599 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1600 &OPTIONAL VALUES type parameter.
1601 * fix bug 214: algorithm for noting rejected templates is now more
1602 similar to that of template seletion. (also reported by rydis on
1604 * fixed bug 141b: printing backquoted information readably and prettily
1605 inserts a space where necessary.
1606 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1607 (CHARACTER 10) are now reported as errors, rather than propagated
1608 as unknown types. (reported by piso on #lisp)
1609 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1610 with values NIL and :ERROR. (thanks to Milan Zamazal)
1611 * fixed bug 191c: CLOS now does proper keyword argument checking as
1612 described in CLHS 7.6.5 and 7.6.5.1.
1613 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1614 longer raise an error; further, using a list as a for-as-arithmetic
1615 counter now raises a meaningful error.
1616 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1617 now understood by sequence creation functions such as MAKE-SEQUENCE
1619 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1620 a STRING-INPUT-STREAM.
1621 * compiler enhancement: SIGNUM is now better able to derive the type
1623 * type declarations inside WITH-SLOTS are checked. (reported by
1625 * fixed some bugs revealed by Paul Dietz' test suite:
1626 ** incorrect optimization of TRUNCATE for a positive first
1627 argument and negative second.
1628 ** compiler failure in let-convertion during flushing dead code.
1629 ** compiler failure while deriving type of TRUNCATE on an
1630 interval, containing 0.
1631 ** ASH of a negative bignum by a negative bignum count now returns
1633 ** intersection of CONS types now canonicalizes properly, fixing
1634 inconsistencies in SUBTYPEP.
1636 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1637 * incompatible change: The --disable-debugger command line
1638 option now clobbers the debugger at a more fundamental
1639 level, by redefining #'INVOKE-DEBUGGER instead of by
1640 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1641 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1642 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1643 enter the debugger REPL and then suffer recursive errors
1644 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1645 while under the new system, BREAK in --disable-debugger mode
1646 terminates the system just as an unhandled error would.
1647 * fixed compiler performance when processing loops with a step >1;
1648 * bug fix: DOCUMENTATION now retrieves generic function
1649 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1650 support has been systematized, and now supports the methods
1651 specified by ANSI, along with a default method and a method for
1652 slot documentation. (reported by Nathan Froyd)
1653 * bug fix: effective methods associated with a generic function are
1654 no longer cached over a change of that generic function's method
1655 combination. (reported by Andreas Fuchs)
1656 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1657 names a non-existent directory. (thanks to Andreas Fuchs)
1658 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1659 the PPC platform, lead to stack corruption; this has been fixed.
1660 (reported by Rainer Joswig)
1661 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1662 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1663 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1664 arithmetic in general, is now much more reliable on the PPC
1666 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1667 platform now returns the right answer.
1668 * optimization: restored some effective method precomputation in
1669 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1670 precomputation is now tunable.
1671 * optimization: compiler-internal data structure use has been
1672 reviewed, and changes have been made that should improve the
1673 performance of the compiler by about 20%.
1674 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1675 simple-base-strings and simple-bit-vectors is improved.
1676 * optimization: the optimization of 32-bit logical and arithmetic
1677 functions introduced in version 0.8.3 on the x86 has been
1678 implemented on the mips, ppc and sparc platforms; an
1679 implementation of the same facility, but for 64-bit arithmetic,
1680 has been added for the alpha.
1681 * microoptimization: the compiler is better able to make use of the
1682 x86 LEA instruction for multiplication by constants.
1683 * bug fix: in some situations compiler did not report usage of
1684 generic arithmetic in (SPEED 3) policy.
1685 * bug 145b fix: compiler used wrong type specifier while converting
1686 MEMBER-types to numeric.
1687 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1689 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1690 index argument now works properly on the Alpha platform.
1691 * bug fix: floating point exception treatment on the Alpha platform
1693 * bug fix: FILE-POSITION works much better on string input and
1694 output streams. (thanks to Nikodemus Siivola)
1695 * bug fix: many threading/garbage collection symptoms sorted.
1696 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1697 might be pseudo-atomic.
1698 * internal change: Stopping for GC is now done with signals not
1699 ptrace. GC is now done in whichever thread wanted it, instead of
1701 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1703 * bug fix: result form in DO is not contained in the implicit
1705 * incompatible change: ICR structure is changed; the value part of
1706 CONTINUATION is now called LVAR; corresponding functions are
1707 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1708 * added type deriver for ISQRT (thanks to Robert E. Brown).
1709 * bug fix: better support for loading from the command line when an
1710 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1712 * fixed some bugs revealed by Paul Dietz' test suite:
1713 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1714 ** ROUND and FROUND now give the right answer when given very
1715 small float arguments.
1716 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1718 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1719 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1720 ** LCM should return a non-negative integer.
1721 ** PARSE-INTEGER returned the index of a terminator instead of the
1722 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1723 ** PARSE-INTEGER returned an incorrect index being applied to a
1725 ** LCM with two arguments of 0 returns 0 rather than signalling
1727 ** unsigned addition of a 32-bit constant with the high bit set no
1728 longer causes an internal compiler error on the x86.
1729 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1730 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1731 with negative last argument.
1732 ** byte specifiers with zero size and position no longer cause
1733 an error during type derivation.
1734 ** bignum multiplication on the Alpha platform now returns the
1736 * porting: The system now builds on SuSE AMD64, although it still
1737 generates a 32-bit binary.
1738 * .fasl file incompatibility: The fasl file version number has
1739 been incremented (because of the changes to internal compiler
1740 data structures referred to above).
1742 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1743 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1744 more accurately, on the Darwin kernel running on PowerPC hardware.
1745 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1746 * Compiler code deletion notes now signal a condition of type
1747 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1748 an associated MUFFLE-WARNING restart.
1749 * The compiler now performs limited argument count validation of
1750 constant format strings in FORMAT, and where appropriate in ERROR,
1751 CERROR and WARN. (thanks to Gerd Moellmann)
1752 * New ASDF-INSTALL contrib can be used for automatic download and
1753 installation of third-party Lisp code from CCLAN or other sites
1755 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1756 which forces another thread to execute a function supplied by the
1758 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1759 now accept and act upon their :ELEMENT-TYPE keyword argument.
1760 (reported by Martin Atzmueller, Edi Weitz)
1761 * bug fix: FILE-POSITION now accepts position designators up to
1762 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1763 is the greater. (thanks to Patrik Nordebo)
1764 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1765 Kalle Olavi Niemitalo)
1766 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1767 printable. (reported by Eric Marsden)
1768 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1769 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1770 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1771 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1772 (thanks to Stig Sandoe)
1773 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1774 UNREAD-CHAR now correctly includes the unread character in the
1775 target sequence. (thanks to Gerd Moellmann)
1776 * bug fix in threaded builds: the system can now be suspended and
1777 resumed by shell job control with minimal disruption.
1778 * bug fixes in times and timezones >2038AD
1779 * better handling of "where is GNU make?" problem in build scripts
1780 (thanks to Nikodemus Siivola)
1781 * new optimization: inside a named function any reference to a
1782 function with the same name is considered to be a self-reference;
1783 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1784 optimization quality.
1785 * new optimization on x86: logical functions and + now have
1786 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1787 used when the result is truncated to 32 bits.
1788 * VALUES declaration is partially enabled.
1789 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1790 * fixed some bugs revealed by Paul Dietz' test suite:
1791 ** The system now obeys the constraint imposed by
1792 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1793 types form a lattice under type intersection.
1794 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1795 ** ASSOC now ignores NIL elements in an alist.
1796 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1797 and (1+ MOST-POSITIVE-FIXNUM) answers.
1798 ** The addition of a method with invalid qualifiers to a generic
1799 function does not cause an error to be signalled immediately;
1800 a warning is signalled, and the error is generated only on
1801 calling the generic function.
1802 * changed .fasl file version number, in part to add type codes for
1803 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1804 obscure ANSI requirements
1806 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1807 * fixed bug 148: failure to inline-expand a local function left
1808 garbage, confusing the compiler.
1809 * fixed bugs 3cd: structure slot readers perform type check if the
1810 slot can have an invalid value (i.e. it is either not initialized
1811 or can be written with a less specific slot writer).
1812 * bug fix: the compiler now traps array references to elements off
1813 the end of an array; previously, the bounds checking in some
1814 circumstances could go off-by-one.
1815 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1817 * type declarations for array element types now obey the description
1818 on the CLHS page "Declaration TYPE", as per discussions on
1819 sbcl-help around 2003-05-08. This means that a declaration
1820 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1821 declaration, all references to BAR will be asserted or assumed
1822 (with THE, so dependent on compiler policy) to involve objects of
1823 type FOO. Note that no such declaration is implied in
1824 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1825 * declared types of functions from the "Conditions"
1826 chapter. (reported by Paul Dietz)
1827 * bug fix: CERROR accepts a function as its first argument.
1828 * bug fix: NTH an NTHCDR accept a bignum as index
1829 arguments. (reported by Adam Warner)
1830 * optimization: character compare routines now optimize comparing
1831 against a constant character. (reported by Gilbert Baumann)
1832 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1833 argument now works properly on the MIPS platform.
1834 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1835 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1836 declared to have a functional type, does not cause an error
1838 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1839 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1840 * fixed bug: if last continuation of a deleted block has a
1841 destination, this destination should be deleted too. (reported by
1843 * fixed a bug in the bootstrap process: the host compiler's values
1844 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1845 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1846 test case from Patrik Nordebo)
1847 * improved the ability of the disassembler on the PPC platform to
1848 provide helpful disassembly notes.
1849 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1850 the class in more cases than previously.
1851 * bug fix: FILE-POSITION now understands :START and :END for
1852 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1853 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1854 8) data. (thanks to David Lichteblau)
1855 * bug fix: it is possible to add a method to a generic function
1856 without lambda list.
1857 * bug fix: reader failed to signal END-OF-FILE inside an
1858 object representation. (reported by Nikodemus Siivola)
1859 * fixed some bugs revealed by Paul Dietz' test suite:
1860 ** LAST and [N]BUTLAST should accept a bignum.
1861 ** condition slot accessors are methods.
1862 ** (VECTOR NIL) is a subtype of STRING.
1864 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1865 * minor incompatible change: some nonsensical specialized lambda
1866 lists (used in DEFMETHOD) which were previously ignored now signal
1868 * minor incompatible change: the system is now aware of the types of
1869 variables in the COMMON-LISP package, and will signal errors for
1870 most violations of these type constraints (where previously they
1871 were silently accepted).
1872 * minor incompatible change: COMPILE-FILE now uses the freedom
1873 afforded (ANSI 3.2.2.3) to use derived function types for
1874 functions defined in the same file. This also permits the system
1875 to warn on static type mismatches and function
1876 redefinition. (Currently it does not work with high DEBUG level.)
