1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
3 * feature: thread support on Solaris/x86, and experimental thread support
5 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
7 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
8 default method for SLOT-UNBOUND.
9 * minor incompatible change: prevent the user from specializing the
10 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
11 somewhat counter to the intent of the protocol, I (CSR) think, and
12 additionally it just doesn't work in SBCL as currently
13 implemented, thanks to optimizations (that are always valid for
14 the other three SLOT-VALUEish functions, but not for the setter).
15 * minor incompatibale change: the :SB-LDB feature is now enabled by
16 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
17 the low-level debugger.
18 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
19 with a type error. (reported by blitz_ on #lisp)
20 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
21 failed with a type error. (reported by Pascal Bourguignon)
22 * bug fix: merging pathnames against defaults with :DIRECTORY
23 starting with '(:RELATIVE :BACK) should preserve the :BACK.
24 (reported by James Y Knight)
25 * improved SB-BSD-SOCKETS support on Windows. (thanks to Timothy
27 * bug fix: saving large (>2GB) cores on x86-64 now works
28 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
29 constant index and a (simple-array (signed-byte 32)) array
30 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
31 error (patch by Robert J. Macomber)
32 * feature: TIME now displays the amount of run-time spent in GC
33 * fixed some bugs revealed by Paul Dietz' test suite:
34 ** MISC.641: LET-conversion were not supposed to work in late
37 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
38 * new feature: source path information is generated for macro-expansion
39 errors for use in IDE's like Slime (thanks to Helmut Eller)
40 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
41 * bug fix: compiler does not loop forever on an invalid type in
43 * improvement: compilation of most CLOS applications is significantly
45 * optimization: added a limited bytecode compiler for simple toplevel
46 forms, speeding up compilation and FASL loading
48 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
49 * minor incompatible change: in sbcl-0.9.11 (but not earlier
50 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
51 into subclasses of the finalized class. Now user calls to
52 FINALIZE-INHERITANCE finalize just the one class, and calls by the
53 system return before any subclasses are finalized.
54 * minor incompatible change: The reader no longer ignores errors
55 regarding non-existent packages in #+ and #- feature tests.
56 * new feature: command line options --no-sysinit, --no-userinit to
57 inhibit loading the corresponding init files
58 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
59 for sending data through UDP sockets (thanks to François-René Rideau)
60 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
61 error is signalled instead (thanks to François-René Rideau)
62 * minor incompatible change: Error signalling behaviour of lexical
63 operations violating package locks has changed slightly. Refer to
64 documentation on package locks for details.
65 * enhancement: EVAL can process IF-expressions without resorting to the
67 * bug fix: Some operations involving SIN, COS, or TAN weren't being
68 constant-folded properly due to missing out-of-line functions.
69 (reported by Mika Pihlajamäki)
70 * bug fix: LISTEN sometimes returned T even in cases where no data was
71 immediately available from the stream
72 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
73 were reversed. (reported by Levente Mészáros)
74 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
75 when given a vector displaced to an adjustable array. (reported
77 * bug fix: derivation of float boundaries from numbers outside the
78 appropriate float range (reported by John Wiseman)
79 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
81 * bug fix: precomputing cache entries for generic functions with
82 some subclasses of specializers as yet invalid does not attempt to
83 fill a cache line with a negative offset. (reported by Levente
85 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
86 allow condition class objects as arguments to DOCUMENTATION and
87 (SETF DOCUMENTATION); only find and set documentation for
88 structure names for the STRUCTURE doc-type. (suggested by Gary
90 * improvements to the Win32/x86 port:
91 ** (user-homedir-pathname) and default initialization file
92 locations now know about the user's "Documents and Settings"
93 directory (thanks to Yaroslav Kavenchuk)
94 ** run-program is implemented (thanks to Mike Thomas)
95 ** sockets support (thanks to Timothy Ritchey)
96 ** better backtrace support (thanks to Alastair Bridgewater)
97 ** sb-grovel supported
98 ** asdf-install and sb-posix work somewhat
99 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
100 * improvements to the Solaris/x86 port:
101 ** works on Solaris 11/Solaris Express
102 ** floating-point exception handling support
103 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
104 * fixed some bugs revealed by Paul Dietz' test suite:
105 ** REMOVE-METHOD returns its generic function argument even when
106 the method is not one of the generic functions' methods.
107 ** objects declared as MEMBER types can be admissible arguments to
109 ** printing characters should simply be printed by the FORMAT ~:C
111 ** compiler failure when compiling functions with hairy constant
112 defaults for optional parameters.
113 ** compiler produces wrong code when MAYBE-INLINE-expanding a
114 function, which is already optimized.
116 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
117 * new platform: experimental support for SBCL x86/Darwin, including
118 MacOS X 10.4.5 on Intel.
119 * new feature: Unicode character names are now known to the system
120 (through CHAR-NAME and NAME-CHAR).
121 * new feature: the filesystem location of SBCL's core file is
122 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
123 * minor incompatible change: the contrib modules SB-POSIX and
124 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
125 this change is to make it easier to distribute
126 location-independent binaries.
127 * bug fix: as implied by AMOP, standardized classes no longer have
128 slots named by external symbols of public packages. (reported by
130 * optimization: calling functions via constant symbols -- as in
131 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
132 via the function object as in (FUNCALL #'FOO).
133 * optimization: on x86 and x86-64, the calling convention for the
134 case of dealing with an unknown number of values has been altered
135 to be friendlier to the prediction heuristics implemented,
136 particularly with reference to CALL and RET pairing. (thanks to
137 Alastair Bridgewater)
138 * optimization: on x86, the code for access to thread-local symbol
139 values and binding/unbinding of thread-local symbols is smaller.
140 (thanks to Alastair Bridgewater)
141 * enhancement: CONSTANTP is now able to determine constantness of
142 more complex forms, including calls to constant-foldable standardized
143 functions and some special forms beyond QUOTE.
144 * fixed bug: occasional GC crashes on Solaris/x86
145 * optimization: x86-64 supports stack allocation of results of simple
146 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
147 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
148 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
149 (thanks to James Knight)
150 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
151 by the standard (thanks to Stephen Compall)
153 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
154 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
155 be used for bundling the runtime and the core file into one
156 executable binary. This feature is not currently supported on all SBCL
157 platforms. (thanks to James Bielman and NIIMI Satoshi)
158 * new feature: a generational or ephemeral garbage collector is now
159 the default on the PowerPC platform (both Linux and Darwin). The
160 old Cheney (stop and copy) collector is a build-time option.
161 (thanks to Cyrus Harmon, after Raymond Toy)
162 * minor incompatible change: the method by which SBCL finds its
163 contributed modules has changed; it no longer relies on symbolic
164 links from an $SBCL_HOME/systems directory, but searches directly
165 in the subdirectories of $SBCL_HOME.
166 * enhancement: the dynamic heap size on the Linux/PPC platform is
167 markedly larger, even using the older Cheney garbage collector.
168 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
169 certain complicated string types. (reported by Gary King)
170 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
171 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
172 * fixed bug: DOCUMENTATION on structures no longer signals an
173 error if no structure type of the right name exists. (reported by
175 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
176 applicable non-standard primary or :AROUND method on
177 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
178 no longer cause unbound variable errors. (reported by Kevin Reid)
179 * fixed bug: metacircle resolution in cases where methods have slots
180 added before the slots from STANDARD-METHOD. (reported by Jean
182 * fixed bug: the Power PC assembly code for calling into lisp
183 sometimes computed its return address wrongly (depending on the
184 15th bit of the address, fixed at link time). This used to
185 manifest itself in a segmentation violation while building PCL.
186 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
187 many others over the years)
188 * fixed bug: ROOM no longer reports silly numbers for stack usage if
189 the stack pointer (treated as a signed integer) is negative.
190 (thanks to Peter van Eynde)
192 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
193 * new platform: experimental support for the Windows operating
194 system has been added. (thanks to Alastair Bridgewater)
195 * fixed several bugs in and robustified the PPC FFI (including
196 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
197 * bug fix: allow non-simple string symbol names (reported by Paul
199 * bug fix: interrupt handling on NetBSD (thanks to Richard M
201 * bug fix: saving a core corrupted callbacks on x86/x86-64
202 * bug fix: closed a loophole in metacircularity detection and
203 grounding in the PCL implementation of CLOS.
204 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
205 (reported by tomppa on #lisp)
206 * bug fix: clear the direction flag on Lisp -> C transitions, as
207 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
208 (reported by Andrej Grozin and Hendrik Maryns)
209 * optimization: major improvements to GC efficiency on GENCGC platforms
210 * optimization: faster implementation of EQUAL
211 * optimization: emit more efficient opcodes for some common
212 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
213 * optimization: several other minor code-generation improvements on x86-64
214 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
216 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
217 * minor incompatible change: (SETF CLASS-NAME) and (SETF
218 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
219 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
220 by ANSI as a generic function, and for consistency (SETF
221 GENERIC-FUNCTION-NAME) is treated likewise.
222 * fixed bug #233.b: make constraint propagation notice when a variable
223 value is changed after it is referenced but before it is used
224 * fixed bug #296: no more arbitrary behaviour from filenames with
225 odd characters as --load arguments. (reported by Adam Warner)
226 * fixed bug #390: :CHARACTER-SET pathname components now work as
227 expected. (reported by Tim Daly Jr)
228 * fixed bug #391: complicated :TYPE intersections in slot
229 definitions no longer cause an error in PCL internals.
230 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
231 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
232 returning the number of octets which would be written to the
233 file-stream. (thanks to Robert J. Macomber)
234 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
235 arguments to the internal stream functions. (thanks to David
237 * bug fix: stack exhaustion is now handled on the x86/SunOS
238 platform. (thanks to Daisuke Homma)
239 * bug fix: more accurate ROOM results on GENCGC platforms
240 * optimization: improved type inference for arithmetic-for
241 index variables in LOOP
242 * optimization: faster floating-point SQRT on x86-64
243 * fixed some bugs revealed by Paul Dietz' test suite:
244 ** DOCUMENTATION returns NIL instead of "" for method combinations
245 that don't have a docstring
247 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
248 * minor incompatible change: (SETF CLASS-NAME) and (SETF
249 GENERIC-FUNCTION-NAME) are no longer generic functions, and
250 therefore are not customizeable by user code (as seems to be at
251 least permitted and maybe required by AMOP). As a consolation,
252 however, the SBCL implementation of these functions now calls
253 REINITIALIZE-INSTANCE as specified by AMOP.
254 * bug fix: slot-definition documentation is propagated as per ANSI
255 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
256 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
257 argument, as do eslotd initialization methods. (from Pascal
258 Costanza's "Closer" project)
259 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
260 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
262 * bug fix: it is now possible to have more than one subclass of
263 STANDARD-GENERIC-FUNCTION without causing stack overflow.
264 (reported by Bruno Haible, Pascal Costanza and others)
265 * bug fix: the dependent update protocol now works for generic
266 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
268 * bug fix: condition-class instances corresponding to
269 DEFINE-CONDITION forms are now created eagerly. (reported by
270 Kalle Olavi Niemitalo on comp.lang.lisp)
271 * bug fix: floating point printing is more accurate in some
272 circumstances. (thanks to Simon Alexander)
273 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
274 merged with *DEFAULT-PATHNAME-DEFAULTS*.
275 * bug fix: callbacks on OS X now preserve stack-alignment.
276 (thanks to Cyrus Harmon)
277 * enhancement: the x86-64 disassembler is much better at
278 disassembling SSE instructions. (thanks to Lutz Euler)
279 * enhancement: CHECK-TYPE on a variable now allows type-inference
280 (thanks to Vincent Arkesteijn)
281 * optimization: improved performance of EUC-JP external format.
282 (thanks to NIIMI Satoshi)
283 * optimization: performance improvements to IO on file streams of
284 :ELEMENT-TYPE CHARACTER
285 * optimization: much faster memory allocation on x86-64
286 * optimization: faster garbage collections (latency and throughput)
288 * optimization: faster bignum subtraction and division on x86 and x86-64
289 (thanks to Lutz Euler)
290 * optimization: more accurate type-inference for loops that use a
291 floating point index variable or a negative step.
293 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
294 * bug fix: add a workaround to SBCL looping infinitely at startup on
295 Linux kernels with apparently buggy implementations of personality().
296 (thanks to Svein Ove Aas)
297 * bug fix: Unicode symbols are correctly printed in LDB backtraces
298 (thanks to David Lichteblau)
299 * bug fix: local bindings shadow global symbol macros for
300 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
301 * optimization: non-open coded uses of numeric comparison operators
302 (e.g. >) no longer cons when called with more than one parameter
303 on platforms supporting dynamic-extent allocation.
304 * enhancement: saving cores with foreign code loaded is now supported
305 on MIPS/Linux in addition to the previously supported platforms.
306 * bug fix: FIND-RESTART now tests for activity, not applicability when given
307 a restart object as identifier. (reported by Helmut Eller for CMUCL)
308 * bug fix: division by zero in sb-sprof when no samples were collected
309 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
310 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
311 (thanks to Alexander Kjeldaas)
312 * new feature: ignoring the return values of destructive functions
313 like NREVERSE signals a compile-time style-warning.
314 (thanks to Kevin Reid)
315 * new feature: the alignment of alien structure fields can be
316 explicitly specified. (thanks to Cyrus Harmon)
317 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
318 and dump core on SIGQUIT
320 ** incompatible change: threads do not inherit values of specials
321 from their parents (see manual)
322 ** bug fix: threads stacks belonging to dead threads are freed by the
323 next exiting thread, no need to gc to collect thread stacks anymore
324 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
325 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
326 ** bug fix: dynamic variable and thread start related gc lossage
327 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
329 * fixed some bugs revealed by Paul Dietz' test suite:
330 ** SUBTYPEP is slightly more accurate on heinously complicated
331 CONS types where some of the members have uncertain (in the
332 NIL, NIL sense) type relationships to each other.
334 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
335 no longer purified unless :PURIFY T is explicitly specified.
336 ** Non-purified cores are significantly smaller than before
338 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
339 * new feature: timers based on Zach Beane's excellent timer package
340 * added support for the following external formats: koi8-u,
341 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
342 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
343 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
344 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
345 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
346 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
347 * incompatible change: a threaded SBCL will no longer revert to
348 non-threaded mode on non-NPTL systems, but refuse to start entirely.
349 * bug fix: interrupts are disabled until startup is complete; no
350 more sigsegvs when receiving a signal to soon
351 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
353 * bug fix: add a workaround for the memory randomization features in
354 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
355 workaround will only be in effect on systems with the proc filesystem
357 * bug fix: printing objects of type HASH-TABLE signals a
358 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
360 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
361 have been running for over 50 days. (reported by Gilbert Baumann)
362 * bug fix: the logic for getting names of functions gets less
363 confused when confronded with alternate-metaclass
364 funcallable-instances. (reported by Cyrus Harmon)
365 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
366 consistent, even on internal alternate-metaclass objects.
367 * bug fix: flush closure information collected by physical
368 environment analysis prepass before the main pass. (bug reported
370 * bug fix: compiler pack phase does not modify a hashtable, which is
371 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
372 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
373 may be expanded. (reported by Alexander <malishev> on c.l.l)
374 * bug fix: no more style-warnings when compiling code using LAST
375 under high-SPEED low-SPACE settings. (reported by David Wallin
377 * fixed some incompatibilities between SBCL's MOP and the MOP
379 ** the METAOBJECT class is now implemented;
380 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
381 STANDARD-OBJECT, as required;
382 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
383 now compatible (as required by VALIDATE-SUPERCLASS); there
384 remains a consistency requirement over the presence or absence
385 of the FUNCTION class in the superclass list of finalized
386 classes; see the manual for more details;
387 ** the :AROUND method for COMPUTE-SLOTS on
388 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
389 requested slot ordering.
391 ** bug fix: parent thread now can be gc'ed even with a live
393 ** bug fix: binding a special with PROGV to no value is not
394 the same as that symbol not having been bound (thanks to
396 ** bug fix: binding specials is thread safe (thanks to
398 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
399 and SB-SPROF do not die with 'no handler for signal XX in
400 interrupt_handle_now(..)' anymore
401 ** bug fix: WITH-TIMEOUT works with multiple running threads
402 * fixed some bugs revealed by Paul Dietz' test suite:
403 ** ENSURE-GENERIC-FUNCTION should take a method class object for
404 the :method-class keyword argument.
406 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
407 * new port: the Solaris operating system on x86 processors is now
408 mostly supported, though some rough edges in the environment
409 remain. (thanks to Daisuke Homma)
410 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
411 space for its heap. (thanks to Thiemo Seufer)
412 * optimization: SBCL on MIPS platforms now supports dynamic-extent
413 closures. (thanks to Thiemo Seufer)
414 * minor incompatible change: eof selects abort in the debugger.
415 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
416 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
418 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
419 deprecated, and will go away in a future revision of SBCL.
420 * minor incompatible change: GC-ON and GC-OFF are no longer
421 implemented with a counter, it does not matter how many times gc
422 is switched on or off
423 * bug fix: discriminating functions for generic function classes
424 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
425 make invalid assumptions about method precedence order. (reported
427 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
428 platforms. (thanks to Thiemo Seufer)
429 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
430 * bug fix: correct pathname printing with printer escaping is on.
431 (thanks to Kevin Reid)
432 * bug fix: complex VOP definitions in "user-space" no longer trigger
433 package locks. (reported by Zach Beane)
434 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
435 for funcallable-instance functions, and is no different from
437 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
438 characters in the current readtable. (reported by Nicholas Neuss)
439 * bug fix: SUBTYPEP on various CONS types returns more a more
440 accurate acknowledgment of its certainty.
441 * optimizations: REMOVE-DUPLICATES now runs in linear time on
442 lists in some cases. This partially fixes bug 384.
443 * flush all standard streams before prompting in the REPL and the
445 * bug fix: signal handling and triggering gc do not conflict
446 directly anymore, in particular a high frequency sb-sprof does
447 not prevent gc from running
448 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
449 approximation for timezone and DST information between the
450 universal time epoch and the smallest negative 32-bit time_t.
451 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
452 year 1899 so long as the encoded time is non-negative.
453 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
454 the correct answer for non-constant rotate amounts when performing
455 an inline 32-bit rotation.