1877 * minor incompatible change: VALUES declaration is disabled.
1878 * When issuing notes, the compiler now signals a condition of type
1879 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1880 restart for use in user handlers. It is expected that the
1881 COMPILER-NOTE condition will eventually become a condition
1882 supertype to a hierarchy of note types, which will then be
1883 handleable in a similar fashion. However, at the moment, no such
1884 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1885 but it's an implementation detail, not a classification for the
1887 * Changes in type checking closed the following bugs:
1888 ** type checking of unused values (192b, 194d, 203);
1889 ** template selection based on unsafe type assertions (192c, 236);
1890 ** type checking in branches (194bc).
1891 * A short form of VALUES type specifier has ANSI meaning (it has
1892 increased the number of situations when SBCL cannot perform type
1894 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1895 keywords or constants is permissible.
1896 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1897 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1898 classes. (thanks to Antonio Martinez)
1899 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1900 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1901 argument to a pretty-print stream. Instead, it leaves any such
1902 operations to DESCRIBE-OBJECT methods.
1903 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1905 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1906 lambda lists. (thanks to David Lichteblau)
1907 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1908 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1909 (thanks to Gerd Moellmann)
1910 * fixed bug 246: increased compilation speed of long
1911 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1913 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1915 * DEFCONSTANT now throws a condition of type
1916 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1917 constant to a non-EQL value; CONTINUE and ABORT restarts
1918 respectively change and preserve the value.
1919 * fixed bug 63: The code walker, part of the implementation of CLOS,
1920 is now better at handling symbol macros.
1921 * bug fix: There is no longer an internal implementation type named
1922 CL:LENGTH. (reported by Raymond Toy)
1923 * bug fix: In macro-like defining macros/special operators the
1924 implicit block does not enclose the lambda list.
1925 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1926 enclosed in parentheses) are not suitable as type specifiers, and
1927 their use properly signals an error now.
1928 * bug fix: An argument count mismatch for a type specifier in code
1929 being compiled no longer causes an unhandled error at compile
1930 time, but signals a compile-time warning.
1931 * fixed simple vector readable printing
1932 * bug fix: DESCRIBE takes more care over whether the class
1933 precedence list slot of a class is bound before accessing it.
1934 (reported by Markus Krummenacker)
1935 * bug fix: FORMATTER can successfully compile pretty-printer format
1936 strings which use variants of the ~* directive inside.
1937 * bug fix: SEARCH now applies its TEST predicate to the elements of
1938 the arguments in the correct order. (thanks to Wolfhard Buss)
1939 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1940 in type inference. (thanks to Robert E. Brown)
1941 * bug fix: Reading in symbols with an explicit package name of ""
1942 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1943 * fixed some bugs revealed by Paul Dietz' test suite:
1944 ** NIL is now allowed as a structure slot name.
1945 ** Arbitrary numbers, not just REALs, are allowed in certain
1946 circumstances in LOOP for-as-arithmetic clauses.
1947 ** Multiple class redefinitions before slot access no longer
1948 causes a type error.
1949 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1950 association between the name and a class.
1951 ** Generic functions with non-standard method-combination and over
1952 five methods all of which return constants no longer return NIL
1953 after the first few invocations. (thanks to Gerd Moellmann)
1954 ** CALL-NEXT-METHOD with no arguments now passes the original
1955 values of the arguments, even in the presence of assignment.
1956 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1958 ** DISASSEMBLE works with closures and funcallable instances.
1959 ** ADD-METHOD now returns the generic function, not the new method.
1960 ** FIND-METHOD signals an error if the lengths of the specializers
1961 is incompatible with the generic function, even if the ERRORP
1963 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1964 which its argument is a member.
1965 ** DEFCLASS only redefines the class named by its class-name
1966 argument if that name is the proper name of the class;
1967 otherwise, it creates a new class.
1968 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1969 of the UNBOUND-SLOT condition to the name of the slot.
1970 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1971 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1972 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1973 SLOT-MAKUNBOUND in the specified fashion.
1975 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1976 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1977 cross-compilation host. As a consequence, we can now bootstrap our
1978 way up to SBCL starting with a bare gcc toolchain and human-readable
1979 source code (first the source to CLISP, then the source to SBCL).
1980 * A contributed module containing a partial implementation of the
1981 simple-streams interface has been included. (thanks to Rudi
1983 * A contributed module implementing the RFC1321 Message Digest
1984 Algorithm, known as MD5, has been included.
1985 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1986 no longer has any effect, as the code controlled by this feature
1987 has been deleted. (As far as we know, no-one has ever built using
1988 this feature, and its semantics were confused in any case).
1989 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1990 work on conditions (as required by the ANSI specification),
1991 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1992 expected behaviour on conditions. Users should note, however,
1993 that such behaviour is not required by the ANSI specification,
1994 and so use of this behaviour may render their code unportable.
1995 * fixed some bugs revealed by Paul Dietz' test suite:
1996 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1998 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1999 ** SXHASH on condition objects no longer returns NIL.
2000 ** :ALLOCATION :CLASS slots are better treated; their values are
2001 updated on class redefinition, and initforms inherited from
2002 superclasses are applied.
2003 ** REMOVE-METHOD returns its generic function argument even when
2004 no method was removed.
2005 ** SHARED-INITIALIZE now initializes the values of the requested
2006 slots, including those with :ALLOCATION :CLASS.
2007 ** ALLOCATE-INSTANCE now works on structure classes defined via
2008 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2010 ** SLOT-EXISTS-P now works on conditions, as well as structures
2012 ** MAKE-LOAD-FORM now has the required methods on
2013 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2014 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2015 keyword, and now implements the SLOT-NAMES argument.
2016 ** methods with &OPTIONAL arguments no longer allow too many
2017 arguments to be passed in the call without error.
2018 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2019 option is consistent with the required arguments of the generic
2020 function lambda list.
2021 * bug fix: REQUIRE accepts a string designator. (Thanks to
2023 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2024 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2025 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2026 * bug fix: The compiler now checks for duplicated variables in macro
2028 * bug fix: SETQ on globals returns the correct value.
2029 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2030 not a COMPILER-ERROR (followed by some other strange error on
2031 choosing the CONTINUE restart).
2032 * bug fix: make.sh and friends are now more consistent in the way that
2033 they look for GNU "make".
2035 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2036 * experimental native threads support (on x86 Linux >=2.4 only).
2037 This is not compiled in by default: you need to add :SB-THREAD to
2038 the target features. See the "Beyond ANSI" chapter of the manual
2040 * fix for longstanding nonANSIism: The old distinction between
2041 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2042 The return value from CL:FIND-CLASS is now a CLOS class, and
2043 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2044 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2045 * An interface to the MetaObject Protocol, as described in Kiczales,
2046 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2047 MIT Press, 1991, is available from the SB-MOP package.
2048 * incompatible change: the SB-PCL package should now be considered
2049 a private implementation detail, and no longer a semi-private MOP
2051 * minor incompatible change: due to rearrangement for threads, the
2052 control stack and binding stack are now allocated at arbitrary
2053 addresses instead of being hardcoded per-port. Users affected by
2054 this probably have to be doing advanced things with shared
2055 libraries, and will know who they are.
2056 * minor incompatible change: Previously, all --eval forms used were
2057 processed with READ before any of them were processed with EVAL.
2058 Now each --eval form is processed with both READ and EVAL before
2059 the next --eval form is processed. (Thus package operations like
2060 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2061 work as the user might reasonably expect.)
2062 * minor incompatible change: *STANDARD-INPUT* is now only an
2063 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2065 * minor incompatible change: Y-OR-N-P is now character-oriented,
2066 not line oriented. Also, YES-OR-NO-P now works without errors.
2067 (thanks to Antonio Martinez)
2068 * sb-aclrepl module improvements: an integrated inspector, added
2069 repl features, and a bug fix to :trace command.
2070 * Known functions, which cannot be open coded by the backend, are
2071 considered to be able to check types of their arguments. (fixing
2072 a bug report by Nathan J. Froyd)
2073 * fixed a bug in computing method discriminating functions: It is
2074 now possible to define methods specialized on classes which have
2075 forward-referenced superclasses. (thanks to Gerd Moellmann)
2076 * fixed evaluation order in optional entries (reported by Gilbert
2078 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2079 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2080 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2081 required-by-AMOP NAME argument, as well as CLASS and
2082 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2083 * fixed bug 20: DEFMETHOD can define methods using names that are
2084 not the proper names of classes to designate class specializers.
2085 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2086 Unix file descriptors, instead of blowing up. (thanks to Antonio
2088 * Garbage collection refactoring: user-visible change is that a
2089 call to the GC function during WITHOUT-GCING will not do garbage
2090 collection until the end of the WITHOUT-GCING. If you were doing
2091 this you were probably losing anyway.
2092 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2093 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2094 TYPEP the latter but not the former.
2095 * The compiler issues a full WARNING for calls to undefined functions
2096 with names from the CL package.
2097 * MAP-INTO for a vector destination is open coded. (reported by
2098 Brian Downing on c.l.l)
2099 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2100 documentation string.
2101 * fixed some bugs revealed by Paul Dietz' test suite:
2102 ** COPY-ALIST now signals an error if its argument is a dotted
2104 ** Condition slots are now accessed more correctly in the presence
2105 of multiple initargs for a given slot.
2106 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2107 exclude restarts of the same name associated with a different
2109 ** DEFCLASS of forward-referenced classes with another
2110 forward-referenced class in the superclasses list no longer
2112 ** Condition slots are now initialized once each, not multiple
2113 times. (thanks to Gerd Moellmann)
2114 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2116 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2117 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2118 arguments contain duplicated elements.
2119 ** RESTART-CASE understands local macros.
2120 ** RESTART-CASE associates exactly its own restarts with a condition.
2121 ** ENDP in safe mode checks its argument to be of type LIST.
2122 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2123 in question is unbound.
2124 ** Optimized MAKE-INSTANCE functions no longer cause internal
2125 assertion failures in the presence of duplicate initargs.
2126 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2127 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2128 propagated even in the ctor optimized implementation of
2130 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2132 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2133 * a better implementation of SXHASH on (simple) bit vectors,
2134 measured both in execution speed and in distribution of results
2135 over the positive fixnums, has been installed. Likewise, a better
2136 implementation of EQUAL for simple bit vectors is now available.
2137 * fixed CEILING optimization for a divisor of form 2^k.
2138 * fixed bug 240 (emitting extra style warnings "using the lexical
2139 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2140 by Antonio Martinez)
2141 * fixed SXHASH, giving different results for NIL depending on type
2142 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2143 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2144 variable. (found by Rolf Wester)
2145 * SBCL does not ignore type declarations for special
2146 variables. (reported by rif on c.l.l 2003-03-05)
2147 * some bug fixes in contrib/sb-aclrepl/
2148 * fixed some bugs revealed by Paul Dietz' test suite:
2149 ** a bug in the CONS type specifier, whereby the CAR and CDR
2150 types got intertwined, has been fixed;
2151 ** the type system is now able to reason about the interaction
2152 between INTEGER and RATIO types more completely;
2153 ** APPEND, [N]REVERSE and NRECONC check that those their
2154 arguments, which must be proper lists, are really so;
2155 ** An array specialized to be unable to hold elements has been
2156 implemented, as required -- yes, really -- by ANSI;
2157 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2158 on malformed property lists;
2160 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2161 * incompatible packaging change: in line with Unix convention,
2162 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2163 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2164 * REQUIRE and PROVIDE are now optionally capable of doing something
2165 useful. See the documentation string for REQUIRE.
2166 * infrastructure for a managed SBCL contrib system: contributed
2167 modules in this release include:
2168 ** the ASDF system definition facility;
2169 ** an interface to the BSD Sockets API;
2170 ** an ACL-like convenience interface to the repl;
2171 (thanks to Kevin Rosenberg)
2172 ** an implementation of ROTATE-BYTE, with efficient implementation
2174 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2175 gives a closer approximation to the right answer than previously.
2176 (thanks to Raymond Toy)
2177 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2178 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2179 optional environment arguments, as required by ANSI.
2180 * fixed bugs in other functions taking environment objects, allowing
2181 calls with an explicit NIL environment argument to be compiled
2183 * fixed bug 228: primary return values from
2184 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2185 COMPILE or FUNCTION.
2186 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2187 :TYPE will no longer signal errors on innocuous objects.
2188 * fixed bug 231b: SETQ is better at respecting type declarations in
2189 the lexical environment.
2190 * fixed a bug in DEFCLASS: classes named by symbols with no or
2191 unprintable packages can now be defined.
2192 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2193 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2194 * fixed compiler failure related to checking types of functions.
2195 (reported by Robert E. Brown)
2196 * the compiler is now much more consistent in its error-checking
2197 treatment of bounding index arguments to sequence functions: in
2198 (SAFETY 3) code, errors will be signalled in almost all cases if
2199 invalid sequence bounding indices are passed to functions defined
2200 by ANSI to operate on sequences.
2201 * fixed a bug in the build procedure: documentation of SBCL-specific
2202 packages is now preserved and available in the final Lisp image.
2203 * lifted FDEFINITION lookup out of loops in the implementation of
2204 many list operations. (thanks to Robert E. Brown)
2205 * fixed a bug in the reader: the #n# reader macro now works for
2206 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2207 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2208 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2209 E. Brown and Tony Martinez respectively)
2210 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2211 SLOT-BOUNDP: in particular, it is now able to optimize them much
2212 better, and is now not vulnerable to having packages renamed.
2213 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2214 always called when a slot is not present in an instance. (thanks
2216 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2217 longer signal an error on unreading a character following EOF on
2218 the previous constituent stream. (thanks to Tony Martinez)
2219 * fixed some bugs revealed by Paul Dietz' test suite:
2220 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2221 not just nonnegative fixnums;
2222 ** the logical bit-array operators such as BIT-AND now accept an
2223 explicit NIL for their "opt-arg" argument (to indicate a
2224 freshly-consed result bit-array);
2225 ** ELT now signals an error on an invalid sequence index in safe
2227 ** the type system is now cleverer about negations of numeric
2228 types, and consequently understands the BIGNUM and RATIO types
2230 ** the type system is now cleverer about the interaction between
2231 INTEGER and RATIO types: while bugs still remain, many more
2232 cases are accurately computed;
2233 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2234 if it is in the last clause;
2235 ** CONSTANTLY now correctly returns a side-effect-free function in
2237 ** DECLARE is no longer treated as a special-operator; in
2238 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2239 * incremented fasl file version number due to the change in the
2240 DEFSTRUCT-SLOT-DESCRIPTION structure.
2242 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2243 * minor incompatible change: code processed by the "interpreter" or
2244 EVAL now has a compilation optimization policy of (DEBUG 2)
2245 (changed from (DEBUG 1)) to improve debuggability of interactive
2246 development, and to allow the use of the debug RETURN command in
2248 * an experimental implementation of the RETURN command for the
2249 debugger has been included. (thanks to Frederik Kuivinen)
2250 * fixed bug 62: constraints were not propagated into a loop.
2251 * fixed bug in embedded calls of SORT (reported and investigated by
2253 * fixed some bugs revealed by Paul Dietz' test suite:
2254 ** printing and reading of arrays with some dimensions having
2255 length 0 (thanks to Gerd Moellmann);
2256 ** BOA constructor with &AUX argument without a default value does
2257 not cause a type error;
2258 ** CONSTANTP now returns true for all self-evaluating objects.
2260 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2261 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2262 accessors that are related by inheritance, as specified in the
2263 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2264 Valtteri Vuorikoski)
2265 * The compiler is now able to inline functions that were defined in
2266 a complex lexical environment (e.g. inside a MACROLET).
2267 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2269 * Support for the upcoming FreeBSD-5.0 release has been included.
2270 (thanks to Dag-Erling Smorgrav)
2271 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2272 effect when it is not in a toplevel context.
2273 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2274 stand a better chance of being correct. (thanks to Gerd
2276 * fixed bug in COERCE, which now signals an error on coercing a
2277 rational to a bounded real type which excludes the expected
2279 * The compiler is now able to derive types more accurately from the
2280 COERCE and COMPILE functions.
2281 * fixed bug 223: functional binding is considered to be constant
2282 only for symbols in the CL package.
2283 * fixed bug 231: SETQ did not check the type of a variable being set
2284 (reported by Robert E. Brown)
2285 * A new optimization for MAKE-INSTANCE has been included, fixing
2286 various bugs (including relating to :ALLOCATION :CLASS slots and
2287 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2289 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2290 ** As required by ANSI, LOOP now disallows anonymous collection
2291 clauses such as COLLECT I in conjunction with aggregate boolean
2292 clauses such as THEREIS (= I 1);
2293 ** LOOP now signals an error when any variable is reused in the
2294 same loop (including the potentially useful construct analogous
2295 to WITH A = 1 WITH A = (1+ A);
2296 ** IT is only a special loop symbol within the first clause of a
2297 conditional loop clause;
2298 ** LOOP with a typed iteration variable over a hashtable now
2299 signals a type error iff it should.
2300 * fixed some other bugs revealed by Paul Dietz' test suite:
2301 ** FILE-STREAM now names the class previously known as FD-STREAM;
2302 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2303 argument) no longer signals an error;
2304 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2305 of the slot symbol, rather than using the current package
2306 ((:CONC-NAME "") continues to intern the slot's name in the
2308 * incremented fasl file version number, because of the incompatible
2309 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2310 of the new implementation of DEFINE-COMPILER-MACRO.
2312 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2313 * Support for building SBCL for MIPS platforms running in
2314 little-endian mode has now been checked in, and basic
2315 functionality on said platforms verified.
2316 * minor incompatible change: PCL now records the pathname of a file
2317 in which methods and the like are defined, rather than its
2319 * minor incompatible change: TRUENAME now considers the truename of
2320 a file naming a directory to be the pathname with :DIRECTORY
2321 component indicating that directory.
2322 * minor incompatible change: a NAMED clause in the extended form of
2323 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2324 reason for the previous behaviour is unclear.
2325 * more systematization and improvement of CLOS and MOP conformance
2326 in PCL (thanks to Gerd Moellman and Pierre Mai):
2327 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2329 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2330 primary methods with no specializers;
2331 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2333 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2334 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2335 FORWARD-REFERENCED-CLASSes; error reporting on
2336 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2338 ** SXHASH on CLOS instances now uses a slot internal to the
2339 instance to return different numbers on distinct instances,
2340 while preserving the same return value through invocations of
2342 ** DEFMETHOD signals errors when methods with longer incongruent
2343 lambda lists are added to generic functions;
2344 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2345 CLASS, as specified in AMOP;
2346 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2347 on the order returned by the primary method for classes of
2348 class STANDARD-CLASS;
2349 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2350 * fixed some bugs shown by Paul Dietz' test suite:
2351 ** DOLIST puts its body in TAGBODY;
2352 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2354 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2355 value producing form;
2356 ** if more variables are given to PROGV than values, extra
2357 variables are bound and made to have no value;
2358 ** NSUBSTITUTE on list arguments gets the right answer with
2360 ** ELT signals an error of type TYPE-ERROR when the index argument
2361 is not a valid sequence index;
2362 ** LOOP signals (at macroexpansion time) an error of type
2363 PROGRAM-ERROR when duplicate variable names are found;
2364 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2365 ** FUNCALL of special-operators now cause an error of type
2367 ** PSETQ now works as required in the presence of side-effecting
2368 symbol-macro places;
2369 ** NCONC accepts any object as its last argument;
2370 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2372 ** loop-for-as-package does not require a package to be explicitely
2374 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2376 * fixed bug 166: compiler preserves "there is a way to go"
2377 invariant when deleting code.
2378 * fixed bug 172: macro lambda lists with required arguments after
2379 &REST arguments now cause an error to be signalled. (thanks to
2381 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2382 bug. (thanks to Matthew Danish)
2383 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2385 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2386 arguments are lexically rebound. (thanks to Gerd Moellmann and
2388 * fixed bug 194: error messages are now more informative when there
2389 is no primary method applicable in a call to a generic
2390 function. (thanks to Gerd Moellmann)
2391 * fixed bug in command line argument checking (thanks to Julian
2393 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2394 (thanks to Matthew Danish)
2395 * incremented fasl file version number, because of the
2396 SXHASH-related changes in the layout of CLOS data structures
2398 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2399 * minor incompatible change: The runtime (the Unix executable named
2400 "sbcl") is now much pickier about the .core files it will load.
2401 Essentially it now requires .core files to descend from the same
2402 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2403 as the runtime does. (The intent is to prevent the crashes which
2404 can occur, and which can even be reported as mysterious failures,
2405 when people patch the sources or change the build parameters
2406 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2407 sbcl and .core files.)
2408 * fixed bug: VALUES-LIST is no longer optimized away.
2409 * fixed bug 142: The FFI conversion of C string values to Lisp
2410 string values no longer conses excessively. (thanks to Nathan
2411 Froyd porting Raymond Toy's fix to CMU CL)
2412 * began to systematize and improve MOP conformance in PCL (thanks to
2413 Nathan Froyd, Gerd Moellman and Pierre Mai):
2414 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2416 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2417 ** FINALIZE-INHERITANCE is now called on class finalization;
2418 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2419 argument precedence order.
2420 * fixed bug 202: The compiler no longer fails on functions whose
2421 derived types contradict their declared type.
2422 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2423 so it can be non-toplevel.
2424 * The fasl file version number has changed (because of the new
2425 implementation of DEFMACRO).