457 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
458 there is only one thread in the session
459 ** bug fix: memory leak for streams created in one thread and
460 written to in another
461 ** bug fix: lockup when compiled with gcc4
462 ** bug fix: race that allows the gc to be triggered when gc is
464 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
465 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
467 ** bug fix: finished threads are gc'ed properly
468 * fixed some bugs revealed by Paul Dietz' test suite:
469 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
470 has arguments with a different set of applicable methods from
471 the orignal arguments.
472 ** The type error thrown by MAP now has a correct expected-type
474 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
475 name a compiled function.
476 ** Three MISC tests where a large bignum was improperly coerced to
477 a float that couldn't represent that bignum during type
478 derivation were fixed.
479 ** SUBTYPEP can now handle types involving the negation of a
480 list-form FUNCTION type.
481 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
482 as (COMPLEX (AND RATIO (NOT FIXNUM))).
483 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
485 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
486 * New feature: Experimental support for bivalent streams: streams
487 opened with :element-type :default now allow character and binary
488 (unsigned-byte 8) I/O
489 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
490 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
492 * bug fix: no more highly sporadic "couldn't check whether ~S is
493 readable" when reading a stream and an interrupt hits in the middle
494 of a select system call
495 * compiler better recognizes complex arrays (reported by Eduardo
497 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
498 for variables on SBCL built with linkage-tables. (reported by Luis
500 * various error reporting improvements.
501 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
502 (Thanks to Thiemo Seufer)
503 * bug fix: nested reader invokations work correctly
504 * bug fix: it is possible to have simultaneous references to foreign
505 code and foreign data with the same name.
507 ** added x86-64 support
508 ** incompatible change: the threading api now works with thread
509 objects instead of thread ids
510 ** bug fix: threads are protected from signals and interruption when
511 starting up or going down
512 ** bug fix: a race where an exiting thread could lose its stack to gc
513 ** bug fix: don't halt on infinite error in threads if possible
514 ** fixed numerous gc deadlocks introduced in the pthread merge
515 ** bug fix: fixed thread safety issues in read and print
516 ** bug fix: debugger doesn't hang on session lock if interrupted at
517 an inappropriate moment
518 ** bug fix: run-program is now thread safe(r)
519 ** bug fix: inner with-recursive-lock no longer releases the mutex
520 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
521 * fixed some bugs revealed by Paul Dietz' test suite:
522 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
523 EXPECTED-TYPE slots filled.
524 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
526 ** MAKE-CONDITION accepts classes as type-designators.
527 ** COMPILE may never return NIL.
528 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
529 range before calling Unix time functions
531 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
532 * numerous signal handling fixes to increase stability
533 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
534 * minor incompatible change: we now correctly canonize default
535 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
536 as per the MOP, rather than the historical (INITARG INITFUNCTION
537 INITFORM). (reported by Bruno Haible)
538 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
539 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
540 for more information.
541 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
542 pathname is a directory pathname.
543 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
544 * dynamic space size on PPC has been increased to 768Mb. (thanks to
546 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
547 :METACLASS argument in addition to a class name. (reported by
548 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
549 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
550 (reported by Bruno Haible; patch from Gerd Moellmann)
551 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
553 * bug fix: more cleanups to the floating point exception handling on
554 x86-64 (thanks to James Knight)
555 * bug fix: the compiler does not try to constant fold calls of
556 COERCE and BIT-* functions when they return freshly constructed
557 objects. (reported by Nikodemus Siivola and Paul Dietz)
558 * optimization: improved the MIPS versions of generic (in the
559 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
560 * optimization: direct conversion of (unsigned-byte 32) to floats on
561 the PowerPC platform.
562 * optimization: structure instances with raw slots now use less
563 memory, and probably show better memory locality. (thanks to
565 * optimization: DYNAMIC-EXTENT declarations for lists and closures
566 are treated as requests for stack allocation on the x86-64,
567 Alpha-32, PPC and SPARC platforms.
568 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
569 the wrong file descriptor; implementation of SOCKET-OPEN-P.
570 (thanks to Tony Martinez)
572 ** gcing a dead thread can no longer lead to lockups
573 ** threads block signals until they are set up properly
574 ** errno is no longer shared by threads
575 ** interrupt-thread restores the eflags register on x86
576 ** fixed some lockups due to gc/thread interaction
577 * fixed some bugs revealed by Paul Dietz' test suite:
578 ** invalid dotted lists no longer raise a read error when
580 ** signal an error if a symbol that names a declaration is used
581 as the name of a type, or vice versa
582 ** allow using the (declare (typespec var*)) abbreviation for
583 (declare (type typespec var*)) with all type specifiers
584 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
585 ** always return NIL from PPRINT-POP when OBJECT is NIL
586 ** don't signal errors when pretty-printing malformed LABELS,
587 FLET or MACROLET forms
588 ** declarations in a DOLIST body are also in scope for the
590 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
592 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
595 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
596 * fixed cross-compiler leakages that prevented building a 32-bit
597 target with a 64-bit host compiler.
598 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
599 opened with :IF-EXISTS :OVERWRITE.
600 * fixed bug 281: error for an invalid qualifier in a short-form method
601 combination method is not signalled until the faulty method is called.
602 * bug fix: iteration variable type inferrer failed to deal with open
603 intervals. (reported by Alan Shields)
604 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
605 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
606 or more runtime options were provided to the sbcl binary.
607 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
608 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
609 * optimization: there's now a fast-path for fixnum arguments in the
610 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
611 * optimization: the code generated on x86-64 is more compact thanks
612 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
613 * bug fix: floating-point exception handling now partly works on
614 x86-64. (Thanks to James Knight)
615 * improvement to the MIPS backend from Thiemo Seufer: C-style
616 64-bit long long arguments and return values to alien functions
618 * fixed some bugs revealed by Paul Dietz' test suite:
619 ** the type-error signalled from WARN has a filled-in DATUM slot.
620 ** the type-error required when a stream is not associated with
621 a file has the stream as its datum.
622 ** type-errors on single-floats on x86-64 no longer have
623 :INVALID-OBJECT as the datum
624 ** the type-errors signalled for invalid function names now have
625 a correct expected type
626 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
627 for structures defined with a :TYPE. Documentation strings for
628 typed structures are no longer immediately discarded
629 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
630 broadcast streams and synonym streams. FILE-LENGTH now also works
631 on broadcast streams.
633 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
634 * incompatible change: the --noprogrammer option, deprecated since
635 version 0.7.5, has been removed. Please use the equivalent
636 --disable-debugger option instead.
637 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
639 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
640 inoperational for a while now) has been completely removed.
641 * null lexical environments are now printed as #<NULL-LEXENV>,
642 significantly reducing the amount of clutter in typical backtraces.
643 * documentation on weak pointers, finalization, and after GC hooks
644 has been added to the manual.
645 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
646 as other specialized array types, is much faster. SUBSEQ and
647 COPY-SEQ on such arrays have also been sped up.
648 * optimization: EQL is now more efficient when at least other argument
649 is known to be of type (OR FIXNUM (NOT NUMBER)).
650 * fixed bug: the runtime is now less vulnerable to changes in the
651 size of the SBCL object on OS X, and virtual memory is reserved for
652 all spaces that need to be at a fixed address.
653 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
654 * fixed bug: finalizers and after GC hooks that cause consing are now
656 * fixed bug: compiler error messages and summaries are now printed to
657 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
658 * fixed inference of the upper bound of an iteration variable.
659 (reported by Rajat Datta).
660 * fixed bug 211e: calling local functions with duplicated constant
661 keyword argument no longer causes a bogus style warning about an
663 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
664 ftype declaration to be disregarded. (reported by Dave Roberts)
665 * fixed bug 373: caused by erronous compilation of references to alien
666 variables in the runtime on ppc/darwin.
667 * fixed bug 376: CONJUGATE type deriver.
668 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
669 (reported by Baughn on #lisp)
670 * a cross-compiler bug on non-x86 platforms has been identified and
671 fixed. (thanks to Bruno Haible)
672 * improvements to the MIPS runtime code for increased stability.
673 (thanks to Thiemo Seufer)
674 * increased the maximimum compact environment size to allow
675 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
676 * partial workaround for bug 135: don't name the function we're
677 calling for hairy cases of EVAL, so as not to accumulate one environment
678 entry per call to EVAL. (reported by Kevin Reid)
679 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
680 * optimization: type testing for non-vector arrays should be faster.
681 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
683 * bug fix: setting 31st element of a bit vector to zero did not work
685 * bug fix: redefining a class definition which failed due to a
686 previous accessor / function clash now works (but see BUGS entry
687 #380 for more problems in this area). (thanks to Zach Beane)
688 * the long form of DEFINE-METHOD-COMBINATION disables method group
689 checking when given a single method group with pattern *. (thanks
691 * on x86 compiler supports stack allocation of results of simple
692 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
693 * support for building and running on Mac OS X 10.4 "Tiger" has been added
694 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
695 * fixed some bugs related to Unicode integration:
696 ** the restarts for recovering from input and output encoding
697 errors only appear when there is in fact such an error to
699 * fixed some bugs revealed by Paul Dietz' test suite:
700 ** MISC.549 and similar: late transformation of unsafe type
701 assertions into derived types caused unexpected code
703 ** SCALE-FLOAT type deriver is less wrong.
704 ** type derivers for EXP, LOG and similar functions compute result
705 types for complex arguments better.
706 ** (MISC.563) CONJUGATE type deriver works for very restricted
708 ** out-of-line type testers for character strings are available.
709 ** EQUAL compiler transform understands specialness of objects
711 ** accessing double-floats stored on the stack now works on x86-64.
712 ** debugger internals could sometimes create invalid lispobjs,
713 resulting in GC crashes.
714 ** MISC.548: type check weakening can convert required type into
716 ** initialization forms for bindings are not in scope of free special
719 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
720 * incompatible change: thread support for non-NPTL systems has
721 been removed - locking is buggy and unreliable. A threaded
722 SBCL build will now warn at startup and refuse to create
723 new threads, unless futex support is detected in the kernel
724 * incompatible change: the top level REPL now has only an ABORT
725 restart associated with it, not TOPLEVEL and ABORT as it used to.
726 TOP and TOPLEVEL are now available as debugger commands for
727 returning to the top level.
728 * incompatible change: forms evaluated in the REPL now use the
729 global optimization policy.
730 * incompatible change: user- and system-initialization files are
731 no longer processed with LOAD, but by READ and EVAL; hence the
732 global optimization policy, startup package, readtable, etc,
734 * The .fasl file format number has been incremented because of
735 various incompatible changes.
736 * internal entry point details and argument counts no longer appear
737 in backtraces unless explicitly requested by setting
738 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
739 * built-in and standard functions no longer have names like "top
740 level local call to FOO".
741 * fixed bug 32: functions defined in non-null lexical environments
742 now have more legible printed representation
743 * fixed bug 33: functions defined in non-null lexical environemnts
744 are now more amenable to inspection by INSPECT.
745 * workaround for bug 354: XEPs no longer appear in backtraces unless
746 explicitly requested.
747 * fixed bug: receiving the signal which results from attempting to
748 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
749 SIGBUS on other *BSDs) on architectures where the C stack is also
750 the Lisp stack (x86 and x86-64) from foreign code no longer leads
751 to debugger confusion or wild instability. (reported by Cheuksan
753 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
754 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
755 (reported by Lutz Euler)
756 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
757 compiled code. (reported by Paul Dietz)
758 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
759 specialized parameter to a method to something that is not TYPEP
760 the specializer is now possible.
761 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
762 face of package deletion.
763 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
764 pathname. (reported by David Tolpin on c.l.l)
765 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
766 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
767 STRING-STREAM and FILE-STREAM, respectively.
768 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
769 than previously; in addition, its test suite is now run on build.
770 (thanks to Luke Gorrie)
771 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
772 * optimization: added a immediate representation for single-floats
774 * fixed some bugs revealed by Paul Dietz' test suite:
775 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
776 ** attempting to create a package with a colliding nickname causes
777 correctable errors to be signalled.
778 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
779 ** several x86-64 backend bugs related to sign-extension of immediate
782 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
783 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
784 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
786 * fixed loading of multiply forward-referenced layouts.
787 (thanks to Cheuksan Wang)
788 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
789 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
790 (see COMPILE-FILE documentation for details of :PRINT option)
791 * fixed bugs 19 and 317: fixed-format floating point printing is
792 more accurate. This also fixes a bug reported by Adam Warner
793 related to the ~@F format directive.
794 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
796 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
797 dispatch table if the from-char is a dispatch macro character.
798 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
799 by the wordsize no longer ignore the last word. (reported by Lutz
801 * fixed bug in type checking of dynamic-extent variables. (reported
803 * optimization: sequence traversal functions use their freedom to
804 coerce function designators to functions.
805 * optimization: code with many calls to CLOS methods specialized on
806 CLOS classes has had redundant type checks removed.
807 * optimization: added declarations to speed up operations that access
808 the internal character database (for example STRING-UPCASE)
809 * optimization: comparison operations between floats and sufficiently small
810 fixnums no longer create extra rationals
811 * fixed some bugs related to Unicode integration:
812 ** portions of multibyte characters at the end of buffers for
813 character-based file input are correctly transferred to the
814 start of the buffer at the next read.
815 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
816 passing it through to OPEN.
817 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
818 argument given to internal calls to OPEN.
819 ** the built-in comment readers (introduced by character sequences
820 ";" and "#|") are more forgiving to encoding errors; they will
821 STYLE-WARN and then attempt to resync the stream at a character
822 boundary. (thanks to Teemu Kalvas)
823 * fixed some bugs in the x86-64 port:
824 ** Negative short int return values from c-calls are sign-extended
826 ** The stack is aligned to 16-bytes for c-calls, as required by
827 the ABI. (thanks to Cheuksan Wang)
828 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
829 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
830 for enums. (thanks to Vincent Arkesteijn)
831 ** Multiple small optimizations and bugfixes for floating point
833 * fixed some bugs revealed by Paul Dietz' test suite:
834 ** Space, Tab, Linefeed, Return and Page have the invalid
835 secondary constituent character trait.
836 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
838 ** WITH-INPUT-FROM-STRING should only update the index place on
840 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
842 ** Bit-array operations (BIT-AND and similar) worked incorrectly
843 with one-dimensional arrays with fill pointers.
844 ** TYPE-OF failed on a complex with an integer realpart and a
846 ** compiler failure during type inference for the code of form
847 (IF (EQL X (THE ...)) ...) (MISC.535).
849 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
850 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
851 platform. The port supports 61-bit fixnums, large memory spaces
852 and reloading shared object files.
853 * enhancement: saving cores with foreign code loaded is now
854 supported on ppc/Darwin in addition to the previously supported
856 * enhancement: the statistical profiler now walks deeper into the
857 call stack for more meaningful call-graphs and accrued time
858 reports (x86/x86-64 only). It also now reports time spent in
860 * enhancement: it is now possible to trace most individual methods
861 of a generic function in addition to tracing the generic function
863 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
865 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
866 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
867 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
868 produces an error. (thanks to Vincent Arkesteijn)
869 * bug fix: NAMESTRING on pathnames with :WILD components in their
870 directories works correctly. (thanks to Artem V. Andreev)
871 * fixed bug 125: compiler preserves identity of closures. (reported
873 * bug fixed: functions with &REST arguments sometimes failed with
874 "Undefined function" when compiled with (DEBUG 3). (reported by
876 * bug fix: overflow during compiling of setting element of a bit
877 vector with constant index and value. (reported by Timmy Douglas)
878 * build fix: fixed the dependence on *LOAD-PATHNAME* and
879 *COMPILE-FILE-PATHNAME* being absolute pathnames.
880 * on x86 compiler partially supports stack allocation of dynamic-extent
882 * GO and RETURN-FROM do not check the extent of their exit points
883 when compiled with SAFETY 0.
884 * fixed some bugs related to Unicode integration:
885 ** encoding and decoding errors are now much more robustly
886 handled; it should now be possible to recover even from invalid
887 input or output to the terminal. (thanks to Teemu Kalvas)
888 ** provided a first cut at implementing STRING-TO-OCTETS and
889 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
890 ** altered the SB-MD5 contributed module slightly, changing the
891 interface just enough for it to be supportable for builds where
892 lisp characters are not eight bits.
893 * fixed some bugs revealed by Paul Dietz' test suite:
894 ** the FORMATTER-generated functions for ~V[ conditionals require
895 the correct number of arguments.
896 ** READ-FROM-STRING returns the mandated second value when applied
897 to displaced strings.
898 ** the #\Rubout and #\Backspace characters are treated as invalid
899 constituent characters by the tokenizer.
901 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
902 * new feature: reloading changed shared object files with
903 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
904 * new feature: references to foreign variables and functions
905 can now be compiled and loaded before the corresponding shared
906 object file is loaded, as long as the foreign definitions are
907 available at runtime.
908 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
909 supported. (thanks to Dan Debertin)
910 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
911 just keywords. (thanks to Vincent Arkesteijn)
912 * fixed bug #331: structure-class instances corresponding to
913 DEFSTRUCT forms are now created eagerly.
914 * fixed bug #345: backtraces from calls to undefined functions work
915 on x86 as well. Related bug #61 is now also partially fixed on x86
916 (backtraces from throws to unknown catch tags.)
917 * bug fix: lambda-list parsing is now stricter vrt. order and number
918 of lambda-list keywords.
919 * bug fix: as specified by AMOP, an error is signalled if a
920 class-option appears multiple times in a DEFCLASS form. (reported
922 * bug fix: RANDOM can be compiled when the compiler derives the type
923 of its numeric argument as a disjoint set of small integers.
924 (reported by Paul Dietz)
925 * bug fix: starting a core saved with shared objects loaded when
926 those objects are not available no longer causes threaded SBCL to
927 hang. (reported by Sean Ross)
928 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
930 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
931 argument, which must always be NIL. (reported by Kalle Niemitalo)
932 * bug fix: printing 1.0d+23 no longer results in an error.
933 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
934 * bug fix: structure slot setters preserve evaluation order. (thanks
936 * fixed some bugs related to Unicode integration:
937 ** RUN-PROGRAM can allow its child to take input from a Lisp
938 stream. (reported by Stefan Scholl)
939 ** ASDF-INSTALL successfully downloads .tar.gz files in all
940 locales. (reported by Ken Causey)
941 * fixed some bugs revealed by Paul Dietz' test suite:
942 ** INCF, DECF and REMF evaluate their place form as specified in
944 ** FORMATTER expands ~{ ~} iteration directives with V or #
945 parameters correctly.
946 ** FORMATTER deals with the ~@[ ~] conditional directive where the
947 consequent uses no arguments correctly.
948 ** the system has a partial understanding of the (COMPLEX RATIO)
950 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
953 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
954 * new feature: a build-time option (controlled by the :SB-UNICODE
955 keyword feature, enabled by default) for building the system with
956 support for the entire 21-bit character space defined by the
958 * new feature: the system now has rudimentary external-format
959 support; the primary user-visible change at this time is that
960 characters with the high bit set (such as lower-case-e-acute) will
961 print correctly to a terminal in a UTF-8 environment.
962 * minor incompatible change: BASE-CHAR no longer names a class;
963 however, CHARACTER continues to do so, as required by ANSI.
964 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
965 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
966 has been moved to the SB-EXT package (temporarily re-exported from
968 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
969 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
971 * fixed bug #351: better error-handlind and reporting for malformed
973 * fixed bug #350: bignum-printing is now more memory-efficient,
974 allowing printing of very large bignums, eg. (expt 2 10000000).
975 (reported by Bruno Haible)
976 * fixed bug #302: better primitive-type selection for intersection
978 * fixed bug #308: non-graphic characters now all have names, as
979 required. (reported by Bruno Haible)
980 * bug fix: redefining a class with different superclasses now correctly
981 removes it from the direct-subclasses of its previous superclasses.
982 (reported by David Morse)
983 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
984 new value now works. (reported by Bruno Haible)
985 * bug fix: correct canonicalization of multiple non-standard slot
986 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
987 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
988 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
989 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
990 now exists, an signals an error.
991 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
992 during macroexpansion for non-iteration variables as well. (reported
993 by Bruno Haible for CMUCL)
994 * bug fix: Cyclic structures and unprintable objects in compiler
995 messages no longer cause errors. (reported by Bruno Haible)
996 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
997 and READ-FROM-STRING all now return a primary value of NIL if
998 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
999 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1000 (reported by Bruno Haible for CMUCL)
1001 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1002 do not designate a condition. (reported by Bruno Haible for
1004 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1005 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1006 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1007 permitting resolution in favour of any of the conflicting symbols.
1008 (reported by Bruno Haible for CMUCL)
1009 * bug fix: EQUAL compiler optimizations is less aggressive on
1010 strings which can potentially compare true despite having distinct
1011 specialized array element types.
1012 * bug fix: unit enumerations can be defined without dividing by
1013 zero. (thanks to Vincent Arkesteijn)
1014 * FORMAT compile-time argument count checking has been enhanced.
1015 (report from Bruno Haible for CMUCL)
1016 * a partial workaround for the bug 262: the compiler does not try to
1017 inline-expand a local function doing RETURN-FROM from a deleted
1018 BLOCK. (thanks to Peter Denno for the bug report and to David
1019 Wragg for the simple test case)
1020 * fixed some bugs revealed by Paul Dietz' test suite:
1021 ** RENAME-PACKAGE allows all package designators as new package
1023 ** constraint propagation and lambda variable substitution are
1024 more cautious in dealing with partially deleted code.
1025 ** compiler sometimes lost reoptimization passes.
1026 ** CERROR, given a condition as condition designator, passes its
1027 remaining arguments to the continue format control without
1029 ** Case-altering FORMAT directives work correctly on non-ASCII
1031 ** The REAL type specifier handles bounds outside the floating
1032 point ranges without signalling FLOATING-POINT-OVERFLOW.
1033 ** Functions with IR1-transformations can create intercomponent
1034 references to global functions.
1035 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1037 ** FORMAT ~R treats a nil value for its first parameter correctly.
1039 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1040 * enhancement: saving cores with foreign code loaded is now
1041 supported on x86/NetBSD and sparc/Linux in addition to the previously
1042 supported platforms.
1043 * bug fix: on some platforms repeated installations caused multiple
1044 copies of HTML documentation to be installed -- should not happen
1045 any more. (reported by Stefan Scholl)
1046 * bug fix: parsing self-recursive alien record types multiple times
1047 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1048 original patch by Helmut Eller for CMUCL)
1049 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1050 (thanks to Richard Kreuter)
1051 * bug fix: defining classes whose accessors are methods on existing
1052 generic functions in other (locked) packages no longer signals
1053 bogus package lock violations. (reported by François-René Rideau)
1054 * bug fix: special variables as DEFMETHOD parameters no longer have
1055 associated bogus type declarations. (reported by David Wragg and
1057 * bug fix: read-write consistency on streams of element-type
1058 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1059 * bug fix: redefiniton of the only method of a generic function with
1060 no DEFGENERIC no longer emits a full WARNING. In addition,
1061 redefinition of generic functions with no DEFGENERIC to an
1062 incompatible lambda list now signals an error. (thanks to Zach
1064 * bug fix: DEFGENERIC now works even when there's a function of the
1065 same name in an enclosing lexical environment. (thanks to Zach
1067 * fixed compiler failure, caused by instrumenting code during
1068 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1069 * optimization: added loop analysis and improved register allocation
1070 * fixed some bugs revealed by Paul Dietz' test suite:
1071 ** POSITION on displaced vectors with non-zero displacement
1072 returns the right answer.
1073 ** (SIMPLE-STRING) is a valid type specifier for sequence
1075 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1077 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1078 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1080 ** The justification version of the FORMAT ~< directive treats
1081 non-zero minpad parameter correctly.
1083 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1084 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1085 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1086 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1087 the supported interface.
1088 * new feature: Single-stepping of code compiled with DEBUG 2 or
1089 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1090 * new feature: saving cores with foreign code loaded is now
1091 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1092 Timothy Moore's work for CMUCL)
1093 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1094 parameters to * instead of NIL if no initform is supplied.
1095 (reported by Johan Bockgaard)
1096 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1097 correctly when FUNCALL appears as the car of the form. Note:
1098 despite this FUNCALL forms are not currently subject to
1099 compiler-macro expansion. (port of Raymond Toy's fix for the
1100 same from CMUCL, reported by Johan Bockgaard)
1101 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1102 (thanks for Teemu Kalvas)
1103 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1104 case instead of terminating the iteration (thanks for Julian
1105 Squires, Sean Champ and Raymond Toy)
1106 * bug fix: incorrect expansion of defgeneric that caused a style
1107 warning. (thanks for Zach Beane)
1108 * bug fix: slot accessor effective method computation works properly
1109 for classes with multiple non-standard applicable methods on
1110 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1111 * on x86 compiler supports stack allocation of results of LIST and
1112 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1113 CMUCL implementation by Gerd Moellmann)
1114 * fixed some bugs revealed by Paul Dietz' test suite:
1115 ** FORMAT strings with both the ~<~:;~> form of the justification
1116 directive and pretty-printing directives cause an error.
1118 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1119 * incompatible change: the internal functions
1120 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1121 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1122 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1123 instead of the old functions.
1124 * new feature: on platforms where "dladdr" is available foreign
1125 function names now appear in backtraces. (based on Helmut Eller's
1127 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1128 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1130 * documentation: ASDF manual is now distributed with SBCL.
1131 * bug fix: SBCL can now load its contributed modules with REQUIRE
1132 even if the system-provided entries have been removed from
1133 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1135 * bug fix: backtraces involving undefined functions or assembly
1136 routines are more informative. (thanks to Brian Downing)
1137 * bug fix: mutually referent alien structures now work correctly.
1138 (reported by Rick Taube)
1139 * bug fix: structures defined by WITH-ALIEN can be referred to
1140 within other definitions in the same WITH-ALIEN.
1141 * bug fix: division operators (MOD, TRUNCATE and the like) with
1142 constant zero divisors and integer dividends no longer generate
1144 * bug fix: provide default methods for INPUT-STREAM-P and
1145 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1146 * bug fix: improve the detection and resolution of MOP metacycles
1147 (where effective method computation depends on the generic function
1148 having its effective method computed). (reported by Bruno Haible)
1149 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1150 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1151 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1152 * bug fix: pathnames with relative directory components can be
1153 represented relative to default pathnames.
1154 * optimization: in taking the GCD of bignums, reduce the two bignums
1155 to approximately the same size (using Euclid's algorithm) before
1156 applying the more sophisticated binary GCD. (thanks to Juho
1158 * optimization: COUNT on bitvectors now operates word-at-a-time.
1159 * optimization: ASH with a positive, but not necessarily constant,
1160 (leftwards) shift, when in a modular context, is compiled to a
1162 * fixed some bugs revealed by Paul Dietz' test suite:
1163 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1164 if the corresponding argument is NIL.
1165 ** FORMAT directives accepting a minpad parameter treat negative
1167 ** literal commas as character parameters to FORMAT directives are
1169 ** literal spaces directly after ~<Newline> directives within a
1170 format-logical-block (~:< ~@:>) do not induce :FILL-style
1171 conditional newlines.
1172 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1173 as specified: it no longer includes conditional newlines).
1174 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1176 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1177 :PER-LINE-PREFIX argument does not evaluate to a string.
1179 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1180 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1181 the manual for details; for now, package locks can be disabled by
1182 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1183 if no major problems are found then it is likely that they will be
1184 compiled in unconditionally.
1185 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1186 unsupported operators on all platforms. To load a shared library
1187 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1188 object file, link it into a shared library outside of SBCL and
1189 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1191 * bug workaround: ROOM T and the GC were not working together
1192 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1193 reporting facility aren't preserved. That reporting has been
1194 disabled, so now until and unless someone figures out how to make it
1195 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1196 * minor incompatible change: as threatened around sbcl-0.8.0, the
1197 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1198 The symbols are also exported from SB-PCL for backwards
1199 compatibility, but more so than before SB-PCL should be treated as
1200 an implementation-internal package.
1201 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1203 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1204 It is known as of this release not to work on the Alpha, however.
1205 * fixed bug #167: errors signalled due to illegal syntax in method
1206 bodies are now more legible.
1207 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1208 designators and can hence be used with TYPEP.
1209 * fixed bug #333: CHECK-TYPE now ensures that the type error
1210 signalled, if any, has the right object to be accessed by
1211 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1212 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1213 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1215 * fixed bug #334: programmatic addition of slots using specialized
1216 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1217 and :ALLOCATION :CLASS slots.
1218 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1219 float exponents. (rereported by Peter Seibel)
1220 * fixed a bug: #\Space (and other whitespace characters) are no
1221 longer considered to be macro characters in standard syntax by
1222 GET-MACRO-CHARACTER.
1223 * fixed bug: initialization of condition class metaobjects no longer
1224 causes an instance of the condition to be created. (reported by
1226 * fixed bug: it is now possible to have slots such that
1227 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1228 description is neither :INSTANCE nor :CLASS.
1229 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1230 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1231 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1232 non-standard SLOT-DEFINITION-ALLOCATION values.
1233 * fixed bugs: various race conditions handling exiting threads.
1234 CL-PPCRE's thread torture test now passes.
1235 * fixed arguably-a-bug: GC time proportional to number of threads in
1236 system even when most of them are idle
1237 * optimization: improved performance of BIT and SBIT on bit-vectors.
1238 * .fasl file incompatibility: The fasl file version number has
1239 been incremented because of changes associated with package locks.
1241 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1242 * minor incompatible change: the system no longer provides
1243 optimization hints (conditions which are TYPEP
1244 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1245 mode; these hints are still emitted when higher SPEED optimization
1247 * new contrib module: a sampling profiler (profiling by statistical
1248 sampling, rather than by instrumenting functions) is available as
1249 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1250 * the behaviour of the standard function ED is now customizeable by
1251 third parties through a hook variable: see ED's documentation
1252 string for information on the protocol.
1253 * the compiler no longer emits efficiency notes for (FUNCALL X)
1254 when the type of X is uncertain under default optimization
1256 * fixed bug 276: mutating a binding of a specialized parameter to a
1257 method to something that is not TYPEP the specializer is now
1259 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1260 exhibit the required behaviour on the x86 platform. (thanks to
1261 Peter van Eynde, Eric Marsden and Bruno Haible)
1262 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1263 even for difficult arguments. (reported by Peter Graves)
1264 * fixed bug 141a: the backquote printer now descends quoted
1266 * fixed another bug in backquote printing: no more destructive
1267 modification of the form's list structure. (reported by Brian
1269 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1270 move between its address being taken and the call to
1271 interrupt_thread, fixing a crashing race condition.
1272 * the SB-POSIX contrib implementation has been adjusted so that it
1273 no longer exhibits ridiculously poor performance when constructing
1274 instances corresponding to C structs.
1276 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1277 * minor incompatible change: the sb-grovel contrib now treats C
1278 structures as alien (in the sense of SB-ALIEN) objects rather than
1279 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1280 has implications for memory management of client code
1281 (sb-grovel-returned objects must now be manually managed) and for
1282 type safety (alien objects now have full types).
1283 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1284 used to control emission of compiler diagnostics, rather than the
1285 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1286 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1287 quality should be considered deprecated.
1288 * install.sh now installs the user manual as well
1289 * (not quite a new documentable feature, but worth considering in
1290 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1291 beginnings of a semantically meaningful condition hierarchy is
1292 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1294 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1295 designator as the defaults argument.
1296 * fixed bug: Displaced arrays whose displaced-to array has become
1297 too small now cause ARRAY-DIMENSION to signal an error, providing
1298 for safer bounds-checking. (reported by Bruno Haible)
1299 * fixed bug: DEFCLASS slot definitions with identical :READER and
1300 :WRITER names now signal a reasonable error. (reported by Thomas
1302 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1304 * fixed bug: Invalid slot specification errors now print correctly.
1305 (thanks to Zach Beane)
1306 * fixed bug 320: Shared to local slot value transfers in class
1307 redefinitions now happen corrently from superclasses as
1308 well. (reported by Bruno Haible)
1309 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1311 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1312 improper lists correctly. (reported by Bruno Haible)
1313 * fixed bug 313: source-transform for <fun-name> was erroneously
1314 applied to a call of a value of a variable with name <fun-name>.
1315 (reported by Antonio Menezes Leitao)
1316 * fixed bug 307: The obsolete instance protocol ensures that
1317 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1318 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1319 remove dead unknown-values globs from the middle of the stack.
1320 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1321 easier to resume long computations after using *BREAK-ON-SIGNALS*
1322 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1323 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1324 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1325 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1326 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1327 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1329 * fixed a (fairly theoretical) bug in string printing: if
1330 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1331 does not have array-element-type equal to the most general string
1333 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1334 function name. (thanks to Nikodemus Siivola)
1335 * fixed bug: compile-time format string checker failed on
1336 non-closed ~{. (reported by Thomas F Burdick)
1337 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1338 don't warn on legal loop code involving more than one aggregate
1339 boolean. (thanks to Nikodemus Siivola)
1340 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1341 in keyword position to MAKE-PACKAGE was not being reported as
1343 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1344 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1346 * fixed bugs 280 and 312: the checking for multiple definitions in a
1347 file is less likely to become confused by uses of inline
1349 * fixed bug: the #S reader macro performs the keyword coercion
1350 specified for slot names. (reported by Kalle Niemitalo)
1351 * fixed bug: lambda lists may contain symbols whose names start with
1352 & but are not lambda-list-keywords; their occurrence triggers a
1354 * fixed bug 321: define-method-combination argument lambda lists do
1355 not cause invalid code to be generated when &optional or &aux
1356 variables have default values. (reported by Bruno Haible)
1357 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1358 have CLOS classes; this appears also to have fixed reported
1359 instability in stack exhaustion detection.
1360 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1361 MUFFLE-WARNING when no associated restart is present is now
1363 * optimization: rearranged the expansion of various defining macros
1364 so that each expands into only one top-level form in a
1365 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1367 * optimization: used a previously little-used slot in symbols to
1368 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1370 * fixed some bugs revealed by Paul Dietz' test suite:
1371 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1373 ** Fixed an optimization bug related to inheritance of initforms
1374 from local to shared slots.
1375 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1376 ** CAST optimizer forgot to flush argument derived type.
1377 ** print/read consistency on floats is now orders of magnitude
1378 more likely. (thanks also to Bruno Haible for a similar report
1380 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1381 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1382 for objects of type REAL. Make it so.
1383 ** SXHASH is specified (infelicitously) to respect similarity,
1384 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1385 it so. (thanks to Markus Ziegler)
1386 ** on the Alpha, the compiler succeeds in compiling functions
1387 returning a known number of arguments greater than 63.
1388 ** fixed handling of invalid NIL arguments in keyword position
1390 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1392 ** provided a readably-printable representation for RANDOM-STATE
1394 ** ensured that pathnames, where they have a namestring, always
1395 print using #P"..." syntax.
1397 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1398 * Support for the forthcoming 2.0 version of the NetBSD kernel
1399 running on x86 hardware has been added. (thanks to Perry
1400 E. Metzger most immediately, and others for their past work)
1401 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1402 current release version; SBCL's OpenBSD support had been broken
1403 since about the time of OpenBSD's switch to ELF binary format.)
1404 * [placeholder for DX summary]
1405 ** user code with &REST lists declared dynamic-extent, under high
1406 speed or space and low safety and debug optimization policy.
1407 * The manual has been converted to Texinfo format and the debugger
1408 chapter from the cmucl manual has been added.
1409 * A facility has been added to extract documentation strings from
1410 sbcl and store them as Texinfo-formatted snippets for inclusion in
1411 the manual (via Texinfo's @include directive)
1412 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1413 the test case to Dave Roberts)
1414 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1415 pointers. (reported by Sean Ross)
1416 * bug fix: PROFILE output is printed nicely even for large numerical
1417 values. (thanks to Zach Beane)
1418 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1419 greater than 32 handle EOF correctly.
1420 * bug fix: on X86 an immediate argument of the IMUL instruction is
1421 correctly printed in disassembly. (thanks to Lutz Euler)
1422 * bug fix: class slots in redefined classes preserve their old
1423 values. (thanks to Bruno Haible and Nikodemus Siivola)
1424 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1425 succeeds. (reported by Marco Baringer)
1426 * fixed some bugs revealed by Paul Dietz' test suite:
1427 ** READ-SEQUENCE now works on ECHO-STREAMs.