2426 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2427 safe code, that any length requirement by their type-specifier
2428 argument is valid. The exceptions to this are described in bug
2430 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2431 functions now check the types of their inputs as required by ANSI.
2432 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2433 introduced symbol is DECLAREd to be SPECIAL.
2434 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2435 * fixed bug: :COUNT argument to sequence functions may be negative
2436 * fixed bug: body of DO-SYMBOLS may contain declarations
2437 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2438 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2439 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2440 symbol macro only once
2441 * fixed printing of call frame when argument list is unavailable
2442 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2443 * compiler no longer signals WARNING on unknown keyword
2446 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2447 * A beta-quality port to the mips architecture running Linux,
2448 based on the old CMUCL backend, has been made. It has been tested
2449 on a big-endian kernel, and works sufficiently well to be able to
2450 rebuild itself; it has not been tested in little-endian mode.
2451 * fixed an inconsistency between gencgc.c and purify.c which made
2452 dumping/loading .core files unreliable
2453 * fixed bug 120a: The compiler now deals correctly with IFs where
2454 the consequent is the same as the alternative, instead of
2455 misderiving the return type. (thanks to Alexey Dejneka)
2456 * fixed bug 113: Logical pathnames are now dumpable (the logical
2457 host is resolved at load-time, throwing an error if it is not
2459 * fixed bug 174: FORMAT's error message is slightly clearer when a
2460 non-printing character is used in a format directive.
2461 * fixed several bugs in compiler checking of type declarations, i.e.
2462 violations of the Python "declarations are assertions" principle
2463 (thanks to Alexey Dejneka)
2464 * fixed several bugs in PCL's error checking (thanks to Gerd
2466 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2468 * fixed bug in compilation of functions as first class values
2469 (thanks to Antonio Martinez-Shotton)
2470 * The compiler's handling TYPE-ERRORs which it can prove will
2471 inevitably happen at runtime has been cleaned up and corrected
2472 in several ways. (thanks to Alexey Dejneka)
2473 * improved argument type checking for various basic arithmetic
2474 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2475 lavished on them in the past that they can be compiled in many
2476 ways in different special cases
2477 * fixed bug 181: compiler checks validity of user supplied type
2479 * cleaned up code flushing in optimization: Function calls which
2480 should signal errors for safety purposes (e.g. which ANSI says
2481 should signal errors when their arguments are of incorrect type)
2482 are no longer optimized away.
2483 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2484 * incremented fasl file version number, because changes in the
2485 implementation of sequence functions like COERCE caused
2486 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2487 in old inline expansions) to become undefined. (Actually these
2488 changes were later undone, so we might very well be binary
2489 compatible with 0.7.7 after all, but leaving the version number
2490 incremented seemed like the simplest and most conservative
2493 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2494 * An alpha-quality port to the parisc architecture running Linux,
2495 based on the old CMUCL backend, has been made. This, even more so
2496 than the other backends, should be considered still a work in
2497 progress; known problems include that the Linux kernel in 64-bit
2498 mode does not propagate the correct sigcontext structure to
2499 userspace, and consequently SBCL on a parisc64 kernel will not
2501 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2502 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2503 suppressed.) Also now that the compiler is looking at declarations
2504 in the environment, it checks optimization declarations as well,
2505 and suppresses inlining when (> DEBUG SPEED).
2506 * More fixes have been made to treatment of floating point exception
2507 treatment and other Unix signals. In particular, floating point
2508 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2509 * The detection and handling of control stack exhaustion (infinite
2510 or very deeply nested recursion) has changed. Stack exhaustion
2511 detection is now done by write-protecting pages at the OS level
2512 and applies at all optimization settings; when found, a
2513 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2514 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2515 be caught using IGNORE-ERRORS.
2516 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2517 correctly when transforming recursive LABELS functions to LETs.
2518 (thanks to Alexey Dejneka)
2519 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2520 that are names of constants or global variables.
2521 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2522 alien routines with docstrings.
2523 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2524 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2526 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2527 (thanks to Eric Marsden)
2528 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2529 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2530 * Bug fix: The SPARC backend can now compile functions involving
2531 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2532 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2533 to a functions expecting a PATHNAME-DESIGNATOR.
2534 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2535 lambda lists. (thanks to Alexey Dejneka)
2536 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2537 functions so that EXPORTing the name of the class doesn't cause
2538 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2539 to Antonio Martinez for reporting this)
2540 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2541 constant-folded without failing an assertion. (thanks to Einar
2542 Floystad Dorum for reporting this)
2543 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2544 types are now dealt with more correctly by the compiler.
2545 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2546 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2547 behaviour analogous to RENAME-FILE. This puts its behaviour more
2548 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2550 * The fasl file version number has changed again. (because of the
2551 bug fix involving the names of PCL MAKE-INSTANCE functions)
2553 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2554 * bug fix: Floating point exceptions are treated much more
2555 consistently on the x86/Linux and PPC/Linux platforms.
2556 * Array initialization with :INITIAL-ELEMENT is now much faster for
2557 cases when the compiler cannot open code the array creation, but
2558 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2559 array accesses have also seen a speed increase.
2560 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2561 specified. (This was at the root of some bad interactions between
2562 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2564 * bug fix: Internal error arguments for undefined functions are now
2565 computed correctly on the PPC/Linux platform.
2566 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2567 Raymond Toy's patch for CMU CL)
2568 * Support for the Solaris 9 operating environment has been included
2569 (thanks to Daniel Merritt)
2570 * A very ugly but hopefully complete draft of the missing FFI chapter
2571 of the manual has been created by reformatting the corresponding
2572 CMU CL manual chapter into (currently very ugly and incoherent)
2573 DocBook and bringing it up to date for SBCL behavior. Thus, the
2574 manual is now essentially complete, at least by my extreme
2575 once-and-only-once standards, whereby it's acceptable to refer to
2576 the doc strings of SB-EXT functions as the primary documentation.
2577 * The fasl file version number has changed again, due to cleanup of
2578 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2579 is no longer a static symbol.)
2581 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2582 * SBCL now builds with OpenMCL (version 0.12) as the
2583 cross-compilation host; also, more progress has been made toward
2584 bootstrapping under CLISP.
2585 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2587 * bug 158 fixed: The compiler can now deal with integer loop
2588 increments different from 1; fixing this turned out also to fix
2590 * bug 169 fixed: no more bogus warnings about using lexical bindings
2591 despite the presence of perfectly good SPECIAL declarations (thanks
2592 to David Lichteblau)
2593 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2594 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2595 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2597 * bug 180 fixed: Method combination specifications no longer ignore
2598 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2599 * bug fix: Structure type predicate functions now check their argument
2600 count as they should.
2601 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2602 correctly. (thanks to Pierre Mai)
2603 * minor incompatible change: The --noprogrammer option is deprecated
2604 in favor of the new --disable-debugger option, which is very similar.
2605 (The major difference is that it takes effect at a slightly different
2606 time at startup, causing handling of errors in --sysinit and
2607 --userinit files will be affected differently.) The
2608 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2609 been added to allow this functionality to be controlled from ordinary
2610 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2611 maintainers, who might want to run non-interactive scripts to
2612 build SBCL cores which will later be used interactively.)
2613 * minor incompatible change: The LOAD function no longer, when given
2614 a wild pathname to load, loads all files matching that pathname.
2615 Instead, an error of type FILE-ERROR is signalled.
2617 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2618 * bug 147 fixed: The compiler preserves its block link/count
2619 invariants more correctly now so that it doesn't crash. (thanks
2621 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2623 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2624 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2625 no-value-for-this-character as NIL (as specified by ANSI).
2626 * HOST-NAMESTRING on physical pathnames now returns a string that is
2627 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2628 (thanks to Christophe Rhodes)
2629 * The Alpha port handles icache flushing more correctly. (thanks to
2631 * More progress has been made toward bootstrapping under CLISP. (thanks
2632 to Christophe Rhodes)
2633 * The fasl file format has changed again, because dynamic loading
2634 on OpenBSD (which has non-ELF object files) motivated some cleanups
2635 in the way that foreign symbols are transformed and passed around.
2636 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2637 is no longer treated as whitespace by the reader, but instead as
2638 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2639 |A
\7fB|, instead of A as it used to.
2641 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2642 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2643 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2644 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2645 this as of 0.7.1.45, but was left out of the previous news section
2646 (thanks to Dan Barlow)
2647 * SBCL now runs on the Solaris operating system on SPARC architectures
2648 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2649 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2650 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2651 * SPARC backend cleanups, allowing builds of cores optimized for V8
2652 and V9 SPARCS, and also emission of code targeted to a particular
2653 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2655 * SBCL is closer to bootstrapping under CLISP, thanks to various
2656 fixes by Christophe Rhodes.
2657 * The fasl file format has changed again, to allow the compiler's
2658 INFO database to support symbol macros.
2659 * The user manual (in doc/) is formatted into HTML more nicely.
2660 (thanks to coreythomas)
2661 * The system is smarter about SUBTYPEP relationships, especially
2662 those involving NOT types (including types such as ATOM which are
2663 represented internally using NOT types). Thus SUBTYPEP is less
2664 likely to return (VALUES NIL NIL) in general, and in particular
2665 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2667 * The fasl file format has changed again, because the internal
2668 representation of types now includes a new slot to support the new
2669 SUBTYPEP-of-NOT-types logic.
2670 * (not a change in the main branch of SBCL, but a related prototype
2671 which can hopefully be merged into the main branch of SBCL in the
2672 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2673 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2674 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2675 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2676 dispatch macro character combinations. (thanks to Alexey Dejneka)
2677 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2678 that superficially look like logical namestrings correctly.
2679 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2681 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2682 default supplied-p) for &optional and &key arguments. (thanks to
2685 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2686 * incompatible change: The compiler is now less aggressive about
2687 tail call optimization, doing it only when (> SPACE DEBUG) or
2688 (> SPEED DEBUG). (This is an incompatible change because there are
2689 programs which relied on the old CMU-CL-style behavior to optimize
2690 away their unbounded recursion which will now die of stack overflow.)
2691 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2692 for non-GENCGC systems has been increased to 20M (since that
2693 seems much closer to the likely performance optimum for modern
2694 systems than the old 4M value was)
2695 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2696 and *DEBUG-PRINT-LEVEL*
2697 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2698 of CMU CL's support for SPARC, and various endianness and other
2699 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2700 * new syntactic sugar for the Unix command line: --load foo.bar is now
2701 an alternate notation for --eval '(load "foo.bar")'.
2703 ** The system now detects stack overflow and handles it gracefully,
2704 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2705 optimization settings. (This is a good thing in general, and
2706 its introduction in this version should be particularly timely
2707 for anyone whose code fails because of suppression of tail
2709 ** The system now hunts for the C variable "environ" in a more
2710 devious way, to avoid segfaults when the C library version differs
2711 between compile time and run time. (thanks to Christophe Rhodes)
2712 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2713 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2714 ** The compiler no longer issues bogus style warnings for undefined
2715 classes in the same source file as the DEFCLASSes which defined
2716 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2718 ** fixes in CONDITION class precedence list for undefined function
2719 errors (thanks to Alexei Dejneka)
2720 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2721 correctly. (thanks to Dan Barlow)
2722 ** portability fixes aiming at bootstrapping under CLISP (thanks
2723 to Dave McDonald and Christophe Rhodes)
2724 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2725 ** fixes in type translation and and type inference (thanks to
2727 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2728 ** various fixes in the new ports (thanks to Dan Barlow)
2729 * several changes related to debugging:
2730 ** suppression of tail recursion, as noted above
2731 ** stack overflow detection, as noted above
2732 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2733 is now the default. (For some time encapsulation has been more
2734 reliable than the breakpoint-based :ENCAPSULATE NIL
2735 implementation, at least on X86 systems; and I just noticed that
2736 encapsulation also seems closer to the spirit of the ANSI
2739 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2741 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2742 up properly again. (There was a packaging bug in 0.7.0 which
2743 left their definitions in SB-SYS::LOAD-FOREIGN and
2744 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2745 vital for most things which interface to C-level interfaces,
2746 like extensions working with sockets or databases or
2747 Perl-compatible regexes or whatever, and the need to fix
2748 this bug is the main reason that 0.7.1 was released so
2750 ** DEFGENERIC is now choosier about the methods it redefines, so that
2751 reLOADing a previously-LOADed file containing DEFGENERICs does
2752 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2753 cycle now works as it should. (thanks to Alexey Dejneka)
2754 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2755 fixed by Christophe Rhodes. (He actually submitted this patch
2756 months ago, and I delayed until after 0.7.0.)
2757 ** Bug 111 (internal compiler confusion about runtime checks on
2758 FUNCTION types) was fixed by Alexey Dejneka.
2759 * Some internal cleanups (getting rid of variables which aren't
2760 needed now that the byte interpreter is gone) caused the fasl
2761 file format number to change again.
2763 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2764 * major incompatible change: The default fasl file extension, i.e. the
2765 default extension for files produced by COMPILE-FILE, has changed
2766 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2768 ** There are many changes in the implementation of the compiler.
2769 SBCL is now essentially a compiler-only implementation of ANSI
2770 Common Lisp. EVAL still "interprets" a few special cases, but
2771 almost all the interesting cases are handled by creating
2772 a LAMBDA expression, calling COMPILE on it, then calling
2773 FUNCALL on the result.
2774 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2775 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2776 Since the code is newer, there might still be some new bugs
2777 (though not as many as before Martin Atzmueller's fixes:-). But
2778 the new code is substantially simpler and clearer, and hopefully
2779 any remaining bugs will be simpler, less fundamental, and more
2780 fixable then the bugs in the old code.
2781 ** The revised compiler is still a little unsteady on its feet.
2783 *** The debugging information it produces (particularly the names
2784 of FUNCTION objects) is sometimes much less useful than what
2785 the old compiler produced.
2786 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2787 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2788 within a MACROLET) has been temporarily weakened.
2789 ** There are new compiler optimizations for various functions:
2790 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2791 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2792 *** the math functions TRUNCATE, FLOOR, and CEILING
2793 *** the function-of-all-trades COERCE
2794 Mostly these should be transparent, but there's one
2795 potentially-annoying problem (bug 117): when the compiler
2796 inline-expands a function and does type analysis on the result,
2797 it can create control paths which have type mismatches, and
2798 when it can't prove that those control paths aren't taken,
2799 it will issue WARNINGs about the type mismatches. This is
2800 a particular problem in practice for the new sequence functions.
2801 It's not clear how this should be fixed, and for now, a
2802 workaround is given in the entry for 117 in the BUGS file.
2803 ** (Because of the interaction between the two previous items --
2804 occasional inlining problems and new inline expansions -- some
2805 of the new sequence function optimizations won't really kick in
2806 completely until debugging information, and then inlining, are
2807 straightened out in some future version.)
2808 * minor incompatible changes:
2809 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2810 DIRECTORY no longer implicitly promotes NIL slots of its
2811 pathname argument to :WILD. In particular, when you ask for the
2812 contents of a directory (which you used to be able to do without
2813 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2814 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2815 ** changes in behavior that ANSI explicitly defines to be
2816 implementation dependent:
2817 *** The new compiler-only implementation still conforms with ANSI,
2818 but acts a little different than before. Besides the obvious
2819 changes in performance tradeoffs (that the cost per form passed
2820 to EVAL has gone up, and the cost per form executed by EVAL
2821 has gone down), the behavior of the system changes a little
2822 because there are no longer any interpreted function objects.
2823 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2824 e.g. doing COMPILE on the output of interactive DEFUN is
2826 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2828 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2829 has changed from (:CL) to NIL.
2830 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2831 e.g. #\Newline and #\Tab, don't have names specified in the
2832 ANSI Common Lisp standard, is now based on their ASCII symbolic
2833 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2834 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2835 are no longer used for output.
2836 ** changes in internal implementation constants:
2837 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2838 4 million. (If your application spends a lot of time GCing and
2839 you have a lot of RAM, you might want to experiment with
2840 increasing it even more.)
2841 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2842 However, almost all old code should still continue to work without
2843 immediate update, as SB-C-CALL is now a (deprecated) nickname
2845 ** Old operator names in the style DEF-FOO are now deprecated in
2846 favor of new corresponding names DEFINE-FOO, for consistency with
2847 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2848 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2849 internal symbols, but a few supported extensions like
2850 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2851 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2852 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2853 etc. as you get deeper into recursive calls to the debugger
2854 command loop, instead of the old "5]", "5]]", "5]]]"
2855 sequence. (I was motivated to do this when squabbles between
2856 ILISP and SBCL left me very deeply nested in the debugger. In the
2857 short term, this change will probably provoke more ILISP/SBCL
2858 squabbles, but hopefully it will be an improvement in the long run.)
2859 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2860 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2861 function, since declaiming return types involving aliens is
2862 (1) annoyingly messy to do by hand and (2) vital to efficient
2863 compilation of code which calls such functions.
2864 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2865 longer reexported by the SB-EXT package. They're solely useful
2866 for alien code, so it seems more logical that you should get
2867 them from the SB-ALIEN package, not in SB-EXT.
2868 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2869 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2870 features. Instead, the code that they used to control is always
2871 built into the system.
2872 * many other bug fixes
2873 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2874 advantage of the new EVAL-WHEN stuff and to clean them up in
2875 general, and they are now more ANSI-compliant in a number of
2876 ways. Martin Atzmueller is responsible for a lot of this.
2877 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2879 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2880 *** correct ERROR type for various file operations
2881 *** some fixes for Lisp streams
2882 *** DEFMETHOD syntax checking
2883 *** changing old weird representation of debug information as
2884 strings (which, among their other deficiencies, don't transform
2885 correctly when you rename packages, and don't change their
2886 print representation when you change things like *PACKAGE*
2887 and *PRINT-LENGTH*) to symbols and lists of symbols
2888 He also made several improvements and fixed several bugs in DESCRIBE.
2889 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2891 *** misbehavior of WRITE-STRING/WRITE-LINE
2892 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2893 and several other LOOP problems as well
2894 *** DIRECTORY when similar filenames are present
2895 *** DEFGENERIC with :METHOD options
2896 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2897 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2898 *** argument ordering in FIND with :TEST option
2899 *** mishandled package designator argument in APROPOS-LIST
2900 *** various problems in the backquote readmacro
2901 *** a bug in APROPOS
2902 *** probably some others that I'm not describing very well here,
2903 since the CVS log documents them by reference to sbcl-devel
2904 messages, and the SourceForge archives aren't working well.:-(
2905 ** Dan Barlow improved the Alpha port (and is making progress on the
2906 PPC port, for those of you who think different).
2907 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2908 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2909 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2910 continues to work on the SPARC port (for those of you in a position
2911 to look down upon our little PC-compatible boxes from a great height).
2912 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2913 and uses that copy, rather than the current dynamic value, when
2914 it's trying to decide whether to truncate output. Thus e.g.
2915 (let ((*print-lines* 50))
2916 (pprint-logical-block (stream nil)
2918 (let ((*print-lines* 8))
2919 (print (aref possiblybigthings i) stream)))))
2920 should now truncate the logical block only at 50 lines, instead of
2921 often truncating it at 8 lines, as it did before.
2922 * The doc/cmucl/ directory, containing old CMU CL documentation
2923 from the time of the fork, is no longer part of the base system.
2924 SourceForge has shut down its anonymous FTP service, and with it
2925 my original plan for distributing the old CMU CL documentation
2926 there. For now, if you need these files you can download an old
2927 SBCL source release and extract them from it.
2928 * The fasl file version number changed again, for dozens of reasons,
2929 some of which are apparent above.
2931 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2932 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2933 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2934 patch, so that DISASSEMBLE output is much nicer.
2935 * The code in the SB-PROFILE package now seems reasonably stable.
2936 I still haven't decided what the final interface should look like
2937 (I'd like PROFILE to interact cleanly with TRACE, since both
2938 facilities use function encapsulation) but if you have a need
2939 for profiling now, you can probably use it successfully with
2940 the current CMU-CL-style interface.
2941 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2942 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2943 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2944 link now returns the dangling link itself, and for similar
2945 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2946 link itself. (In these cases the old code signalled an error and
2947 looped endlessly, respectively.) Thus, DIRECTORY now works even
2948 in the presence of dangling and cyclic symbolic links.
2949 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2950 is now a supported extension again, since the consensus on
2951 sbcl-devel was that it can be useful for ordinary development
2952 work, not just for debugging SBCL itself.
2953 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2954 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2955 that currently-defined functions might be redefined later with
2956 different return types.
2957 * Hash tables can be printed readably, as inspired by CMU CL code
2958 of Eric Marsden and SBCL code of Martin Atzmueller.
2959 * better error handling in CLOS method combination, thanks to
2960 Martin Atzmueller porting Pierre Mai's CMU CL patches
2961 * more overflow fixes for >16Mbyte I/O buffers
2962 * A bug in READ has been fixed, so that now a single Ctrl-D
2963 character suffices to cause end-of-file on character streams.
2964 In particular, now you only need one Ctrl-D at the command
2965 line (not two) to exit SBCL.
2966 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2968 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2969 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2970 does the right thing.
2971 * The implementation of some type tests, especially for CONDITION
2972 types, is now tidier and maybe faster, due to CMU CL code
2973 originally by Douglas Crosher, ported by Martin Atzmueller.