1428 ** RATIONALIZE works more according to its specification. (thanks
1431 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1432 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1433 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1434 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1435 should matter to you only if you rebind the printer control
1436 variables and then find you want different bindings in the
1437 debugger than in the ordinary execution of your program.)
1438 * The runtime build system has been tweaked to support building
1439 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1440 assembler and linker. (thanks to Nikodemus Siivola)
1441 * Unbound, undefined, undeclared variables now trigger full
1442 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1443 more often programmer error than deliberate exploitation of undefined
1445 * optimization: the hash algorithm for strings has changed to one
1446 that is less vulnerable to spurious collisions. (thanks to Juho
1448 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1449 needless bounds checking. (thanks to Juho Snellman)
1450 * optimization: implemented multiplication as a modular
1451 (UNSIGNED-BYTE 32) operation on the PPC backend.
1452 * fixed some bugs revealed by Paul Dietz' test suite:
1453 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1455 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1456 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1457 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1459 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1461 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1463 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1464 their output stream on EOF from read.
1465 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1466 have been read to end-of-file.
1467 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1469 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1470 description of determination of which consecutive characters
1472 ** Printing the "Space" character with escaping on now yields "#\\ ",
1473 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1474 ** Reading floating-point numbers with *READ-BASE* set to a number
1475 less than 10 works correctly.
1476 ** Reading floating-point numbers with *READ-BASE* set to a number
1477 more than 10 works correctly.
1478 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1479 the readtable currently in effect.
1481 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1482 * minor incompatible change: parsing of namestrings on a physical
1483 (Unix) host has changed; numbers after the final #\. in a
1484 namestring are no longer interpreted as a version field. This is
1485 intented to be largely invisible to the user, except that the
1486 meaning of the namestring "*.*.*" has changed: it now refers to a
1487 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1488 should usually be replaced by
1489 (make-pathname :name :wild :type :wild :version :wild)
1490 with the added benefit that this is more likely to be portable.
1491 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1492 option to OPEN now signals an error if the file being opened
1493 exists; this may have an impact on existing code.
1494 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1495 signals received, on the PowerPC platforms (both Linux and
1496 Darwin). (thanks to Pierre Mai for pointing out the location of
1498 * several fixes on OS X: The system now builds and runs cleanly on
1499 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1500 (thanks to Brian Mastenbrook)
1501 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1502 second-resolution: integer multiples of 1/3600 between -24 and 24.
1503 (thanks to Vincent Arkesteijn)
1504 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1505 type when called with 1 argument; PEEK-CHAR checked type of
1506 PEEK-TYPE only after having read first character from a
1507 stream. (reported by Peter Graves)
1508 * bug fix: the garbage collector now has much better locality
1509 behaviour, and in particular no longer treats all memory as being
1510 exhausted when many small objects point to each other in a deeply
1512 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1513 recognized as being TYPEP their class.
1514 * bug fix: the PUSHNEW documentation string has been corrected.
1515 (thanks to Vincent Arkesteijn)
1516 * bug fix: defaulting of the value for the last of an atypically
1517 large number of multiple values being bound was not being
1518 performed correctly on the Alpha or PPC platforms
1519 * optimization: implemented multiplication as a modular
1520 (UNSIGNED-BYTE 32) operation on the x86 backend.
1521 * optimization: SEARCH on simple-base-strings can now be open-coded.
1522 (see also contrib/compiler-extras.lisp for inspiration for
1523 teaching the compiler about the Boyer-Moore algorithm).
1524 * value, returned by MAX (and MIN) called with several EQUALP, but
1525 not EQL, arguments now does not depend on compiler settings.
1526 * fixed some bugs revealed by Paul Dietz' test suite:
1527 ** in stack analysis liveness information is propagated from
1528 non-local entry points.
1529 ** pathwise CAST removing failed when the CAST node did not start
1531 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1532 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1534 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1535 host is already defined.
1536 ** RENAME-FILE works on streams instead of signalling an internal
1538 ** PEEK-CHAR uses the current readtable when determining whether
1539 or not a character is whitespace.
1540 ** MERGE-PATHNAMES handles the case when the pathname does not
1541 specify a name while the default-pathname specifies a version.
1542 ** Pathnames now stand a better chance of respecting print/read
1544 ** Attempting to use standardized file system operators with a
1545 pathname with invalid :DIRECTORY components signals a
1547 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1549 ** DIRECTORY on logical pathnames is more correct.
1550 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1551 signal a TYPE-ERROR if their argument is not a stream.
1552 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1553 designator argument does not designate a stream.
1554 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1555 examining the synonym.
1556 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1558 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1559 element-type larger than ([UN]SIGNED-BYTE 32).
1561 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1562 * When built with the :SB-FUTEX feature, threaded builds now take
1563 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1564 for faster/more reliable mutex and condition variable support.
1565 * Incompatible change (but one you probably shouldn't have been using
1566 anyway): the interface and code for arbitrating between multiple
1567 threads in the same user session has been redesigned.
1568 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1569 error when called without an explicit environment argument.
1570 (thanks to Nikodemus Siivola)
1571 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1572 confused about their position. (thanks to Adam Warner and Gerd
1574 * bug fix: Pretty printing backquoted forms with unquotations in the
1575 argument list position of various code constructs such as LAMBDA
1576 now works correctly. (reported by Paul Dietz)
1577 * bug fix: Pretty printing unquotations no longer loses all
1578 stream position information.
1579 * optimization: performance of string output streams is now less
1580 poor for multiple small sequence writes.
1581 * optimization: performance of CSUBTYPEP in the presence of complex
1582 expressions involving CONS and NOT many times has been improved.
1583 (reported by Paul Dietz)
1584 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1586 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1587 streams for the REPL, for the debugger, and so on. (thanks to
1589 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1590 * fixed some bugs revealed by Paul Dietz' test suite:
1591 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1592 is ignored if it is too small, rather than propagating through
1593 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1594 ** extremely complex negations of CONS types were not being
1595 sufficiently canonicalized, leading to inconsistencies in
1597 ** VALUES tranformer lost derived type.
1599 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1600 * fixed a bootstrapping bug: the build process no longer assumes
1601 that the various BOOLE-related constants have the same value in
1602 host and target lisps. (noted by Paul Dietz' test suite on an
1603 SBCL binary built from CLISP)
1604 * The system can now be dynamically linked on the MIPS platform,
1605 which enables dynamic loading of foreign code from Lisp. (thanks
1606 to Ralf Baechle for discussions on the MIPS ABI)
1607 * The system now records debugging information for its own source
1608 files in a filesystem-position-independent manner, relative to
1609 the "SYS" logical host.
1610 * fixed a compiler bug: MV-LET convertion did not check references
1611 to the "max args" entry point. (reported by Brian Downing)
1612 * tweaked disassembly notes to be less confident about proclaiming
1613 some instruction as an LRA. (thanks to Brian Downing)
1614 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1615 now each have their own history, command character, and other
1616 characteristics. (thanks to David Lichteblau)
1617 * fixed some bugs revealed by Paul Dietz' test suite:
1618 ** compiler failure in compiling LOGAND expressions including a
1620 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1621 shift greater than 32.
1622 ** FLUSH-DEST did not mark blocks for type check regeneration.
1623 ** HANDLER-CASE failed to accept declarations in handler clauses
1624 in some circumstances.
1626 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1627 * New code in contrib/sb-introspect (still probably not entirely
1628 stable yet) provides some support for smart Lisp development
1629 environments like SLIME.
1630 * The conditions signalled for errors occurring when loading .fasl
1631 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1632 in a way which should help ASDF recover gracefully.
1633 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1634 stuff has been cleaned up. If you code contrib/ stuff, this might
1635 affect you, and you can look at contrib/README, contrib/STANDARDS,
1636 and/or the 0.8.4.27 diff to check.
1637 * In full calls the compiler now does not generate checks for declared
1638 argument types for all arguments.
1639 * various threading fixes
1640 ** and some experimental patches which didn't make it into
1641 the main tree for this release, but which are shipped in
1642 contrib/experimental-thread.patch as a possible fix for some
1643 failures (deadlock, spinning...) in GC-intensive multithreaded
1645 * fixed PPC build problem (source code incompatibility of different
1646 library versions): added offsetof() hackery which attempts to divine
1647 where glibc maintainers put uc_mcontext today
1648 * fixed bug 282: compiler does not trust type assertions while passing
1649 arguments to a full call.
1650 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1651 &OPTIONAL VALUES type parameter.
1652 * fix bug 214: algorithm for noting rejected templates is now more
1653 similar to that of template seletion. (also reported by rydis on
1655 * fixed bug 141b: printing backquoted information readably and prettily
1656 inserts a space where necessary.
1657 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1658 (CHARACTER 10) are now reported as errors, rather than propagated
1659 as unknown types. (reported by piso on #lisp)
1660 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1661 with values NIL and :ERROR. (thanks to Milan Zamazal)
1662 * fixed bug 191c: CLOS now does proper keyword argument checking as
1663 described in CLHS 7.6.5 and 7.6.5.1.
1664 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1665 longer raise an error; further, using a list as a for-as-arithmetic
1666 counter now raises a meaningful error.
1667 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1668 now understood by sequence creation functions such as MAKE-SEQUENCE
1670 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1671 a STRING-INPUT-STREAM.
1672 * compiler enhancement: SIGNUM is now better able to derive the type
1674 * type declarations inside WITH-SLOTS are checked. (reported by
1676 * fixed some bugs revealed by Paul Dietz' test suite:
1677 ** incorrect optimization of TRUNCATE for a positive first
1678 argument and negative second.
1679 ** compiler failure in let-convertion during flushing dead code.
1680 ** compiler failure while deriving type of TRUNCATE on an
1681 interval, containing 0.
1682 ** ASH of a negative bignum by a negative bignum count now returns
1684 ** intersection of CONS types now canonicalizes properly, fixing
1685 inconsistencies in SUBTYPEP.
1687 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1688 * incompatible change: The --disable-debugger command line
1689 option now clobbers the debugger at a more fundamental
1690 level, by redefining #'INVOKE-DEBUGGER instead of by
1691 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1692 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1693 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1694 enter the debugger REPL and then suffer recursive errors
1695 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1696 while under the new system, BREAK in --disable-debugger mode
1697 terminates the system just as an unhandled error would.
1698 * fixed compiler performance when processing loops with a step >1;
1699 * bug fix: DOCUMENTATION now retrieves generic function
1700 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1701 support has been systematized, and now supports the methods
1702 specified by ANSI, along with a default method and a method for
1703 slot documentation. (reported by Nathan Froyd)
1704 * bug fix: effective methods associated with a generic function are
1705 no longer cached over a change of that generic function's method
1706 combination. (reported by Andreas Fuchs)
1707 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1708 names a non-existent directory. (thanks to Andreas Fuchs)
1709 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1710 the PPC platform, lead to stack corruption; this has been fixed.
1711 (reported by Rainer Joswig)
1712 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1713 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1714 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1715 arithmetic in general, is now much more reliable on the PPC
1717 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1718 platform now returns the right answer.
1719 * optimization: restored some effective method precomputation in
1720 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1721 precomputation is now tunable.
1722 * optimization: compiler-internal data structure use has been
1723 reviewed, and changes have been made that should improve the
1724 performance of the compiler by about 20%.
1725 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1726 simple-base-strings and simple-bit-vectors is improved.
1727 * optimization: the optimization of 32-bit logical and arithmetic
1728 functions introduced in version 0.8.3 on the x86 has been
1729 implemented on the mips, ppc and sparc platforms; an
1730 implementation of the same facility, but for 64-bit arithmetic,
1731 has been added for the alpha.
1732 * microoptimization: the compiler is better able to make use of the
1733 x86 LEA instruction for multiplication by constants.
1734 * bug fix: in some situations compiler did not report usage of
1735 generic arithmetic in (SPEED 3) policy.
1736 * bug 145b fix: compiler used wrong type specifier while converting
1737 MEMBER-types to numeric.
1738 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1740 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1741 index argument now works properly on the Alpha platform.
1742 * bug fix: floating point exception treatment on the Alpha platform
1744 * bug fix: FILE-POSITION works much better on string input and
1745 output streams. (thanks to Nikodemus Siivola)
1746 * bug fix: many threading/garbage collection symptoms sorted.
1747 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1748 might be pseudo-atomic.
1749 * internal change: Stopping for GC is now done with signals not
1750 ptrace. GC is now done in whichever thread wanted it, instead of
1752 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1754 * bug fix: result form in DO is not contained in the implicit
1756 * incompatible change: ICR structure is changed; the value part of
1757 CONTINUATION is now called LVAR; corresponding functions are
1758 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1759 * added type deriver for ISQRT (thanks to Robert E. Brown).
1760 * bug fix: better support for loading from the command line when an
1761 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1763 * fixed some bugs revealed by Paul Dietz' test suite:
1764 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1765 ** ROUND and FROUND now give the right answer when given very
1766 small float arguments.
1767 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1769 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1770 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1771 ** LCM should return a non-negative integer.
1772 ** PARSE-INTEGER returned the index of a terminator instead of the
1773 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1774 ** PARSE-INTEGER returned an incorrect index being applied to a
1776 ** LCM with two arguments of 0 returns 0 rather than signalling
1778 ** unsigned addition of a 32-bit constant with the high bit set no
1779 longer causes an internal compiler error on the x86.
1780 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1781 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1782 with negative last argument.
1783 ** byte specifiers with zero size and position no longer cause
1784 an error during type derivation.
1785 ** bignum multiplication on the Alpha platform now returns the
1787 * porting: The system now builds on SuSE AMD64, although it still
1788 generates a 32-bit binary.
1789 * .fasl file incompatibility: The fasl file version number has
1790 been incremented (because of the changes to internal compiler
1791 data structures referred to above).
1793 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1794 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1795 more accurately, on the Darwin kernel running on PowerPC hardware.
1796 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1797 * Compiler code deletion notes now signal a condition of type
1798 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1799 an associated MUFFLE-WARNING restart.
1800 * The compiler now performs limited argument count validation of
1801 constant format strings in FORMAT, and where appropriate in ERROR,
1802 CERROR and WARN. (thanks to Gerd Moellmann)
1803 * New ASDF-INSTALL contrib can be used for automatic download and
1804 installation of third-party Lisp code from CCLAN or other sites
1806 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1807 which forces another thread to execute a function supplied by the
1809 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1810 now accept and act upon their :ELEMENT-TYPE keyword argument.
1811 (reported by Martin Atzmueller, Edi Weitz)
1812 * bug fix: FILE-POSITION now accepts position designators up to
1813 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1814 is the greater. (thanks to Patrik Nordebo)
1815 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1816 Kalle Olavi Niemitalo)
1817 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1818 printable. (reported by Eric Marsden)
1819 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1820 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1821 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1822 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1823 (thanks to Stig Sandoe)
1824 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1825 UNREAD-CHAR now correctly includes the unread character in the
1826 target sequence. (thanks to Gerd Moellmann)
1827 * bug fix in threaded builds: the system can now be suspended and
1828 resumed by shell job control with minimal disruption.
1829 * bug fixes in times and timezones >2038AD
1830 * better handling of "where is GNU make?" problem in build scripts
1831 (thanks to Nikodemus Siivola)
1832 * new optimization: inside a named function any reference to a
1833 function with the same name is considered to be a self-reference;
1834 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1835 optimization quality.
1836 * new optimization on x86: logical functions and + now have
1837 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1838 used when the result is truncated to 32 bits.
1839 * VALUES declaration is partially enabled.
1840 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1841 * fixed some bugs revealed by Paul Dietz' test suite:
1842 ** The system now obeys the constraint imposed by
1843 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1844 types form a lattice under type intersection.
1845 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1846 ** ASSOC now ignores NIL elements in an alist.
1847 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1848 and (1+ MOST-POSITIVE-FIXNUM) answers.
1849 ** The addition of a method with invalid qualifiers to a generic
1850 function does not cause an error to be signalled immediately;
1851 a warning is signalled, and the error is generated only on
1852 calling the generic function.
1853 * changed .fasl file version number, in part to add type codes for
1854 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1855 obscure ANSI requirements
1857 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1858 * fixed bug 148: failure to inline-expand a local function left
1859 garbage, confusing the compiler.
1860 * fixed bugs 3cd: structure slot readers perform type check if the
1861 slot can have an invalid value (i.e. it is either not initialized
1862 or can be written with a less specific slot writer).
1863 * bug fix: the compiler now traps array references to elements off
1864 the end of an array; previously, the bounds checking in some
1865 circumstances could go off-by-one.
1866 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1868 * type declarations for array element types now obey the description
1869 on the CLHS page "Declaration TYPE", as per discussions on
1870 sbcl-help around 2003-05-08. This means that a declaration
1871 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1872 declaration, all references to BAR will be asserted or assumed
1873 (with THE, so dependent on compiler policy) to involve objects of
1874 type FOO. Note that no such declaration is implied in
1875 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1876 * declared types of functions from the "Conditions"
1877 chapter. (reported by Paul Dietz)
1878 * bug fix: CERROR accepts a function as its first argument.
1879 * bug fix: NTH an NTHCDR accept a bignum as index
1880 arguments. (reported by Adam Warner)
1881 * optimization: character compare routines now optimize comparing
1882 against a constant character. (reported by Gilbert Baumann)
1883 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1884 argument now works properly on the MIPS platform.
1885 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1886 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1887 declared to have a functional type, does not cause an error
1889 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1890 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1891 * fixed bug: if last continuation of a deleted block has a
1892 destination, this destination should be deleted too. (reported by
1894 * fixed a bug in the bootstrap process: the host compiler's values
1895 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1896 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1897 test case from Patrik Nordebo)
1898 * improved the ability of the disassembler on the PPC platform to
1899 provide helpful disassembly notes.
1900 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1901 the class in more cases than previously.
1902 * bug fix: FILE-POSITION now understands :START and :END for
1903 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1904 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1905 8) data. (thanks to David Lichteblau)
1906 * bug fix: it is possible to add a method to a generic function
1907 without lambda list.
1908 * bug fix: reader failed to signal END-OF-FILE inside an
1909 object representation. (reported by Nikodemus Siivola)
1910 * fixed some bugs revealed by Paul Dietz' test suite:
1911 ** LAST and [N]BUTLAST should accept a bignum.
1912 ** condition slot accessors are methods.
1913 ** (VECTOR NIL) is a subtype of STRING.
1915 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1916 * minor incompatible change: some nonsensical specialized lambda
1917 lists (used in DEFMETHOD) which were previously ignored now signal
1919 * minor incompatible change: the system is now aware of the types of
1920 variables in the COMMON-LISP package, and will signal errors for
1921 most violations of these type constraints (where previously they
1922 were silently accepted).
1923 * minor incompatible change: COMPILE-FILE now uses the freedom
1924 afforded (ANSI 3.2.2.3) to use derived function types for
1925 functions defined in the same file. This also permits the system
1926 to warn on static type mismatches and function
1927 redefinition. (Currently it does not work with high DEBUG level.)
1928 * minor incompatible change: VALUES declaration is disabled.
1929 * When issuing notes, the compiler now signals a condition of type
1930 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1931 restart for use in user handlers. It is expected that the
1932 COMPILER-NOTE condition will eventually become a condition
1933 supertype to a hierarchy of note types, which will then be
1934 handleable in a similar fashion. However, at the moment, no such
1935 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1936 but it's an implementation detail, not a classification for the
1938 * Changes in type checking closed the following bugs:
1939 ** type checking of unused values (192b, 194d, 203);
1940 ** template selection based on unsafe type assertions (192c, 236);
1941 ** type checking in branches (194bc).
1942 * A short form of VALUES type specifier has ANSI meaning (it has
1943 increased the number of situations when SBCL cannot perform type
1945 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1946 keywords or constants is permissible.
1947 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1948 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1949 classes. (thanks to Antonio Martinez)
1950 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1951 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1952 argument to a pretty-print stream. Instead, it leaves any such
1953 operations to DESCRIBE-OBJECT methods.
1954 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1956 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1957 lambda lists. (thanks to David Lichteblau)
1958 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1959 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1960 (thanks to Gerd Moellmann)
1961 * fixed bug 246: increased compilation speed of long
1962 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1964 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1966 * DEFCONSTANT now throws a condition of type
1967 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1968 constant to a non-EQL value; CONTINUE and ABORT restarts
1969 respectively change and preserve the value.
1970 * fixed bug 63: The code walker, part of the implementation of CLOS,
1971 is now better at handling symbol macros.
1972 * bug fix: There is no longer an internal implementation type named
1973 CL:LENGTH. (reported by Raymond Toy)
1974 * bug fix: In macro-like defining macros/special operators the
1975 implicit block does not enclose the lambda list.
1976 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1977 enclosed in parentheses) are not suitable as type specifiers, and
1978 their use properly signals an error now.
1979 * bug fix: An argument count mismatch for a type specifier in code
1980 being compiled no longer causes an unhandled error at compile
1981 time, but signals a compile-time warning.
1982 * fixed simple vector readable printing
1983 * bug fix: DESCRIBE takes more care over whether the class
1984 precedence list slot of a class is bound before accessing it.
1985 (reported by Markus Krummenacker)
1986 * bug fix: FORMATTER can successfully compile pretty-printer format
1987 strings which use variants of the ~* directive inside.
1988 * bug fix: SEARCH now applies its TEST predicate to the elements of
1989 the arguments in the correct order. (thanks to Wolfhard Buss)
1990 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1991 in type inference. (thanks to Robert E. Brown)
1992 * bug fix: Reading in symbols with an explicit package name of ""
1993 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1994 * fixed some bugs revealed by Paul Dietz' test suite:
1995 ** NIL is now allowed as a structure slot name.
1996 ** Arbitrary numbers, not just REALs, are allowed in certain
1997 circumstances in LOOP for-as-arithmetic clauses.
1998 ** Multiple class redefinitions before slot access no longer
1999 causes a type error.
2000 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2001 association between the name and a class.
2002 ** Generic functions with non-standard method-combination and over
2003 five methods all of which return constants no longer return NIL
2004 after the first few invocations. (thanks to Gerd Moellmann)
2005 ** CALL-NEXT-METHOD with no arguments now passes the original
2006 values of the arguments, even in the presence of assignment.
2007 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2009 ** DISASSEMBLE works with closures and funcallable instances.
2010 ** ADD-METHOD now returns the generic function, not the new method.
2011 ** FIND-METHOD signals an error if the lengths of the specializers
2012 is incompatible with the generic function, even if the ERRORP
2014 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2015 which its argument is a member.
2016 ** DEFCLASS only redefines the class named by its class-name
2017 argument if that name is the proper name of the class;
2018 otherwise, it creates a new class.
2019 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2020 of the UNBOUND-SLOT condition to the name of the slot.
2021 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2022 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2023 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2024 SLOT-MAKUNBOUND in the specified fashion.
2026 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2027 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2028 cross-compilation host. As a consequence, we can now bootstrap our
2029 way up to SBCL starting with a bare gcc toolchain and human-readable
2030 source code (first the source to CLISP, then the source to SBCL).
2031 * A contributed module containing a partial implementation of the
2032 simple-streams interface has been included. (thanks to Rudi
2034 * A contributed module implementing the RFC1321 Message Digest
2035 Algorithm, known as MD5, has been included.
2036 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2037 no longer has any effect, as the code controlled by this feature
2038 has been deleted. (As far as we know, no-one has ever built using
2039 this feature, and its semantics were confused in any case).
2040 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2041 work on conditions (as required by the ANSI specification),
2042 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2043 expected behaviour on conditions. Users should note, however,
2044 that such behaviour is not required by the ANSI specification,
2045 and so use of this behaviour may render their code unportable.
2046 * fixed some bugs revealed by Paul Dietz' test suite:
2047 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2049 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2050 ** SXHASH on condition objects no longer returns NIL.
2051 ** :ALLOCATION :CLASS slots are better treated; their values are
2052 updated on class redefinition, and initforms inherited from
2053 superclasses are applied.
2054 ** REMOVE-METHOD returns its generic function argument even when
2055 no method was removed.
2056 ** SHARED-INITIALIZE now initializes the values of the requested
2057 slots, including those with :ALLOCATION :CLASS.
2058 ** ALLOCATE-INSTANCE now works on structure classes defined via
2059 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2061 ** SLOT-EXISTS-P now works on conditions, as well as structures
2063 ** MAKE-LOAD-FORM now has the required methods on
2064 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2065 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2066 keyword, and now implements the SLOT-NAMES argument.
2067 ** methods with &OPTIONAL arguments no longer allow too many
2068 arguments to be passed in the call without error.
2069 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2070 option is consistent with the required arguments of the generic
2071 function lambda list.
2072 * bug fix: REQUIRE accepts a string designator. (Thanks to
2074 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2075 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2076 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2077 * bug fix: The compiler now checks for duplicated variables in macro
2079 * bug fix: SETQ on globals returns the correct value.
2080 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2081 not a COMPILER-ERROR (followed by some other strange error on
2082 choosing the CONTINUE restart).
2083 * bug fix: make.sh and friends are now more consistent in the way that
2084 they look for GNU "make".
2086 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2087 * experimental native threads support (on x86 Linux >=2.4 only).
2088 This is not compiled in by default: you need to add :SB-THREAD to
2089 the target features. See the "Beyond ANSI" chapter of the manual
2091 * fix for longstanding nonANSIism: The old distinction between
2092 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2093 The return value from CL:FIND-CLASS is now a CLOS class, and
2094 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2095 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2096 * An interface to the MetaObject Protocol, as described in Kiczales,
2097 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2098 MIT Press, 1991, is available from the SB-MOP package.
2099 * incompatible change: the SB-PCL package should now be considered
2100 a private implementation detail, and no longer a semi-private MOP
2102 * minor incompatible change: due to rearrangement for threads, the
2103 control stack and binding stack are now allocated at arbitrary
2104 addresses instead of being hardcoded per-port. Users affected by
2105 this probably have to be doing advanced things with shared
2106 libraries, and will know who they are.
2107 * minor incompatible change: Previously, all --eval forms used were
2108 processed with READ before any of them were processed with EVAL.
2109 Now each --eval form is processed with both READ and EVAL before
2110 the next --eval form is processed. (Thus package operations like
2111 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2112 work as the user might reasonably expect.)
2113 * minor incompatible change: *STANDARD-INPUT* is now only an
2114 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2116 * minor incompatible change: Y-OR-N-P is now character-oriented,
2117 not line oriented. Also, YES-OR-NO-P now works without errors.
2118 (thanks to Antonio Martinez)
2119 * sb-aclrepl module improvements: an integrated inspector, added
2120 repl features, and a bug fix to :trace command.
2121 * Known functions, which cannot be open coded by the backend, are
2122 considered to be able to check types of their arguments. (fixing
2123 a bug report by Nathan J. Froyd)
2124 * fixed a bug in computing method discriminating functions: It is
2125 now possible to define methods specialized on classes which have
2126 forward-referenced superclasses. (thanks to Gerd Moellmann)
2127 * fixed evaluation order in optional entries (reported by Gilbert
2129 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2130 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2131 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2132 required-by-AMOP NAME argument, as well as CLASS and
2133 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2134 * fixed bug 20: DEFMETHOD can define methods using names that are
2135 not the proper names of classes to designate class specializers.
2136 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2137 Unix file descriptors, instead of blowing up. (thanks to Antonio
2139 * Garbage collection refactoring: user-visible change is that a
2140 call to the GC function during WITHOUT-GCING will not do garbage
2141 collection until the end of the WITHOUT-GCING. If you were doing
2142 this you were probably losing anyway.
2143 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2144 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2145 TYPEP the latter but not the former.
2146 * The compiler issues a full WARNING for calls to undefined functions
2147 with names from the CL package.
2148 * MAP-INTO for a vector destination is open coded. (reported by
2149 Brian Downing on c.l.l)
2150 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2151 documentation string.
2152 * fixed some bugs revealed by Paul Dietz' test suite:
2153 ** COPY-ALIST now signals an error if its argument is a dotted
2155 ** Condition slots are now accessed more correctly in the presence
2156 of multiple initargs for a given slot.
2157 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2158 exclude restarts of the same name associated with a different
2160 ** DEFCLASS of forward-referenced classes with another
2161 forward-referenced class in the superclasses list no longer
2163 ** Condition slots are now initialized once each, not multiple
2164 times. (thanks to Gerd Moellmann)
2165 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2167 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2168 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2169 arguments contain duplicated elements.
2170 ** RESTART-CASE understands local macros.
2171 ** RESTART-CASE associates exactly its own restarts with a condition.
2172 ** ENDP in safe mode checks its argument to be of type LIST.
2173 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2174 in question is unbound.
2175 ** Optimized MAKE-INSTANCE functions no longer cause internal
2176 assertion failures in the presence of duplicate initargs.
2177 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2178 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2179 propagated even in the ctor optimized implementation of
2181 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2183 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2184 * a better implementation of SXHASH on (simple) bit vectors,
2185 measured both in execution speed and in distribution of results
2186 over the positive fixnums, has been installed. Likewise, a better
2187 implementation of EQUAL for simple bit vectors is now available.
2188 * fixed CEILING optimization for a divisor of form 2^k.
2189 * fixed bug 240 (emitting extra style warnings "using the lexical
2190 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2191 by Antonio Martinez)
2192 * fixed SXHASH, giving different results for NIL depending on type
2193 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2194 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2195 variable. (found by Rolf Wester)
2196 * SBCL does not ignore type declarations for special
2197 variables. (reported by rif on c.l.l 2003-03-05)
2198 * some bug fixes in contrib/sb-aclrepl/
2199 * fixed some bugs revealed by Paul Dietz' test suite:
2200 ** a bug in the CONS type specifier, whereby the CAR and CDR
2201 types got intertwined, has been fixed;
2202 ** the type system is now able to reason about the interaction
2203 between INTEGER and RATIO types more completely;
2204 ** APPEND, [N]REVERSE and NRECONC check that those their
2205 arguments, which must be proper lists, are really so;
2206 ** An array specialized to be unable to hold elements has been
2207 implemented, as required -- yes, really -- by ANSI;
2208 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2209 on malformed property lists;
2211 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2212 * incompatible packaging change: in line with Unix convention,
2213 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2214 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2215 * REQUIRE and PROVIDE are now optionally capable of doing something
2216 useful. See the documentation string for REQUIRE.
2217 * infrastructure for a managed SBCL contrib system: contributed
2218 modules in this release include:
2219 ** the ASDF system definition facility;
2220 ** an interface to the BSD Sockets API;
2221 ** an ACL-like convenience interface to the repl;
2222 (thanks to Kevin Rosenberg)
2223 ** an implementation of ROTATE-BYTE, with efficient implementation
2225 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2226 gives a closer approximation to the right answer than previously.
2227 (thanks to Raymond Toy)
2228 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2229 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2230 optional environment arguments, as required by ANSI.
2231 * fixed bugs in other functions taking environment objects, allowing
2232 calls with an explicit NIL environment argument to be compiled
2234 * fixed bug 228: primary return values from
2235 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2236 COMPILE or FUNCTION.
2237 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2238 :TYPE will no longer signal errors on innocuous objects.
2239 * fixed bug 231b: SETQ is better at respecting type declarations in
2240 the lexical environment.
2241 * fixed a bug in DEFCLASS: classes named by symbols with no or
2242 unprintable packages can now be defined.
2243 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2244 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2245 * fixed compiler failure related to checking types of functions.
2246 (reported by Robert E. Brown)
2247 * the compiler is now much more consistent in its error-checking
2248 treatment of bounding index arguments to sequence functions: in
2249 (SAFETY 3) code, errors will be signalled in almost all cases if
2250 invalid sequence bounding indices are passed to functions defined
2251 by ANSI to operate on sequences.
2252 * fixed a bug in the build procedure: documentation of SBCL-specific
2253 packages is now preserved and available in the final Lisp image.
2254 * lifted FDEFINITION lookup out of loops in the implementation of
2255 many list operations. (thanks to Robert E. Brown)
2256 * fixed a bug in the reader: the #n# reader macro now works for
2257 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2258 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2259 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2260 E. Brown and Tony Martinez respectively)
2261 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2262 SLOT-BOUNDP: in particular, it is now able to optimize them much
2263 better, and is now not vulnerable to having packages renamed.
2264 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2265 always called when a slot is not present in an instance. (thanks
2267 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2268 longer signal an error on unreading a character following EOF on
2269 the previous constituent stream. (thanks to Tony Martinez)
2270 * fixed some bugs revealed by Paul Dietz' test suite:
2271 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2272 not just nonnegative fixnums;
2273 ** the logical bit-array operators such as BIT-AND now accept an
2274 explicit NIL for their "opt-arg" argument (to indicate a
2275 freshly-consed result bit-array);
2276 ** ELT now signals an error on an invalid sequence index in safe
2278 ** the type system is now cleverer about negations of numeric
2279 types, and consequently understands the BIGNUM and RATIO types
2281 ** the type system is now cleverer about the interaction between
2282 INTEGER and RATIO types: while bugs still remain, many more
2283 cases are accurately computed;
2284 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2285 if it is in the last clause;
2286 ** CONSTANTLY now correctly returns a side-effect-free function in
2288 ** DECLARE is no longer treated as a special-operator; in
2289 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2290 * incremented fasl file version number due to the change in the
2291 DEFSTRUCT-SLOT-DESCRIPTION structure.
2293 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2294 * minor incompatible change: code processed by the "interpreter" or
2295 EVAL now has a compilation optimization policy of (DEBUG 2)
2296 (changed from (DEBUG 1)) to improve debuggability of interactive
2297 development, and to allow the use of the debug RETURN command in
2299 * an experimental implementation of the RETURN command for the
2300 debugger has been included. (thanks to Frederik Kuivinen)
2301 * fixed bug 62: constraints were not propagated into a loop.
2302 * fixed bug in embedded calls of SORT (reported and investigated by
2304 * fixed some bugs revealed by Paul Dietz' test suite:
2305 ** printing and reading of arrays with some dimensions having
2306 length 0 (thanks to Gerd Moellmann);
2307 ** BOA constructor with &AUX argument without a default value does
2308 not cause a type error;
2309 ** CONSTANTP now returns true for all self-evaluating objects.
2311 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2312 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2313 accessors that are related by inheritance, as specified in the
2314 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2315 Valtteri Vuorikoski)
2316 * The compiler is now able to inline functions that were defined in
2317 a complex lexical environment (e.g. inside a MACROLET).
2318 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2320 * Support for the upcoming FreeBSD-5.0 release has been included.
2321 (thanks to Dag-Erling Smorgrav)
2322 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2323 effect when it is not in a toplevel context.
2324 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2325 stand a better chance of being correct. (thanks to Gerd
2327 * fixed bug in COERCE, which now signals an error on coercing a
2328 rational to a bounded real type which excludes the expected
2330 * The compiler is now able to derive types more accurately from the
2331 COERCE and COMPILE functions.
2332 * fixed bug 223: functional binding is considered to be constant
2333 only for symbols in the CL package.
2334 * fixed bug 231: SETQ did not check the type of a variable being set
2335 (reported by Robert E. Brown)
2336 * A new optimization for MAKE-INSTANCE has been included, fixing
2337 various bugs (including relating to :ALLOCATION :CLASS slots and
2338 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2340 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2341 ** As required by ANSI, LOOP now disallows anonymous collection
2342 clauses such as COLLECT I in conjunction with aggregate boolean
2343 clauses such as THEREIS (= I 1);
2344 ** LOOP now signals an error when any variable is reused in the
2345 same loop (including the potentially useful construct analogous
2346 to WITH A = 1 WITH A = (1+ A);
2347 ** IT is only a special loop symbol within the first clause of a
2348 conditional loop clause;
2349 ** LOOP with a typed iteration variable over a hashtable now
2350 signals a type error iff it should.
2351 * fixed some other bugs revealed by Paul Dietz' test suite:
2352 ** FILE-STREAM now names the class previously known as FD-STREAM;
2353 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2354 argument) no longer signals an error;
2355 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2356 of the slot symbol, rather than using the current package
2357 ((:CONC-NAME "") continues to intern the slot's name in the
2359 * incremented fasl file version number, because of the incompatible
2360 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2361 of the new implementation of DEFINE-COMPILER-MACRO.