2974 * Some math functions have been fixed, and there are new
2975 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2976 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2977 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2978 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2979 likely to be built into sbcl-0.7.0.)
2980 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2981 CL, ported to SBCL by Martin Atzmueller)
2982 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2983 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2984 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2985 * There's a new slam.sh hack to shorten the edit/compile/debug
2986 cycle for low-level changes to SBCL itself, and a new
2987 :SB-AFTER-XC-CORE target feature to control the generation of
2988 the after-xc.core file needed by slam.sh.
2989 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2990 COMPILE-FILE is no longer supported, so that now every function
2991 gets an entry point, so that block compilation looks a little
2992 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2993 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2994 returns the number of bytes consed since the system started,
2995 rather than the number consed since the first time the function
2996 was called. (The new definition parallels ANSI functions like
2997 CL:GET-INTERNAL-RUN-TIME.)
2998 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2999 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3000 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3001 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3002 * Fasl file version numbers are now independent of the target CPU,
3003 since historically most system changes which required version
3004 number changes have affected all CPUs equally. Similarly,
3005 the byte fasl file version is now equal to the ordinary
3008 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3009 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3010 is no longer recognized. I apologize for this, because it was
3011 listed in SB-EXT as a supported extension, but I found that
3012 its existing behavior was poorly specified, as well as incorrectly
3013 specified, and it looked like too much of a mess to straighten it
3014 out. I have enough on my hands trying to get ANSI stuff to work..
3015 * many patches ported from CMU CL by Martin Atzmueller, with
3016 half a dozen bug fixes in pretty-printing and the debugger, and
3017 half a dozen others elsewhere
3018 * fixed bug 13: Floating point infinities are now supported again.
3019 They might still be a little bit flaky, but thanks to bug reports
3020 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3021 as flaky as they were.
3022 * The --noprogrammer command line option is now supported. (Its
3023 behavior is slightly different in detail from what the old man
3024 page claimed it would do, but it's still appropriate under the
3025 same circumstances that the man page talks about.)
3026 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3027 are now supported, and enabled by default. Thus, the compiler can
3028 handle many floating point and complex operations much less
3029 inefficiently. (Thus e.g. you can implement a complex FFT
3031 * The compiler now detects type mismatches between DECLAIM FTYPE
3032 and DEFUN better, and implements CHECK-TYPE more correctly, and
3033 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3034 thanks to patches from Martin Atzmueller.
3035 * various fixes to make the cross-compiler more portable to
3036 ANSI-conforming-but-different cross-compilation hosts (notably
3037 Lispworks for Windows, following bug reports from Arthur Lemmens)
3038 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3039 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3040 fixed thanks to Pierre Mai's CMU CL patches.
3041 * fixes to keep the system from overflowing internal counters when
3042 it tries to use i/o buffers larger than 16M bytes
3043 * fixed bug 45a: Various internal functions required to support
3044 complex special functions have been merged from CMU CL sources.
3045 (When I was first setting up SBCL, I misunderstood a compile-time
3046 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3047 * improved support for type intersection and union, fixing bug 12
3048 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3049 more obscure bugs as well
3050 * some steps toward byte-compiling non-performance-critical
3051 parts of the system, courtesy of patches from Martin Atzmueller
3052 * Christophe Rhodes has made some debian packages of sbcl at
3053 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3054 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3055 stable, but are nonetheless usable. When he's ready, I'd be happy
3056 to add them to the SourceForge "File Releases" section. (And if
3057 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3058 * new fasl file format version number (because of changes in
3059 internal representation of (OR ..) types to accommodate the new
3060 support for (AND ..) types, among other things)
3062 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3063 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3065 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3067 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3068 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3069 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3070 * DESCRIBE now works on structure objects again.
3071 * Most function call argument type mismatches are now handled as
3072 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3073 know whether the function will be redefined before the call is
3074 executed. (The compiler could flag local calls with full WARNINGs,
3075 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3076 it doesn't keep track of enough information to know whether calls
3077 are local in this sense.)
3078 * Compiler output is now more verbose, with messages truncated
3079 later than before. (There should be some supported way for users
3080 to override the default verbosity, but I haven't decided how to
3081 provide it yet, so this behavior is still controlled by the internal
3082 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3083 src/compiler/ir1util.lisp.)
3084 * Fasl file format version numbers have increased again, because
3085 support for the Gray streams extension changes the layout of the
3086 system's STREAM objects.
3087 * The Gray subclassable streams extension now works, thanks to a
3088 patch from Martin Atzmueller.
3089 * The full LOAD-FOREIGN extension (not just the primitive
3090 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3091 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3092 but like most other programs, it defaults to copying the Unix
3093 environment from the original process instead of starting the
3094 new process in an empty environment.
3095 * Extensions which manipulate the Unix environment now support
3096 an :ENVIRONMENT keyword option which doesn't smash case or
3097 do other bad things. The CMU-CL-style :ENV option is retained
3098 for porting convenience.
3099 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3100 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3102 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3104 * A patch from Martin Atzmueller seems to have solved the SIGINT
3105 problem, and as far as we know, signal-handling now works cleanly.
3106 (If you find any new bugs, please report them!)
3107 * The system no longer defaults Lisp source file names to types
3108 ".l", ".cl", or ".lsp", but only to ".lisp".
3109 * The compiler no longer uses special default file extensions for
3110 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3111 seems to expect a single default extension for all compiled code,
3112 and there's no compelling reason to try to stretch the standard
3113 to allow two different extensions.) Instead, byte-compiled files
3114 default to the same extension as native-compiled files.
3115 * Fasl file format version numbers have increased again, because
3116 a rearrangement of internal implementation packages made some
3117 dumped symbols in old fasl files unreadable in new cores.
3118 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3119 many fewer weird special cases.
3120 * Bug #17 (differing COMPILE-FILE behavior between logical and
3121 physical pathnames) has been fixed, and some related misbehavior too,
3122 thanks to a patch from Martin Atzmueller.
3123 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3124 by Tim Moore, ported to SBCL by Martin Atzmueller.
3125 * Martin Atzmueller fixed several filesystem-related problems,
3126 including bug #36, in part by porting CMU CL patches, which were
3127 written in part by Paul Werkowski.
3128 * More compiler warnings in src/runtime/ are gone, thanks to
3129 more patches from Martin Atzmueller.
3130 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3133 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3135 * DESCRIBE now works on CONDITION objects.
3136 * The debugger now handles errors which arise when trying to print
3137 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3139 * The build system now uses an additional file, customize-target-features.lisp,
3140 to allow local modifications to the target *FEATURES* list. (The point of
3141 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3142 debugging features enabled, without having to worry about propagating it
3143 into everyone's system when I do a "cvs update".) When no
3144 customize-target-features.lisp file exists, the target *FEATURES* list
3145 should be constructed the same way as before.
3146 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3147 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3148 warn about behavior which is undefined under the ANSI spec. Note: This
3149 is specified by ANSI, but it's not very popular with programmers.
3150 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3151 macro in the SBCL sources for an example of a workaround which you
3152 might use to make portable ANSI-standard code which does what you want.
3153 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3154 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3155 when it's used as a non-toplevel form. (This is required in order
3156 to implement the DEFCONSTANT-EQX macro.)
3157 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3158 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3159 funny things, due to bugs in the implementation of EVAL-WHEN
3160 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3161 non-toplevel DEFCONSTANTs are uncommon.)
3162 * The core file version number and fasl file version number have been
3163 incremented, because the old noncompliant DEFCONSTANT behavior involved
3164 calling functions which no longer exist, and because I also took the
3165 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3166 * fixed bug 1 (error handling before read-eval-print loop starts), and
3167 redid debugger restarts and related debugger commands somewhat while
3169 ** The QUIT debugger command is gone, since it did something
3170 rather different than the SB-EXT:QUIT command, and since it never
3171 worked properly outside the main toplevel read/eval/print loop.
3172 Invoking the new TOPLEVEL restart provides the same functionality.
3173 ** The GO debugger command is also gone, since you can just invoke
3174 the CONTINUE restart directly instead.
3175 ** The TOP debugger command is also gone, since it's redundant with the
3176 FRAME 0 command, and since it interfered with abbreviations for the
3178 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3180 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3181 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3182 some porting work by Martin Atzmueller)
3183 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3184 with a boatload of patches which he ported to SBCL. Now that those
3186 ** The system tries to make sure that its low-priority messages
3187 are prefixed by semicolons, to help people who like to use
3188 syntax highlighting in their ILISP buffer. (This patch
3189 was originally due to Raymond Toy.)
3190 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3191 patches originally written by Raymond Toy.
3192 ** The compiler understands coercion between single-value and
3193 multiple-VALUES type expressions better, getting rid of some very
3194 weird behavior, thanks to patches originally by Robert MacLachlan
3195 and Douglas Crosher.
3196 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3197 lambda lists, thanks to a patch originally by Pierre Mai.
3198 ** The system no longer bogusly warns about "abbreviated type
3200 ** The compiler gets less confused by inlining and RETURN-FROM,
3201 thanks to some patches originally by Tim Moore.
3202 ** The system no longer hangs when dumping circular lists to fasl
3203 files, thanks to a patch originally from Douglas Crosher.
3204 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3205 undefined function error.
3206 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3207 in structure slot initforms) for now, documented workaround instead:-|
3208 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3209 * fixed bug 5: added stubs for various Gray stream functions called
3210 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3211 installed, at least appropriate type errors are generated
3212 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3213 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3214 consistently in DEFMETHOD forms.
3215 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3216 it doesn't seem to affect SBCL after all
3217 * The C runtime system now builds with better optimization and many
3218 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3220 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3222 * The system is now under CVS at SourceForge (instead of the
3223 CVS repository on my home machine).
3224 * The new signal handling code has been tweaked to treat register
3225 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3226 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3227 downstream, e.g. in debug-int.lisp, has implicit dependencies
3228 on the unsignedness of integer representation of machine words,
3229 and that caused the system to bomb out with infinite regress
3230 when trying to recover from type errors involving signed values,
3231 e.g. (BUTLAST '(1 2 3) -1).)
3232 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3233 (This was one of the bugs Peter Van Eynde reported back in July.)
3234 * The system now uses code inspired by Colin Walters' O(N)
3235 implementation of MAP (from the cmucl-imp@cons.org mailing
3236 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3237 inline the MAP operation, and there is more than one
3238 sequence argument to the MAP call (so that it can't just
3239 do ETYPECASE once and for all based on the type of the
3240 single sequence argument). (The old non-inline implementation
3241 of the general M-argument sequence-of-length-N case required
3242 O(M*N*N) time when any of the sequence arguments were LISTs.)
3243 * The QUIT :UNIX-CODE keyword argument has been renamed to
3244 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3245 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3247 * Raymond Wiker's patches to port dynamic loading from Linux to
3248 FreeBSD have been added.
3249 * The BUGS file is now more nearly up to date, thanks in large part
3250 to Martin Atzmueller's review of it.