2363 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2364 * Support for building SBCL for MIPS platforms running in
2365 little-endian mode has now been checked in, and basic
2366 functionality on said platforms verified.
2367 * minor incompatible change: PCL now records the pathname of a file
2368 in which methods and the like are defined, rather than its
2370 * minor incompatible change: TRUENAME now considers the truename of
2371 a file naming a directory to be the pathname with :DIRECTORY
2372 component indicating that directory.
2373 * minor incompatible change: a NAMED clause in the extended form of
2374 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2375 reason for the previous behaviour is unclear.
2376 * more systematization and improvement of CLOS and MOP conformance
2377 in PCL (thanks to Gerd Moellman and Pierre Mai):
2378 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2380 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2381 primary methods with no specializers;
2382 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2384 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2385 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2386 FORWARD-REFERENCED-CLASSes; error reporting on
2387 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2389 ** SXHASH on CLOS instances now uses a slot internal to the
2390 instance to return different numbers on distinct instances,
2391 while preserving the same return value through invocations of
2393 ** DEFMETHOD signals errors when methods with longer incongruent
2394 lambda lists are added to generic functions;
2395 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2396 CLASS, as specified in AMOP;
2397 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2398 on the order returned by the primary method for classes of
2399 class STANDARD-CLASS;
2400 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2401 * fixed some bugs shown by Paul Dietz' test suite:
2402 ** DOLIST puts its body in TAGBODY;
2403 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2405 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2406 value producing form;
2407 ** if more variables are given to PROGV than values, extra
2408 variables are bound and made to have no value;
2409 ** NSUBSTITUTE on list arguments gets the right answer with
2411 ** ELT signals an error of type TYPE-ERROR when the index argument
2412 is not a valid sequence index;
2413 ** LOOP signals (at macroexpansion time) an error of type
2414 PROGRAM-ERROR when duplicate variable names are found;
2415 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2416 ** FUNCALL of special-operators now cause an error of type
2418 ** PSETQ now works as required in the presence of side-effecting
2419 symbol-macro places;
2420 ** NCONC accepts any object as its last argument;
2421 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2423 ** loop-for-as-package does not require a package to be explicitely
2425 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2427 * fixed bug 166: compiler preserves "there is a way to go"
2428 invariant when deleting code.
2429 * fixed bug 172: macro lambda lists with required arguments after
2430 &REST arguments now cause an error to be signalled. (thanks to
2432 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2433 bug. (thanks to Matthew Danish)
2434 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2436 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2437 arguments are lexically rebound. (thanks to Gerd Moellmann and
2439 * fixed bug 194: error messages are now more informative when there
2440 is no primary method applicable in a call to a generic
2441 function. (thanks to Gerd Moellmann)
2442 * fixed bug in command line argument checking (thanks to Julian
2444 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2445 (thanks to Matthew Danish)
2446 * incremented fasl file version number, because of the
2447 SXHASH-related changes in the layout of CLOS data structures
2449 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2450 * minor incompatible change: The runtime (the Unix executable named
2451 "sbcl") is now much pickier about the .core files it will load.
2452 Essentially it now requires .core files to descend from the same
2453 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2454 as the runtime does. (The intent is to prevent the crashes which
2455 can occur, and which can even be reported as mysterious failures,
2456 when people patch the sources or change the build parameters
2457 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2458 sbcl and .core files.)
2459 * fixed bug: VALUES-LIST is no longer optimized away.
2460 * fixed bug 142: The FFI conversion of C string values to Lisp
2461 string values no longer conses excessively. (thanks to Nathan
2462 Froyd porting Raymond Toy's fix to CMU CL)
2463 * began to systematize and improve MOP conformance in PCL (thanks to
2464 Nathan Froyd, Gerd Moellman and Pierre Mai):
2465 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2467 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2468 ** FINALIZE-INHERITANCE is now called on class finalization;
2469 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2470 argument precedence order.
2471 * fixed bug 202: The compiler no longer fails on functions whose
2472 derived types contradict their declared type.
2473 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2474 so it can be non-toplevel.
2475 * The fasl file version number has changed (because of the new
2476 implementation of DEFMACRO).
2477 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2478 safe code, that any length requirement by their type-specifier
2479 argument is valid. The exceptions to this are described in bug
2481 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2482 functions now check the types of their inputs as required by ANSI.
2483 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2484 introduced symbol is DECLAREd to be SPECIAL.
2485 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2486 * fixed bug: :COUNT argument to sequence functions may be negative
2487 * fixed bug: body of DO-SYMBOLS may contain declarations
2488 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2489 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2490 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2491 symbol macro only once
2492 * fixed printing of call frame when argument list is unavailable
2493 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2494 * compiler no longer signals WARNING on unknown keyword
2497 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2498 * A beta-quality port to the mips architecture running Linux,
2499 based on the old CMUCL backend, has been made. It has been tested
2500 on a big-endian kernel, and works sufficiently well to be able to
2501 rebuild itself; it has not been tested in little-endian mode.
2502 * fixed an inconsistency between gencgc.c and purify.c which made
2503 dumping/loading .core files unreliable
2504 * fixed bug 120a: The compiler now deals correctly with IFs where
2505 the consequent is the same as the alternative, instead of
2506 misderiving the return type. (thanks to Alexey Dejneka)
2507 * fixed bug 113: Logical pathnames are now dumpable (the logical
2508 host is resolved at load-time, throwing an error if it is not
2510 * fixed bug 174: FORMAT's error message is slightly clearer when a
2511 non-printing character is used in a format directive.
2512 * fixed several bugs in compiler checking of type declarations, i.e.
2513 violations of the Python "declarations are assertions" principle
2514 (thanks to Alexey Dejneka)
2515 * fixed several bugs in PCL's error checking (thanks to Gerd
2517 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2519 * fixed bug in compilation of functions as first class values
2520 (thanks to Antonio Martinez-Shotton)
2521 * The compiler's handling TYPE-ERRORs which it can prove will
2522 inevitably happen at runtime has been cleaned up and corrected
2523 in several ways. (thanks to Alexey Dejneka)
2524 * improved argument type checking for various basic arithmetic
2525 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2526 lavished on them in the past that they can be compiled in many
2527 ways in different special cases
2528 * fixed bug 181: compiler checks validity of user supplied type
2530 * cleaned up code flushing in optimization: Function calls which
2531 should signal errors for safety purposes (e.g. which ANSI says
2532 should signal errors when their arguments are of incorrect type)
2533 are no longer optimized away.
2534 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2535 * incremented fasl file version number, because changes in the
2536 implementation of sequence functions like COERCE caused
2537 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2538 in old inline expansions) to become undefined. (Actually these
2539 changes were later undone, so we might very well be binary
2540 compatible with 0.7.7 after all, but leaving the version number
2541 incremented seemed like the simplest and most conservative
2544 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2545 * An alpha-quality port to the parisc architecture running Linux,
2546 based on the old CMUCL backend, has been made. This, even more so
2547 than the other backends, should be considered still a work in
2548 progress; known problems include that the Linux kernel in 64-bit
2549 mode does not propagate the correct sigcontext structure to
2550 userspace, and consequently SBCL on a parisc64 kernel will not
2552 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2553 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2554 suppressed.) Also now that the compiler is looking at declarations
2555 in the environment, it checks optimization declarations as well,
2556 and suppresses inlining when (> DEBUG SPEED).
2557 * More fixes have been made to treatment of floating point exception
2558 treatment and other Unix signals. In particular, floating point
2559 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2560 * The detection and handling of control stack exhaustion (infinite
2561 or very deeply nested recursion) has changed. Stack exhaustion
2562 detection is now done by write-protecting pages at the OS level
2563 and applies at all optimization settings; when found, a
2564 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2565 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2566 be caught using IGNORE-ERRORS.
2567 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2568 correctly when transforming recursive LABELS functions to LETs.
2569 (thanks to Alexey Dejneka)
2570 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2571 that are names of constants or global variables.
2572 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2573 alien routines with docstrings.
2574 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2575 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2577 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2578 (thanks to Eric Marsden)
2579 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2580 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2581 * Bug fix: The SPARC backend can now compile functions involving
2582 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2583 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2584 to a functions expecting a PATHNAME-DESIGNATOR.
2585 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2586 lambda lists. (thanks to Alexey Dejneka)
2587 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2588 functions so that EXPORTing the name of the class doesn't cause
2589 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2590 to Antonio Martinez for reporting this)
2591 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2592 constant-folded without failing an assertion. (thanks to Einar
2593 Floystad Dorum for reporting this)
2594 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2595 types are now dealt with more correctly by the compiler.
2596 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2597 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2598 behaviour analogous to RENAME-FILE. This puts its behaviour more
2599 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2601 * The fasl file version number has changed again. (because of the
2602 bug fix involving the names of PCL MAKE-INSTANCE functions)
2604 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2605 * bug fix: Floating point exceptions are treated much more
2606 consistently on the x86/Linux and PPC/Linux platforms.
2607 * Array initialization with :INITIAL-ELEMENT is now much faster for
2608 cases when the compiler cannot open code the array creation, but
2609 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2610 array accesses have also seen a speed increase.
2611 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2612 specified. (This was at the root of some bad interactions between
2613 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2615 * bug fix: Internal error arguments for undefined functions are now
2616 computed correctly on the PPC/Linux platform.
2617 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2618 Raymond Toy's patch for CMU CL)
2619 * Support for the Solaris 9 operating environment has been included
2620 (thanks to Daniel Merritt)
2621 * A very ugly but hopefully complete draft of the missing FFI chapter
2622 of the manual has been created by reformatting the corresponding
2623 CMU CL manual chapter into (currently very ugly and incoherent)
2624 DocBook and bringing it up to date for SBCL behavior. Thus, the
2625 manual is now essentially complete, at least by my extreme
2626 once-and-only-once standards, whereby it's acceptable to refer to
2627 the doc strings of SB-EXT functions as the primary documentation.
2628 * The fasl file version number has changed again, due to cleanup of
2629 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2630 is no longer a static symbol.)
2632 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2633 * SBCL now builds with OpenMCL (version 0.12) as the
2634 cross-compilation host; also, more progress has been made toward
2635 bootstrapping under CLISP.
2636 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2638 * bug 158 fixed: The compiler can now deal with integer loop
2639 increments different from 1; fixing this turned out also to fix
2641 * bug 169 fixed: no more bogus warnings about using lexical bindings
2642 despite the presence of perfectly good SPECIAL declarations (thanks
2643 to David Lichteblau)
2644 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2645 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2646 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2648 * bug 180 fixed: Method combination specifications no longer ignore
2649 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2650 * bug fix: Structure type predicate functions now check their argument
2651 count as they should.
2652 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2653 correctly. (thanks to Pierre Mai)
2654 * minor incompatible change: The --noprogrammer option is deprecated
2655 in favor of the new --disable-debugger option, which is very similar.
2656 (The major difference is that it takes effect at a slightly different
2657 time at startup, causing handling of errors in --sysinit and
2658 --userinit files will be affected differently.) The
2659 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2660 been added to allow this functionality to be controlled from ordinary
2661 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2662 maintainers, who might want to run non-interactive scripts to
2663 build SBCL cores which will later be used interactively.)
2664 * minor incompatible change: The LOAD function no longer, when given
2665 a wild pathname to load, loads all files matching that pathname.
2666 Instead, an error of type FILE-ERROR is signalled.
2668 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2669 * bug 147 fixed: The compiler preserves its block link/count
2670 invariants more correctly now so that it doesn't crash. (thanks
2672 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2674 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2675 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2676 no-value-for-this-character as NIL (as specified by ANSI).
2677 * HOST-NAMESTRING on physical pathnames now returns a string that is
2678 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2679 (thanks to Christophe Rhodes)
2680 * The Alpha port handles icache flushing more correctly. (thanks to
2682 * More progress has been made toward bootstrapping under CLISP. (thanks
2683 to Christophe Rhodes)
2684 * The fasl file format has changed again, because dynamic loading
2685 on OpenBSD (which has non-ELF object files) motivated some cleanups
2686 in the way that foreign symbols are transformed and passed around.
2687 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2688 is no longer treated as whitespace by the reader, but instead as
2689 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2690 |A
\7fB|, instead of A as it used to.
2692 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2693 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2694 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2695 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2696 this as of 0.7.1.45, but was left out of the previous news section
2697 (thanks to Dan Barlow)
2698 * SBCL now runs on the Solaris operating system on SPARC architectures
2699 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2700 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2701 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2702 * SPARC backend cleanups, allowing builds of cores optimized for V8
2703 and V9 SPARCS, and also emission of code targeted to a particular
2704 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2706 * SBCL is closer to bootstrapping under CLISP, thanks to various
2707 fixes by Christophe Rhodes.
2708 * The fasl file format has changed again, to allow the compiler's
2709 INFO database to support symbol macros.
2710 * The user manual (in doc/) is formatted into HTML more nicely.
2711 (thanks to coreythomas)
2712 * The system is smarter about SUBTYPEP relationships, especially
2713 those involving NOT types (including types such as ATOM which are
2714 represented internally using NOT types). Thus SUBTYPEP is less
2715 likely to return (VALUES NIL NIL) in general, and in particular
2716 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2718 * The fasl file format has changed again, because the internal
2719 representation of types now includes a new slot to support the new
2720 SUBTYPEP-of-NOT-types logic.
2721 * (not a change in the main branch of SBCL, but a related prototype
2722 which can hopefully be merged into the main branch of SBCL in the
2723 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2724 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2725 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2726 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2727 dispatch macro character combinations. (thanks to Alexey Dejneka)
2728 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2729 that superficially look like logical namestrings correctly.
2730 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2732 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2733 default supplied-p) for &optional and &key arguments. (thanks to
2736 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2737 * incompatible change: The compiler is now less aggressive about
2738 tail call optimization, doing it only when (> SPACE DEBUG) or
2739 (> SPEED DEBUG). (This is an incompatible change because there are
2740 programs which relied on the old CMU-CL-style behavior to optimize
2741 away their unbounded recursion which will now die of stack overflow.)
2742 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2743 for non-GENCGC systems has been increased to 20M (since that
2744 seems much closer to the likely performance optimum for modern
2745 systems than the old 4M value was)
2746 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2747 and *DEBUG-PRINT-LEVEL*
2748 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2749 of CMU CL's support for SPARC, and various endianness and other
2750 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2751 * new syntactic sugar for the Unix command line: --load foo.bar is now
2752 an alternate notation for --eval '(load "foo.bar")'.
2754 ** The system now detects stack overflow and handles it gracefully,
2755 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2756 optimization settings. (This is a good thing in general, and
2757 its introduction in this version should be particularly timely
2758 for anyone whose code fails because of suppression of tail
2760 ** The system now hunts for the C variable "environ" in a more
2761 devious way, to avoid segfaults when the C library version differs
2762 between compile time and run time. (thanks to Christophe Rhodes)
2763 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2764 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2765 ** The compiler no longer issues bogus style warnings for undefined
2766 classes in the same source file as the DEFCLASSes which defined
2767 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2769 ** fixes in CONDITION class precedence list for undefined function
2770 errors (thanks to Alexei Dejneka)
2771 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2772 correctly. (thanks to Dan Barlow)
2773 ** portability fixes aiming at bootstrapping under CLISP (thanks
2774 to Dave McDonald and Christophe Rhodes)
2775 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2776 ** fixes in type translation and and type inference (thanks to
2778 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2779 ** various fixes in the new ports (thanks to Dan Barlow)
2780 * several changes related to debugging:
2781 ** suppression of tail recursion, as noted above
2782 ** stack overflow detection, as noted above
2783 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2784 is now the default. (For some time encapsulation has been more
2785 reliable than the breakpoint-based :ENCAPSULATE NIL
2786 implementation, at least on X86 systems; and I just noticed that
2787 encapsulation also seems closer to the spirit of the ANSI
2790 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2792 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2793 up properly again. (There was a packaging bug in 0.7.0 which
2794 left their definitions in SB-SYS::LOAD-FOREIGN and
2795 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2796 vital for most things which interface to C-level interfaces,
2797 like extensions working with sockets or databases or
2798 Perl-compatible regexes or whatever, and the need to fix
2799 this bug is the main reason that 0.7.1 was released so
2801 ** DEFGENERIC is now choosier about the methods it redefines, so that
2802 reLOADing a previously-LOADed file containing DEFGENERICs does
2803 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2804 cycle now works as it should. (thanks to Alexey Dejneka)
2805 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2806 fixed by Christophe Rhodes. (He actually submitted this patch
2807 months ago, and I delayed until after 0.7.0.)
2808 ** Bug 111 (internal compiler confusion about runtime checks on
2809 FUNCTION types) was fixed by Alexey Dejneka.
2810 * Some internal cleanups (getting rid of variables which aren't
2811 needed now that the byte interpreter is gone) caused the fasl
2812 file format number to change again.
2814 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2815 * major incompatible change: The default fasl file extension, i.e. the
2816 default extension for files produced by COMPILE-FILE, has changed
2817 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2819 ** There are many changes in the implementation of the compiler.
2820 SBCL is now essentially a compiler-only implementation of ANSI
2821 Common Lisp. EVAL still "interprets" a few special cases, but
2822 almost all the interesting cases are handled by creating
2823 a LAMBDA expression, calling COMPILE on it, then calling
2824 FUNCALL on the result.
2825 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2826 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2827 Since the code is newer, there might still be some new bugs
2828 (though not as many as before Martin Atzmueller's fixes:-). But
2829 the new code is substantially simpler and clearer, and hopefully
2830 any remaining bugs will be simpler, less fundamental, and more
2831 fixable then the bugs in the old code.
2832 ** The revised compiler is still a little unsteady on its feet.
2834 *** The debugging information it produces (particularly the names
2835 of FUNCTION objects) is sometimes much less useful than what
2836 the old compiler produced.
2837 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2838 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2839 within a MACROLET) has been temporarily weakened.
2840 ** There are new compiler optimizations for various functions:
2841 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2842 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2843 *** the math functions TRUNCATE, FLOOR, and CEILING
2844 *** the function-of-all-trades COERCE
2845 Mostly these should be transparent, but there's one
2846 potentially-annoying problem (bug 117): when the compiler
2847 inline-expands a function and does type analysis on the result,
2848 it can create control paths which have type mismatches, and
2849 when it can't prove that those control paths aren't taken,
2850 it will issue WARNINGs about the type mismatches. This is
2851 a particular problem in practice for the new sequence functions.
2852 It's not clear how this should be fixed, and for now, a
2853 workaround is given in the entry for 117 in the BUGS file.
2854 ** (Because of the interaction between the two previous items --
2855 occasional inlining problems and new inline expansions -- some
2856 of the new sequence function optimizations won't really kick in
2857 completely until debugging information, and then inlining, are
2858 straightened out in some future version.)
2859 * minor incompatible changes:
2860 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2861 DIRECTORY no longer implicitly promotes NIL slots of its
2862 pathname argument to :WILD. In particular, when you ask for the
2863 contents of a directory (which you used to be able to do without
2864 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2865 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2866 ** changes in behavior that ANSI explicitly defines to be
2867 implementation dependent:
2868 *** The new compiler-only implementation still conforms with ANSI,
2869 but acts a little different than before. Besides the obvious
2870 changes in performance tradeoffs (that the cost per form passed
2871 to EVAL has gone up, and the cost per form executed by EVAL
2872 has gone down), the behavior of the system changes a little
2873 because there are no longer any interpreted function objects.
2874 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2875 e.g. doing COMPILE on the output of interactive DEFUN is
2877 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2879 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2880 has changed from (:CL) to NIL.
2881 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2882 e.g. #\Newline and #\Tab, don't have names specified in the
2883 ANSI Common Lisp standard, is now based on their ASCII symbolic
2884 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2885 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2886 are no longer used for output.
2887 ** changes in internal implementation constants:
2888 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2889 4 million. (If your application spends a lot of time GCing and
2890 you have a lot of RAM, you might want to experiment with
2891 increasing it even more.)
2892 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2893 However, almost all old code should still continue to work without
2894 immediate update, as SB-C-CALL is now a (deprecated) nickname
2896 ** Old operator names in the style DEF-FOO are now deprecated in
2897 favor of new corresponding names DEFINE-FOO, for consistency with
2898 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2899 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2900 internal symbols, but a few supported extensions like
2901 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2902 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2903 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2904 etc. as you get deeper into recursive calls to the debugger
2905 command loop, instead of the old "5]", "5]]", "5]]]"
2906 sequence. (I was motivated to do this when squabbles between
2907 ILISP and SBCL left me very deeply nested in the debugger. In the
2908 short term, this change will probably provoke more ILISP/SBCL
2909 squabbles, but hopefully it will be an improvement in the long run.)
2910 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2911 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2912 function, since declaiming return types involving aliens is
2913 (1) annoyingly messy to do by hand and (2) vital to efficient
2914 compilation of code which calls such functions.
2915 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2916 longer reexported by the SB-EXT package. They're solely useful
2917 for alien code, so it seems more logical that you should get
2918 them from the SB-ALIEN package, not in SB-EXT.
2919 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2920 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2921 features. Instead, the code that they used to control is always
2922 built into the system.
2923 * many other bug fixes
2924 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2925 advantage of the new EVAL-WHEN stuff and to clean them up in
2926 general, and they are now more ANSI-compliant in a number of
2927 ways. Martin Atzmueller is responsible for a lot of this.
2928 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2930 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2931 *** correct ERROR type for various file operations
2932 *** some fixes for Lisp streams
2933 *** DEFMETHOD syntax checking
2934 *** changing old weird representation of debug information as
2935 strings (which, among their other deficiencies, don't transform
2936 correctly when you rename packages, and don't change their
2937 print representation when you change things like *PACKAGE*
2938 and *PRINT-LENGTH*) to symbols and lists of symbols
2939 He also made several improvements and fixed several bugs in DESCRIBE.
2940 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2942 *** misbehavior of WRITE-STRING/WRITE-LINE
2943 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2944 and several other LOOP problems as well
2945 *** DIRECTORY when similar filenames are present
2946 *** DEFGENERIC with :METHOD options
2947 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2948 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2949 *** argument ordering in FIND with :TEST option
2950 *** mishandled package designator argument in APROPOS-LIST
2951 *** various problems in the backquote readmacro
2952 *** a bug in APROPOS
2953 *** probably some others that I'm not describing very well here,
2954 since the CVS log documents them by reference to sbcl-devel
2955 messages, and the SourceForge archives aren't working well.:-(
2956 ** Dan Barlow improved the Alpha port (and is making progress on the
2957 PPC port, for those of you who think different).
2958 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2959 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2960 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2961 continues to work on the SPARC port (for those of you in a position
2962 to look down upon our little PC-compatible boxes from a great height).
2963 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2964 and uses that copy, rather than the current dynamic value, when
2965 it's trying to decide whether to truncate output. Thus e.g.
2966 (let ((*print-lines* 50))
2967 (pprint-logical-block (stream nil)
2969 (let ((*print-lines* 8))
2970 (print (aref possiblybigthings i) stream)))))
2971 should now truncate the logical block only at 50 lines, instead of
2972 often truncating it at 8 lines, as it did before.
2973 * The doc/cmucl/ directory, containing old CMU CL documentation
2974 from the time of the fork, is no longer part of the base system.
2975 SourceForge has shut down its anonymous FTP service, and with it
2976 my original plan for distributing the old CMU CL documentation
2977 there. For now, if you need these files you can download an old
2978 SBCL source release and extract them from it.
2979 * The fasl file version number changed again, for dozens of reasons,
2980 some of which are apparent above.
2982 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2983 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2984 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2985 patch, so that DISASSEMBLE output is much nicer.
2986 * The code in the SB-PROFILE package now seems reasonably stable.
2987 I still haven't decided what the final interface should look like
2988 (I'd like PROFILE to interact cleanly with TRACE, since both
2989 facilities use function encapsulation) but if you have a need
2990 for profiling now, you can probably use it successfully with
2991 the current CMU-CL-style interface.
2992 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2993 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2994 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2995 link now returns the dangling link itself, and for similar
2996 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2997 link itself. (In these cases the old code signalled an error and
2998 looped endlessly, respectively.) Thus, DIRECTORY now works even
2999 in the presence of dangling and cyclic symbolic links.
3000 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3001 is now a supported extension again, since the consensus on
3002 sbcl-devel was that it can be useful for ordinary development
3003 work, not just for debugging SBCL itself.
3004 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3005 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3006 that currently-defined functions might be redefined later with
3007 different return types.
3008 * Hash tables can be printed readably, as inspired by CMU CL code
3009 of Eric Marsden and SBCL code of Martin Atzmueller.
3010 * better error handling in CLOS method combination, thanks to
3011 Martin Atzmueller porting Pierre Mai's CMU CL patches
3012 * more overflow fixes for >16Mbyte I/O buffers
3013 * A bug in READ has been fixed, so that now a single Ctrl-D
3014 character suffices to cause end-of-file on character streams.
3015 In particular, now you only need one Ctrl-D at the command
3016 line (not two) to exit SBCL.
3017 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3019 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3020 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3021 does the right thing.
3022 * The implementation of some type tests, especially for CONDITION
3023 types, is now tidier and maybe faster, due to CMU CL code
3024 originally by Douglas Crosher, ported by Martin Atzmueller.
3025 * Some math functions have been fixed, and there are new
3026 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3027 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3028 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3029 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3030 likely to be built into sbcl-0.7.0.)
3031 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3032 CL, ported to SBCL by Martin Atzmueller)
3033 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3034 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3035 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3036 * There's a new slam.sh hack to shorten the edit/compile/debug
3037 cycle for low-level changes to SBCL itself, and a new
3038 :SB-AFTER-XC-CORE target feature to control the generation of
3039 the after-xc.core file needed by slam.sh.
3040 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3041 COMPILE-FILE is no longer supported, so that now every function
3042 gets an entry point, so that block compilation looks a little
3043 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3044 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3045 returns the number of bytes consed since the system started,
3046 rather than the number consed since the first time the function
3047 was called. (The new definition parallels ANSI functions like
3048 CL:GET-INTERNAL-RUN-TIME.)
3049 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3050 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3051 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3052 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3053 * Fasl file version numbers are now independent of the target CPU,
3054 since historically most system changes which required version
3055 number changes have affected all CPUs equally. Similarly,
3056 the byte fasl file version is now equal to the ordinary
3059 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3060 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3061 is no longer recognized. I apologize for this, because it was
3062 listed in SB-EXT as a supported extension, but I found that
3063 its existing behavior was poorly specified, as well as incorrectly
3064 specified, and it looked like too much of a mess to straighten it
3065 out. I have enough on my hands trying to get ANSI stuff to work..
3066 * many patches ported from CMU CL by Martin Atzmueller, with
3067 half a dozen bug fixes in pretty-printing and the debugger, and
3068 half a dozen others elsewhere
3069 * fixed bug 13: Floating point infinities are now supported again.
3070 They might still be a little bit flaky, but thanks to bug reports
3071 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3072 as flaky as they were.
3073 * The --noprogrammer command line option is now supported. (Its
3074 behavior is slightly different in detail from what the old man
3075 page claimed it would do, but it's still appropriate under the
3076 same circumstances that the man page talks about.)
3077 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3078 are now supported, and enabled by default. Thus, the compiler can
3079 handle many floating point and complex operations much less
3080 inefficiently. (Thus e.g. you can implement a complex FFT
3082 * The compiler now detects type mismatches between DECLAIM FTYPE
3083 and DEFUN better, and implements CHECK-TYPE more correctly, and
3084 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3085 thanks to patches from Martin Atzmueller.
3086 * various fixes to make the cross-compiler more portable to
3087 ANSI-conforming-but-different cross-compilation hosts (notably
3088 Lispworks for Windows, following bug reports from Arthur Lemmens)
3089 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3090 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3091 fixed thanks to Pierre Mai's CMU CL patches.
3092 * fixes to keep the system from overflowing internal counters when
3093 it tries to use i/o buffers larger than 16M bytes
3094 * fixed bug 45a: Various internal functions required to support
3095 complex special functions have been merged from CMU CL sources.
3096 (When I was first setting up SBCL, I misunderstood a compile-time
3097 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3098 * improved support for type intersection and union, fixing bug 12
3099 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3100 more obscure bugs as well
3101 * some steps toward byte-compiling non-performance-critical
3102 parts of the system, courtesy of patches from Martin Atzmueller
3103 * Christophe Rhodes has made some debian packages of sbcl at
3104 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3105 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3106 stable, but are nonetheless usable. When he's ready, I'd be happy
3107 to add them to the SourceForge "File Releases" section. (And if
3108 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3109 * new fasl file format version number (because of changes in
3110 internal representation of (OR ..) types to accommodate the new
3111 support for (AND ..) types, among other things)
3113 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3114 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3116 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3118 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3119 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3120 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3121 * DESCRIBE now works on structure objects again.
3122 * Most function call argument type mismatches are now handled as
3123 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3124 know whether the function will be redefined before the call is
3125 executed. (The compiler could flag local calls with full WARNINGs,
3126 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3127 it doesn't keep track of enough information to know whether calls
3128 are local in this sense.)
3129 * Compiler output is now more verbose, with messages truncated
3130 later than before. (There should be some supported way for users
3131 to override the default verbosity, but I haven't decided how to
3132 provide it yet, so this behavior is still controlled by the internal
3133 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3134 src/compiler/ir1util.lisp.)
3135 * Fasl file format version numbers have increased again, because
3136 support for the Gray streams extension changes the layout of the
3137 system's STREAM objects.
3138 * The Gray subclassable streams extension now works, thanks to a
3139 patch from Martin Atzmueller.
3140 * The full LOAD-FOREIGN extension (not just the primitive
3141 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3142 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3143 but like most other programs, it defaults to copying the Unix
3144 environment from the original process instead of starting the
3145 new process in an empty environment.
3146 * Extensions which manipulate the Unix environment now support
3147 an :ENVIRONMENT keyword option which doesn't smash case or
3148 do other bad things. The CMU-CL-style :ENV option is retained
3149 for porting convenience.
3150 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3151 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3153 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3155 * A patch from Martin Atzmueller seems to have solved the SIGINT
3156 problem, and as far as we know, signal-handling now works cleanly.
3157 (If you find any new bugs, please report them!)
3158 * The system no longer defaults Lisp source file names to types
3159 ".l", ".cl", or ".lsp", but only to ".lisp".
3160 * The compiler no longer uses special default file extensions for
3161 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3162 seems to expect a single default extension for all compiled code,
3163 and there's no compelling reason to try to stretch the standard
3164 to allow two different extensions.) Instead, byte-compiled files
3165 default to the same extension as native-compiled files.
3166 * Fasl file format version numbers have increased again, because
3167 a rearrangement of internal implementation packages made some
3168 dumped symbols in old fasl files unreadable in new cores.
3169 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3170 many fewer weird special cases.
3171 * Bug #17 (differing COMPILE-FILE behavior between logical and
3172 physical pathnames) has been fixed, and some related misbehavior too,
3173 thanks to a patch from Martin Atzmueller.
3174 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3175 by Tim Moore, ported to SBCL by Martin Atzmueller.
3176 * Martin Atzmueller fixed several filesystem-related problems,
3177 including bug #36, in part by porting CMU CL patches, which were
3178 written in part by Paul Werkowski.
3179 * More compiler warnings in src/runtime/ are gone, thanks to
3180 more patches from Martin Atzmueller.
3181 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3184 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3186 * DESCRIBE now works on CONDITION objects.
3187 * The debugger now handles errors which arise when trying to print
3188 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3190 * The build system now uses an additional file, customize-target-features.lisp,
3191 to allow local modifications to the target *FEATURES* list. (The point of
3192 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3193 debugging features enabled, without having to worry about propagating it
3194 into everyone's system when I do a "cvs update".) When no
3195 customize-target-features.lisp file exists, the target *FEATURES* list
3196 should be constructed the same way as before.
3197 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3198 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3199 warn about behavior which is undefined under the ANSI spec. Note: This
3200 is specified by ANSI, but it's not very popular with programmers.
3201 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3202 macro in the SBCL sources for an example of a workaround which you
3203 might use to make portable ANSI-standard code which does what you want.
3204 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3205 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3206 when it's used as a non-toplevel form. (This is required in order
3207 to implement the DEFCONSTANT-EQX macro.)
3208 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3209 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3210 funny things, due to bugs in the implementation of EVAL-WHEN
3211 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3212 non-toplevel DEFCONSTANTs are uncommon.)
3213 * The core file version number and fasl file version number have been
3214 incremented, because the old noncompliant DEFCONSTANT behavior involved
3215 calling functions which no longer exist, and because I also took the
3216 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3217 * fixed bug 1 (error handling before read-eval-print loop starts), and
3218 redid debugger restarts and related debugger commands somewhat while
3220 ** The QUIT debugger command is gone, since it did something
3221 rather different than the SB-EXT:QUIT command, and since it never
3222 worked properly outside the main toplevel read/eval/print loop.
3223 Invoking the new TOPLEVEL restart provides the same functionality.
3224 ** The GO debugger command is also gone, since you can just invoke
3225 the CONTINUE restart directly instead.
3226 ** The TOP debugger command is also gone, since it's redundant with the
3227 FRAME 0 command, and since it interfered with abbreviations for the
3229 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3231 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3232 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3233 some porting work by Martin Atzmueller)
3234 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3235 with a boatload of patches which he ported to SBCL. Now that those
3237 ** The system tries to make sure that its low-priority messages
3238 are prefixed by semicolons, to help people who like to use
3239 syntax highlighting in their ILISP buffer. (This patch
3240 was originally due to Raymond Toy.)
3241 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3242 patches originally written by Raymond Toy.
3243 ** The compiler understands coercion between single-value and
3244 multiple-VALUES type expressions better, getting rid of some very
3245 weird behavior, thanks to patches originally by Robert MacLachlan
3246 and Douglas Crosher.
3247 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3248 lambda lists, thanks to a patch originally by Pierre Mai.
3249 ** The system no longer bogusly warns about "abbreviated type
3251 ** The compiler gets less confused by inlining and RETURN-FROM,
3252 thanks to some patches originally by Tim Moore.
3253 ** The system no longer hangs when dumping circular lists to fasl
3254 files, thanks to a patch originally from Douglas Crosher.
3255 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3256 undefined function error.
3257 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3258 in structure slot initforms) for now, documented workaround instead:-|
3259 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3260 * fixed bug 5: added stubs for various Gray stream functions called
3261 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3262 installed, at least appropriate type errors are generated
3263 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3264 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3265 consistently in DEFMETHOD forms.
3266 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3267 it doesn't seem to affect SBCL after all
3268 * The C runtime system now builds with better optimization and many
3269 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3271 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3273 * The system is now under CVS at SourceForge (instead of the
3274 CVS repository on my home machine).
3275 * The new signal handling code has been tweaked to treat register
3276 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3277 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3278 downstream, e.g. in debug-int.lisp, has implicit dependencies
3279 on the unsignedness of integer representation of machine words,
3280 and that caused the system to bomb out with infinite regress
3281 when trying to recover from type errors involving signed values,
3282 e.g. (BUTLAST '(1 2 3) -1).)
3283 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3284 (This was one of the bugs Peter Van Eynde reported back in July.)
3285 * The system now uses code inspired by Colin Walters' O(N)
3286 implementation of MAP (from the cmucl-imp@cons.org mailing
3287 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3288 inline the MAP operation, and there is more than one
3289 sequence argument to the MAP call (so that it can't just
3290 do ETYPECASE once and for all based on the type of the
3291 single sequence argument). (The old non-inline implementation
3292 of the general M-argument sequence-of-length-N case required
3293 O(M*N*N) time when any of the sequence arguments were LISTs.)
3294 * The QUIT :UNIX-CODE keyword argument has been renamed to
3295 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3296 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3298 * Raymond Wiker's patches to port dynamic loading from Linux to
3299 FreeBSD have been added.
3300 * The BUGS file is now more nearly up to date, thanks in large part
3301 to Martin Atzmueller's review of it.
3302 * The debugger now flushes standard output streams before it begins
3303 its output ("debugger invoked" and so forth).
3304 * The core version number and fasl file version number have both
3305 been incremented, because of incompatible changes in the layout
3307 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3308 used to not be called for a saved Lisp image.)
3309 * Martin Atzmueller's version of a patch to fix a compiler crash,
3310 as posted on sbcl-devel 13 September 2000, has been installed.