3251 * The debugger now flushes standard output streams before it begins
3252 its output ("debugger invoked" and so forth).
3253 * The core version number and fasl file version number have both
3254 been incremented, because of incompatible changes in the layout
3256 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3257 used to not be called for a saved Lisp image.)
3258 * Martin Atzmueller's version of a patch to fix a compiler crash,
3259 as posted on sbcl-devel 13 September 2000, has been installed.
3260 * Instead of installing Martin Atzmueller's patch for the
3261 compiler transform for SUBSEQ, I deleted the compiler transform,
3262 and transforms for some similar consing operations.
3263 * A bug in signal handling which kept TRACE from working on OpenBSD
3265 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3266 away by constant folding
3267 * The system now defines its address space constants in one place
3268 (in the Lisp sources), and propagates them automatically elsewhere
3269 (through GENESIS and the sbcl.h file). Therefore, patching the
3270 address map is less unnecessarily tedious and error-prone. The
3271 Lisp names of address space constants have also been systematized.
3272 * CVS tags like dollar-Header-dollar have been removed from
3273 the sources, because they have never saved me trouble and
3274 they've been source of trouble working with patches and other
3275 diff-related operations.
3276 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3277 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3279 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3281 * The system has been ported to OpenBSD.
3282 * The system now compiles with a simple "sh make.sh" on the systems
3283 that it's supported on. I.e., now you no longer need to tweak
3284 text in the target-features.lisp-expr and symlinks in src/runtime/
3285 by hand, the make.sh takes care of it for you.
3286 * The system is no longer so grossly inefficient when compiling code
3287 involving vectors implemented as general (not simple) vectors (VECTOR T),
3288 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3289 which dares to use the various sequence functions on non-simple
3290 vectors, takes less of a performance hit.
3291 * There is now a primitive type predicate VECTOR-T-P
3292 to test for the (VECTOR T) type, so that e.g.
3293 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3294 can now be compiled with some semblance of efficiency. (The old code
3295 turned the type declaration into a full call to %TYPEP at runtime!)
3296 * AREF on (VECTOR T) is still not fast, since it's still compiled
3297 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3298 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3299 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3300 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3301 without making full calls to SUBTYPEP at runtime.
3302 (Some analogous efficiency issues for non-simple vectors specialized to
3303 element types other than T, or for non-simple multidimensional arrays,
3304 have not been addressed. They could almost certainly be handled the
3305 same way if anyone is motivated to do so.)
3306 * The changes in array handling break binary compatibility, so
3307 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3308 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3309 instead of (VALUES T T).
3310 * By following the instructions that Dan Barlow posted to sbcl-devel
3311 on 2 July 2000, I was able to enable primitive dynamic object
3312 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3313 functionality is not implemented (since it calls ld to resolve
3314 library references automatically, requiring RUN-PROGRAM for its
3315 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3316 try to resolve library references) is now supported.
3317 * The system now flushes the standard output streams when it terminates,
3318 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3319 them at several other probably-convenient times, e.g. in each pass of
3320 the toplevel read-eval-print loop, and after evaluating a form given
3321 as an "--eval" command-line option. (These changes were motivated by a
3322 discussion of stream flushing issues on cmucl-imp in August 2000.)
3323 * The source transform for TYPEP of array types no longer assumes
3324 that an array whose element type is a not-yet-defined type
3325 is implemented as an array of T, but instead punts, so that the
3326 type will be interpreted at runtime.
3327 * There is now some support for cross-compiling in make.sh: each of
3328 the phases of make.sh has its own script. (This should be transparent
3329 to people doing ordinary, non-cross-compile builds.)
3330 * Since my laptop doesn't have hundreds of megabytes of memory like
3331 my desktop machine, I became more motivated to do some items on
3332 my to-do list in order to reduce the size of the system a little:
3333 ** Arrange for various needed-only-at-cold-init things to be
3334 uninterned after cold init. To support this, those things have
3335 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3336 symbols with such names are now uninterned after cold init).
3337 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3338 fasl files for cold load.
3339 ** Remove the old compiler structure pooling code (which used to
3340 be conditional on the target feature :SB-ALLOC) completely.
3341 ** Redo the representation of some data in cold init to be more compact.
3342 (I also looked into supporting byte compiled code at bootstrap time,
3343 which would probably reduce the size of the system a lot, but that
3344 looked too complicated, so I punted for now.)
3345 * The maximum signal nesting depth in the src/runtime/ support code has
3346 been reduced from 4096 to 256. (I don't know any reason for the very
3347 large old value. If the new smaller value turns out to break something,
3348 I'll probably just bump it back up.)
3349 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3351 * Many, many bugs reported by Peter Van Eynde have been added to
3352 the BUGS list; some have even been fixed.
3353 * While enabling dynamic object file loading, I tried to make the
3354 code easier to understand, renaming various functions and variables
3355 with less ambiguous names, and changing some function calling
3356 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3357 * While trying to figure out how to do the OpenBSD port, I tried to
3358 clean up some of the code in src/runtime/. In particular, I dropped
3359 support for non-POSIX signal handling, added various comments,
3360 tweaked the code to reduce the number of compilation warnings, and
3361 renamed some files to increase consistency.
3362 * To support the new automatic configuration functionality in make.sh,
3363 the source file target-features.lisp-expr has been replaced with the
3364 source file base-target-features.lisp-expr and the machine-generated
3365 file local-target-features.lisp-expr.
3366 * fixed a stupid quoting error in make.sh so that using CMU CL
3367 "lisp -batch" as cross-compilation host works again
3369 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3371 * DESCRIBE no longer tries to call itself recursively to describe
3372 bound/fbound values, so that it no longer fails on symbols which are
3373 bound to themselves (like keywords, T, and NIL).
3374 * DESCRIBE now works on generic functions.
3375 * The printer now prints less-screwed-up representations of closures
3376 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3377 * A private symbol is used instead of the :EMPTY keyword previously
3378 used to mark empty slots in hash tables. Thus
3379 (DEFVAR *HT* (MAKE-HASH-TABLE))
3380 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3381 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3382 now does what ANSI says that it should. (You can still get
3383 similar noncompliant behavior if bang on the hash table
3384 implementation with all the symbols you get back from
3385 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3386 This breaks binary compatibility, since tests for equality to
3387 :EMPTY are wired into things like the macroexpansion of
3388 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3390 * There's now a minimal placeholder implementation for CL:STEP,
3391 as required by ANSI.
3392 * An obscure bug in the interaction of the normal compiler, the byte
3393 compiler, inlining, and structure predicates has been patched
3394 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3395 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3397 * Missing ordinary arguments in a macro call are now detected even
3398 when the macro lambda list contains &KEY or &REST.
3399 * The debugger no longer complains about encountering the top of the
3400 stack when you type "FRAME 0" to explicitly instruct it to go to
3401 the top of the stack. And it now prints the frame you request even
3402 if it's the current frame (instead of saying "You are here.").
3403 * As specified by ANSI, the system now always prints keywords
3404 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3406 * The default initial SIZE of HASH-TABLEs is now smaller.
3407 * Type information from CLOS class dispatch is now propagated
3408 into DEFMETHOD bodies, so that e.g.
3409 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3411 is now basically equivalent to
3412 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3413 (DECLARE (TYPE SINGLE-FLOAT X))
3415 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3416 operation, without having to do run-time type dispatch.
3417 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3418 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3419 * Since I don't seem to be making big file reorganizations very often
3420 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3421 is overflowing my ability to conveniently back them up), I've finally
3422 checked the system into CVS. (The CVS repository is on my home system,
3423 not at SourceForge -- putting it on SourceForge might come later.)
3424 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3425 high-level GC-NOTIFY-FOO functions send their output. (There's
3426 still very little control of where low-level verbose GC functions
3427 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3428 less than it used to -- the GC-NOTIFY-FOO functions are now under
3429 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3430 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3431 in only one place in the source code, and propagates it automatically
3432 everywhere that it's needed. Thus e.g. when I bump the version from
3433 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3434 * The C source files now include boilerplate legalese and documentation
3435 at the head of each file (just as the Lisp source files already did).
3436 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3437 to his page will be replaced with a link to his new CLiki service.
3439 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3441 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3442 * The build process now looks for GNU make under the default name "gmake",
3443 instead of "make" as it used to. If GNU make is not available as "gmake"
3444 on your system, you can change this default behavior by setting the
3445 GNUMAKE environment variable.
3446 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3447 can build without error under CMU CL.
3449 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3451 * There is now a partial SBCL user manual (with some new text and some
3452 text cribbed from the CMU CL manual).
3453 * The beginnings of a profiler have been added (starting with the
3454 CMU CL profiler and simplifying and cleaning up). Eventually the
3455 main interface should be through the TRACE macro, but for now,
3456 it's still accessed through vaguely CMU-CL-style functions and macros
3457 exported from the package SB-PROFILE.
3458 * Some problems left over from porting CMU CL to the new
3459 cross-compilation bootstrap process have been cleaned up:
3460 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3461 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3462 part when getting rid of the compiler *BACKEND* stuff.)
3463 ** The value of *NULL-TYPE* was screwed up, because it was
3464 being initialized before the type system knew the final
3465 definition of the 'NULL type. This screwed up several key
3466 optimizations in the compiler, causing inefficiency in all sorts
3467 of places. (I found it because I wanted to understand why
3468 GET-INTERNAL-RUN-TIME was consing.)
3469 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3470 PROCLAIM FTYPE information. Unfortunately this broke binary
3471 compatibility again, since now the forms output by DEFGENERIC
3472 to refer to functions which didn't exist in 0.6.3.
3473 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3474 can use the new (as of 0.6.3) transform for SOME into MAP into
3476 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3477 values to (INTEGER 0 1000000), so that the time code will no longer
3478 occasionally get blown up by Linux returning 1000000 microseconds
3479 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3480 its output conform to the ANSI spec. (Alas, this makes its output
3481 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3482 * A full call to MAP NIL with a single sequence argument no longer conses.
3483 * fixes to problems pointed out by Martin Atzmueller:
3484 * The manual page no longer talks about multiprocessing as though
3485 it were currently supported.
3486 * The ILISP support patches have been removed from the distribution,
3487 because as of version 5.10.1, ILISP now supports SBCL without us
3488 having to maintain patches.
3489 * added a modified version of Raymond Toy's recent CMU CL patch for
3490 EQUALP comparison of HASH-TABLE
3492 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3494 * The system still can't cross-compile itself with
3495 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3496 entails), but at least it can compile more of itself that way
3497 than it used to be able to, and various buglets which were uncovered
3498 by trying to cross-compile itself that way have now been fixed.
3499 * This release breaks binary compatibility again. This time
3500 at least I've incremented the FASL file format version to 2, so that the
3501 problem can be detected reliably instead of just causing weird errors.
3502 * various new style warnings:
3503 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3504 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3505 ** using the lexical binding of a variable named in the *FOO* style
3506 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3507 as specified by ANSI.
3508 * *RANDOM-STATE* is no longer automatically initialized from
3509 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3510 default behavior of the system is to repeat its behavior every time
3511 it's run. If you'd like to change this behavior, you can always
3512 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3513 old convention there was no comparably easy way to get the system to
3514 repeat its behavior every time it was run.
3515 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3516 FTYPE declarations has been removed, in favor of their ANSI
3517 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3518 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3519 the types of their sequence arguments can be determined at compile time.
3520 This is done through a new open code expansion for MAP which eliminates
3521 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3522 argument types can be determined at compile time.
3523 * The optimizer now transforms COERCE into an identity operation when it
3524 can prove that the coerced object is already of the correct type. (This
3525 can be a win for machine generated code, including the output of other
3526 optimization transforms, such as the MAP transform above.)
3527 * Credit information has been moved from source file headers into CREDITS.
3528 * Source file headers have been made more standard.
3529 * The CASE macro now compiles without complaining even when it has
3532 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3534 * (Note that the way that the PCL macroexpansions were rewritten
3535 to accommodate the change in DEFGENERIC below breaks binary
3536 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3537 not run under sbcl-0.6.2. Once we get out of alpha releases,
3538 i.e. hit release 1.0.0, we'll probably try to maintain binary
3539 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3540 and sbcl-1.4.4. Until then, however, it might be fairly common
3541 for maintenance releases to break binary compatibility.)
3542 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3544 * The reporting of unhandled signals has been changed to print some
3545 explanatory text as well as the report form. (Previously only
3546 the report form was printed.)
3547 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3548 it defines, so that the compiler no longer issues undefined function
3549 warnings for compiled-but-not-yet-loaded generic functions.
3550 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3551 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3552 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3553 section of the ANSI spec are supported.
3554 * The "" nickname for the "KEYWORD" package has been removed.
3555 The reader still handles symbol tokens which begin with a package marker
3556 as keywords, but it doesn't expose its mechanism for doing so in the
3557 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3558 * The system now issues STYLE-WARNINGs for contradictory TYPE
3559 proclamations. (Warnings for contradictory FTYPE proclamations would
3560 be nice too, but those can't be done usefully unless the type system
3561 is made smarter about FUNCTION types.)
3562 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3563 systematized, so that "*target-*.lisp is supposed to exist only on the
3564 target and imply that there's a related file which exists on the
3565 host, and *host-*.lisp is supposed to exist only on the host and imply
3566 that there's a related file which exists on the target. This involves a
3567 lot of renaming. Hopefully the acute confusion caused by the renaming
3568 will be justified by the reduction in chronic confusion..
3569 ** runtime-type.lisp -> early-target-type.lisp
3570 ** target-type.lisp -> late-target-type.lisp
3571 ** early-host-format.lisp -> early-format.lisp
3572 ** late-host-format.lisp -> late-format.lisp
3573 ** host-error.lisp -> misc-error.lisp
3574 ** early-error.lisp -> early-target-error.lisp
3575 ** late-error.lisp -> late-target-error.lisp
3576 ** host-defboot.lisp -> early-defboot.lisp
3577 ** code/misc.lisp -> code/target-misc.lisp
3578 ** code/host-misc.lisp -> code/misc.lisp
3579 ** code/numbers.lisp -> code/target-numbers.lisp
3580 ** code/early-numbers.lisp -> numbers.lisp
3581 ** early-host-type.lisp -> early-type.lisp
3582 ** late-host-type.lisp -> late-type.lisp
3583 ** host-typep.lisp -> typep.lisp
3584 ** load.lisp -> target-load.lisp
3585 ** host-load.lisp -> load.lisp
3586 ** host-disassem.lisp -> disassem.lisp
3587 ** host-insts.lisp -> insts.lisp
3588 ** byte-comp.lisp -> target-byte-comp.lisp
3589 ** host-byte-comp.lisp -> byte-comp.lisp
3590 ** host-signal.lisp -> signal.lisp
3591 ** host-defstruct.lisp -> defstruct.lisp
3592 ** late-target-type.lisp -> deftypes-for-target.lisp
3593 Furthermore, several other previously target-only files foo.lisp (e.g.
3594 hash-table.lisp and random.lisp) have been split into a target-and-host
3595 foo.lisp file and a target-only target-foo.lisp file, with their key type
3596 definitions in the target-and-host part, so that the cross-compiler will
3597 know more about target types.
3598 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3599 gone away. In their place are various *BACKEND-FOO* variables
3600 corresponding to the slots of the old structure.
3601 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3602 into the target SBCL has been fixed.
3603 * The chill.lisp system for loading cold code into a running SBCL
3605 * Support for the CMU CL "scavenger hook" extension has been removed.
3606 (It was undocumented and unused in the CMU CL sources that SBCL was
3607 derived from, and stale in sbcl-0.6.1.)
3608 * Various errors in the cross-compiler type system were detected
3609 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3610 (enabling various consistency checks). Many of them were fixed,
3611 but some hard problems remain, so the compiler is back to
3612 running without *TYPE-SYSTEM-INITIALIZED* for now.
3613 * As part of the cross-compiler type system cleanup, I implemented
3614 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3615 * I have started adding UNCROSS calls throughout the type system
3616 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3617 on cross-compiler input files will be able to go away, and various
3619 * CONSTANTP now returns true for quoted forms (as explicitly required
3622 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3624 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3625 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3626 (MAKE-LIST -1). In the longer term, it ought to become true
3627 that declarations are assertions even at SAFETY 1. For now, it's not
3628 quite true even at SAFETY 3, but it's at least more nearly true..
3629 (Note that this change seems to increases the size of the system by
3630 O(5%) and to decrease the speed of the compiler by 20% or more.)
3631 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3632 for the problem of printing dozens of lines of distracting information
3633 about low-level system machinery as part of the top stack frame
3634 on entry to the debugger when an undefined function was called.
3635 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3636 is not reset to COMMON-LISP-USER.
3637 * Compilation of stuff related to dyncount.lisp has been made conditional
3638 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3639 smaller. The various dyncount-related symbols have been moved into
3640 a new "SB-DYNCOUNT" package.
3641 * tty-inspect.lisp has been renamed to inspect.lisp.
3642 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3643 feature has gone away. (When we eventually port to other flavors of
3644 libc and/or Unix, we'll try to make the differences between flavors
3645 invisible at the user level.)
3646 * Various other *FEATURES* tags, and/or their associated conditionals,
3647 have been removed if obsolescent, or given better documentation, or
3648 sometimes given more-mnemonic names.
3650 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3652 * tidied up "make.sh" script
3653 * tidied up system directory structure
3654 * better "clean.sh" behavior
3655 * added doc/FOR-CMUCL-DEVELOPERS
3656 * many many small tweaks to output format, e.g. removing possibly-confusing
3657 trailing #\. character in DESCRIBE-INSTANCE
3658 * (EQUALP #\A 'A) no longer signals an error.
3659 * new hashing code, including EQUALP hashing
3660 * tidied up Lisp initialization and toplevel
3661 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3662 * command line argument processing
3663 * added POSIX-GETENV function to deal with Unix-ish environment variables
3664 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3665 terminating SBCL on EOF
3666 * non-verbose GC by default
3667 * There is no more "sbcl" shell script; the sbcl file is now the C
3668 runtime executable (just like CMU CL).
3669 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3671 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3672 symbol and package information as Lisp native symbol and package objects
3673 instead of strings naming symbols and strings naming packages. This way,
3674 whenever packages are renamed (as in warm init), debug information is
3675 transformed along with everything else.
3676 * tweaked the optimization policy declarations which control the building
3677 of SBCL itself. Now, among other things, the system no longer saves
3678 source location debugging information. (This helps two problems at once
3679 by reducing SBCL size and by keeping SBCL from trying to look for its
3680 sources -- which may not exist -- when reporting errors.)
3681 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3682 debugging and testing purposes
3683 * cleaned up printing, making the printer call PRINT-OBJECT for
3684 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3685 methods, giving nearly-ANSI behavior
3686 * converted almost all special variables to use *FOO* naming convention
3687 * deleted PARSE-TIME functionality, since it can be done portably
3688 * moved some files out of cold init into warm init
3689 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3691 * regularized formatting of source files
3692 * added an install.sh script
3693 * fixed ridiculous memory usage of cross-compiler by making
3694 compiler/alloc.lisp not try to do pooling unless it can hook
3695 itself into the GC of the cross-compilation host. Now the system
3696 builds nicely on my old laptop.
3697 * added :SB-ALLOC in target-features.lisp-expr
3698 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3699 was not implemented)
3700 * re-did condition handling and note reporting in the compiler. Notes
3701 are no longer handled by signalling conditions. Style warnings
3702 and warnings are handled more correctly and reported in such a way
3703 that it's easy to find one or the other in your output (so that you
3704 can e.g. figure out which of many problems caused COMPILE-FILE to
3706 * changed the severity of several compiler warnings from full WARNING
3707 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3708 compiler note reporting so that it doesn't use the condition system
3709 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3710 COMPILE-FILE command)
3711 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3712 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3713 no longer supported, since their implementation was deeply intertwingled
3714 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3715 * removed bogus "support" for compiler macros named (SETF FOO), and
3716 removed the compiler macro for SETF INFO (but only after making a fool
3717 of myself on the cmucl-imp mailing list by posting a bogus patch for
3718 DEFINE-COMPILER-MACRO..)
3719 * Compiled files containing forms which have side effects on the Lisp
3720 reader (such as DEFPACKAGE forms) are now handled more correctly.
3721 (Compiler queuing of top level lambdas has been suppressed by setting
3722 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3723 * deleted various currently-unused source files, e.g. gengc.lisp. They
3724 may be added back at some point e.g. when porting to other architectures,
3725 but until they are it's distracting to distribute them and to try to
3727 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3728 there were so many of them they're just distractions, and UNCROSS is
3729 known to be able to handle the current sources
3730 * moved PROFILE functionality into TRACE, so that it will be clear
3731 how the wrapping and unwrapping of functions when you profile them
3732 interacts with the wrapping and unwrapping of functions when you
3733 trace them. (Actually, the functionality isn't there yet, but at least
3734 the interface specification is there. Hopefully, the functionality will
3735 arrive with some future maintenance release.)
3736 * removed host-oops.lisp
3737 * changed signature of QUIT function to allow UNIX-CODE argument
3738 * fixed READ-SEQUENCE bug
3739 * tweaked verbose GC output so that it looks more like the progress
3740 output that ANSI specifies for functions like LOAD
3741 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3742 * added <http://sbcl.sourceforge.com> to the banner information printed by