3311 * Instead of installing Martin Atzmueller's patch for the
3312 compiler transform for SUBSEQ, I deleted the compiler transform,
3313 and transforms for some similar consing operations.
3314 * A bug in signal handling which kept TRACE from working on OpenBSD
3316 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3317 away by constant folding
3318 * The system now defines its address space constants in one place
3319 (in the Lisp sources), and propagates them automatically elsewhere
3320 (through GENESIS and the sbcl.h file). Therefore, patching the
3321 address map is less unnecessarily tedious and error-prone. The
3322 Lisp names of address space constants have also been systematized.
3323 * CVS tags like dollar-Header-dollar have been removed from
3324 the sources, because they have never saved me trouble and
3325 they've been source of trouble working with patches and other
3326 diff-related operations.
3327 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3328 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3330 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3332 * The system has been ported to OpenBSD.
3333 * The system now compiles with a simple "sh make.sh" on the systems
3334 that it's supported on. I.e., now you no longer need to tweak
3335 text in the target-features.lisp-expr and symlinks in src/runtime/
3336 by hand, the make.sh takes care of it for you.
3337 * The system is no longer so grossly inefficient when compiling code
3338 involving vectors implemented as general (not simple) vectors (VECTOR T),
3339 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3340 which dares to use the various sequence functions on non-simple
3341 vectors, takes less of a performance hit.
3342 * There is now a primitive type predicate VECTOR-T-P
3343 to test for the (VECTOR T) type, so that e.g.
3344 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3345 can now be compiled with some semblance of efficiency. (The old code
3346 turned the type declaration into a full call to %TYPEP at runtime!)
3347 * AREF on (VECTOR T) is still not fast, since it's still compiled
3348 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3349 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3350 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3351 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3352 without making full calls to SUBTYPEP at runtime.
3353 (Some analogous efficiency issues for non-simple vectors specialized to
3354 element types other than T, or for non-simple multidimensional arrays,
3355 have not been addressed. They could almost certainly be handled the
3356 same way if anyone is motivated to do so.)
3357 * The changes in array handling break binary compatibility, so
3358 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3359 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3360 instead of (VALUES T T).
3361 * By following the instructions that Dan Barlow posted to sbcl-devel
3362 on 2 July 2000, I was able to enable primitive dynamic object
3363 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3364 functionality is not implemented (since it calls ld to resolve
3365 library references automatically, requiring RUN-PROGRAM for its
3366 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3367 try to resolve library references) is now supported.
3368 * The system now flushes the standard output streams when it terminates,
3369 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3370 them at several other probably-convenient times, e.g. in each pass of
3371 the toplevel read-eval-print loop, and after evaluating a form given
3372 as an "--eval" command-line option. (These changes were motivated by a
3373 discussion of stream flushing issues on cmucl-imp in August 2000.)
3374 * The source transform for TYPEP of array types no longer assumes
3375 that an array whose element type is a not-yet-defined type
3376 is implemented as an array of T, but instead punts, so that the
3377 type will be interpreted at runtime.
3378 * There is now some support for cross-compiling in make.sh: each of
3379 the phases of make.sh has its own script. (This should be transparent
3380 to people doing ordinary, non-cross-compile builds.)
3381 * Since my laptop doesn't have hundreds of megabytes of memory like
3382 my desktop machine, I became more motivated to do some items on
3383 my to-do list in order to reduce the size of the system a little:
3384 ** Arrange for various needed-only-at-cold-init things to be
3385 uninterned after cold init. To support this, those things have
3386 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3387 symbols with such names are now uninterned after cold init).
3388 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3389 fasl files for cold load.
3390 ** Remove the old compiler structure pooling code (which used to
3391 be conditional on the target feature :SB-ALLOC) completely.
3392 ** Redo the representation of some data in cold init to be more compact.
3393 (I also looked into supporting byte compiled code at bootstrap time,
3394 which would probably reduce the size of the system a lot, but that
3395 looked too complicated, so I punted for now.)
3396 * The maximum signal nesting depth in the src/runtime/ support code has
3397 been reduced from 4096 to 256. (I don't know any reason for the very
3398 large old value. If the new smaller value turns out to break something,
3399 I'll probably just bump it back up.)
3400 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3402 * Many, many bugs reported by Peter Van Eynde have been added to
3403 the BUGS list; some have even been fixed.
3404 * While enabling dynamic object file loading, I tried to make the
3405 code easier to understand, renaming various functions and variables
3406 with less ambiguous names, and changing some function calling
3407 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3408 * While trying to figure out how to do the OpenBSD port, I tried to
3409 clean up some of the code in src/runtime/. In particular, I dropped
3410 support for non-POSIX signal handling, added various comments,
3411 tweaked the code to reduce the number of compilation warnings, and
3412 renamed some files to increase consistency.
3413 * To support the new automatic configuration functionality in make.sh,
3414 the source file target-features.lisp-expr has been replaced with the
3415 source file base-target-features.lisp-expr and the machine-generated
3416 file local-target-features.lisp-expr.
3417 * fixed a stupid quoting error in make.sh so that using CMU CL
3418 "lisp -batch" as cross-compilation host works again
3420 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3422 * DESCRIBE no longer tries to call itself recursively to describe
3423 bound/fbound values, so that it no longer fails on symbols which are
3424 bound to themselves (like keywords, T, and NIL).
3425 * DESCRIBE now works on generic functions.
3426 * The printer now prints less-screwed-up representations of closures
3427 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3428 * A private symbol is used instead of the :EMPTY keyword previously
3429 used to mark empty slots in hash tables. Thus
3430 (DEFVAR *HT* (MAKE-HASH-TABLE))
3431 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3432 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3433 now does what ANSI says that it should. (You can still get
3434 similar noncompliant behavior if bang on the hash table
3435 implementation with all the symbols you get back from
3436 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3437 This breaks binary compatibility, since tests for equality to
3438 :EMPTY are wired into things like the macroexpansion of
3439 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3441 * There's now a minimal placeholder implementation for CL:STEP,
3442 as required by ANSI.
3443 * An obscure bug in the interaction of the normal compiler, the byte
3444 compiler, inlining, and structure predicates has been patched
3445 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3446 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3448 * Missing ordinary arguments in a macro call are now detected even
3449 when the macro lambda list contains &KEY or &REST.
3450 * The debugger no longer complains about encountering the top of the
3451 stack when you type "FRAME 0" to explicitly instruct it to go to
3452 the top of the stack. And it now prints the frame you request even
3453 if it's the current frame (instead of saying "You are here.").
3454 * As specified by ANSI, the system now always prints keywords
3455 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3457 * The default initial SIZE of HASH-TABLEs is now smaller.
3458 * Type information from CLOS class dispatch is now propagated
3459 into DEFMETHOD bodies, so that e.g.
3460 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3462 is now basically equivalent to
3463 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3464 (DECLARE (TYPE SINGLE-FLOAT X))
3466 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3467 operation, without having to do run-time type dispatch.
3468 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3469 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3470 * Since I don't seem to be making big file reorganizations very often
3471 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3472 is overflowing my ability to conveniently back them up), I've finally
3473 checked the system into CVS. (The CVS repository is on my home system,
3474 not at SourceForge -- putting it on SourceForge might come later.)
3475 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3476 high-level GC-NOTIFY-FOO functions send their output. (There's
3477 still very little control of where low-level verbose GC functions
3478 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3479 less than it used to -- the GC-NOTIFY-FOO functions are now under
3480 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3481 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3482 in only one place in the source code, and propagates it automatically
3483 everywhere that it's needed. Thus e.g. when I bump the version from
3484 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3485 * The C source files now include boilerplate legalese and documentation
3486 at the head of each file (just as the Lisp source files already did).
3487 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3488 to his page will be replaced with a link to his new CLiki service.
3490 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3492 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3493 * The build process now looks for GNU make under the default name "gmake",
3494 instead of "make" as it used to. If GNU make is not available as "gmake"
3495 on your system, you can change this default behavior by setting the
3496 GNUMAKE environment variable.
3497 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3498 can build without error under CMU CL.
3500 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3502 * There is now a partial SBCL user manual (with some new text and some
3503 text cribbed from the CMU CL manual).
3504 * The beginnings of a profiler have been added (starting with the
3505 CMU CL profiler and simplifying and cleaning up). Eventually the
3506 main interface should be through the TRACE macro, but for now,
3507 it's still accessed through vaguely CMU-CL-style functions and macros
3508 exported from the package SB-PROFILE.
3509 * Some problems left over from porting CMU CL to the new
3510 cross-compilation bootstrap process have been cleaned up:
3511 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3512 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3513 part when getting rid of the compiler *BACKEND* stuff.)
3514 ** The value of *NULL-TYPE* was screwed up, because it was
3515 being initialized before the type system knew the final
3516 definition of the 'NULL type. This screwed up several key
3517 optimizations in the compiler, causing inefficiency in all sorts
3518 of places. (I found it because I wanted to understand why
3519 GET-INTERNAL-RUN-TIME was consing.)
3520 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3521 PROCLAIM FTYPE information. Unfortunately this broke binary
3522 compatibility again, since now the forms output by DEFGENERIC
3523 to refer to functions which didn't exist in 0.6.3.
3524 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3525 can use the new (as of 0.6.3) transform for SOME into MAP into
3527 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3528 values to (INTEGER 0 1000000), so that the time code will no longer
3529 occasionally get blown up by Linux returning 1000000 microseconds
3530 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3531 its output conform to the ANSI spec. (Alas, this makes its output
3532 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3533 * A full call to MAP NIL with a single sequence argument no longer conses.
3534 * fixes to problems pointed out by Martin Atzmueller:
3535 * The manual page no longer talks about multiprocessing as though
3536 it were currently supported.
3537 * The ILISP support patches have been removed from the distribution,
3538 because as of version 5.10.1, ILISP now supports SBCL without us
3539 having to maintain patches.
3540 * added a modified version of Raymond Toy's recent CMU CL patch for
3541 EQUALP comparison of HASH-TABLE
3543 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3545 * The system still can't cross-compile itself with
3546 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3547 entails), but at least it can compile more of itself that way
3548 than it used to be able to, and various buglets which were uncovered
3549 by trying to cross-compile itself that way have now been fixed.
3550 * This release breaks binary compatibility again. This time
3551 at least I've incremented the FASL file format version to 2, so that the
3552 problem can be detected reliably instead of just causing weird errors.
3553 * various new style warnings:
3554 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3555 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3556 ** using the lexical binding of a variable named in the *FOO* style
3557 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3558 as specified by ANSI.
3559 * *RANDOM-STATE* is no longer automatically initialized from
3560 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3561 default behavior of the system is to repeat its behavior every time
3562 it's run. If you'd like to change this behavior, you can always
3563 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3564 old convention there was no comparably easy way to get the system to
3565 repeat its behavior every time it was run.
3566 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3567 FTYPE declarations has been removed, in favor of their ANSI
3568 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3569 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3570 the types of their sequence arguments can be determined at compile time.
3571 This is done through a new open code expansion for MAP which eliminates
3572 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3573 argument types can be determined at compile time.
3574 * The optimizer now transforms COERCE into an identity operation when it
3575 can prove that the coerced object is already of the correct type. (This
3576 can be a win for machine generated code, including the output of other
3577 optimization transforms, such as the MAP transform above.)
3578 * Credit information has been moved from source file headers into CREDITS.
3579 * Source file headers have been made more standard.
3580 * The CASE macro now compiles without complaining even when it has
3583 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3585 * (Note that the way that the PCL macroexpansions were rewritten
3586 to accommodate the change in DEFGENERIC below breaks binary
3587 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3588 not run under sbcl-0.6.2. Once we get out of alpha releases,
3589 i.e. hit release 1.0.0, we'll probably try to maintain binary
3590 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3591 and sbcl-1.4.4. Until then, however, it might be fairly common
3592 for maintenance releases to break binary compatibility.)
3593 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3595 * The reporting of unhandled signals has been changed to print some
3596 explanatory text as well as the report form. (Previously only
3597 the report form was printed.)
3598 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3599 it defines, so that the compiler no longer issues undefined function
3600 warnings for compiled-but-not-yet-loaded generic functions.
3601 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3602 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3603 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3604 section of the ANSI spec are supported.
3605 * The "" nickname for the "KEYWORD" package has been removed.
3606 The reader still handles symbol tokens which begin with a package marker
3607 as keywords, but it doesn't expose its mechanism for doing so in the
3608 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3609 * The system now issues STYLE-WARNINGs for contradictory TYPE
3610 proclamations. (Warnings for contradictory FTYPE proclamations would
3611 be nice too, but those can't be done usefully unless the type system
3612 is made smarter about FUNCTION types.)
3613 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3614 systematized, so that "*target-*.lisp is supposed to exist only on the
3615 target and imply that there's a related file which exists on the
3616 host, and *host-*.lisp is supposed to exist only on the host and imply
3617 that there's a related file which exists on the target. This involves a
3618 lot of renaming. Hopefully the acute confusion caused by the renaming
3619 will be justified by the reduction in chronic confusion..
3620 ** runtime-type.lisp -> early-target-type.lisp
3621 ** target-type.lisp -> late-target-type.lisp
3622 ** early-host-format.lisp -> early-format.lisp
3623 ** late-host-format.lisp -> late-format.lisp
3624 ** host-error.lisp -> misc-error.lisp
3625 ** early-error.lisp -> early-target-error.lisp
3626 ** late-error.lisp -> late-target-error.lisp
3627 ** host-defboot.lisp -> early-defboot.lisp
3628 ** code/misc.lisp -> code/target-misc.lisp
3629 ** code/host-misc.lisp -> code/misc.lisp
3630 ** code/numbers.lisp -> code/target-numbers.lisp
3631 ** code/early-numbers.lisp -> numbers.lisp
3632 ** early-host-type.lisp -> early-type.lisp
3633 ** late-host-type.lisp -> late-type.lisp
3634 ** host-typep.lisp -> typep.lisp
3635 ** load.lisp -> target-load.lisp
3636 ** host-load.lisp -> load.lisp
3637 ** host-disassem.lisp -> disassem.lisp
3638 ** host-insts.lisp -> insts.lisp
3639 ** byte-comp.lisp -> target-byte-comp.lisp
3640 ** host-byte-comp.lisp -> byte-comp.lisp
3641 ** host-signal.lisp -> signal.lisp
3642 ** host-defstruct.lisp -> defstruct.lisp
3643 ** late-target-type.lisp -> deftypes-for-target.lisp
3644 Furthermore, several other previously target-only files foo.lisp (e.g.
3645 hash-table.lisp and random.lisp) have been split into a target-and-host
3646 foo.lisp file and a target-only target-foo.lisp file, with their key type
3647 definitions in the target-and-host part, so that the cross-compiler will
3648 know more about target types.
3649 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3650 gone away. In their place are various *BACKEND-FOO* variables
3651 corresponding to the slots of the old structure.
3652 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3653 into the target SBCL has been fixed.
3654 * The chill.lisp system for loading cold code into a running SBCL
3656 * Support for the CMU CL "scavenger hook" extension has been removed.
3657 (It was undocumented and unused in the CMU CL sources that SBCL was
3658 derived from, and stale in sbcl-0.6.1.)
3659 * Various errors in the cross-compiler type system were detected
3660 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3661 (enabling various consistency checks). Many of them were fixed,
3662 but some hard problems remain, so the compiler is back to
3663 running without *TYPE-SYSTEM-INITIALIZED* for now.
3664 * As part of the cross-compiler type system cleanup, I implemented
3665 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3666 * I have started adding UNCROSS calls throughout the type system
3667 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3668 on cross-compiler input files will be able to go away, and various
3670 * CONSTANTP now returns true for quoted forms (as explicitly required
3673 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3675 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3676 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3677 (MAKE-LIST -1). In the longer term, it ought to become true
3678 that declarations are assertions even at SAFETY 1. For now, it's not
3679 quite true even at SAFETY 3, but it's at least more nearly true..
3680 (Note that this change seems to increases the size of the system by
3681 O(5%) and to decrease the speed of the compiler by 20% or more.)
3682 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3683 for the problem of printing dozens of lines of distracting information
3684 about low-level system machinery as part of the top stack frame
3685 on entry to the debugger when an undefined function was called.
3686 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3687 is not reset to COMMON-LISP-USER.
3688 * Compilation of stuff related to dyncount.lisp has been made conditional
3689 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3690 smaller. The various dyncount-related symbols have been moved into
3691 a new "SB-DYNCOUNT" package.
3692 * tty-inspect.lisp has been renamed to inspect.lisp.
3693 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3694 feature has gone away. (When we eventually port to other flavors of
3695 libc and/or Unix, we'll try to make the differences between flavors
3696 invisible at the user level.)
3697 * Various other *FEATURES* tags, and/or their associated conditionals,
3698 have been removed if obsolescent, or given better documentation, or
3699 sometimes given more-mnemonic names.
3701 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3703 * tidied up "make.sh" script
3704 * tidied up system directory structure
3705 * better "clean.sh" behavior
3706 * added doc/FOR-CMUCL-DEVELOPERS
3707 * many many small tweaks to output format, e.g. removing possibly-confusing
3708 trailing #\. character in DESCRIBE-INSTANCE
3709 * (EQUALP #\A 'A) no longer signals an error.
3710 * new hashing code, including EQUALP hashing
3711 * tidied up Lisp initialization and toplevel
3712 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3713 * command line argument processing
3714 * added POSIX-GETENV function to deal with Unix-ish environment variables
3715 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3716 terminating SBCL on EOF
3717 * non-verbose GC by default
3718 * There is no more "sbcl" shell script; the sbcl file is now the C
3719 runtime executable (just like CMU CL).
3720 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3722 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3723 symbol and package information as Lisp native symbol and package objects
3724 instead of strings naming symbols and strings naming packages. This way,
3725 whenever packages are renamed (as in warm init), debug information is
3726 transformed along with everything else.
3727 * tweaked the optimization policy declarations which control the building
3728 of SBCL itself. Now, among other things, the system no longer saves
3729 source location debugging information. (This helps two problems at once
3730 by reducing SBCL size and by keeping SBCL from trying to look for its
3731 sources -- which may not exist -- when reporting errors.)
3732 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3733 debugging and testing purposes
3734 * cleaned up printing, making the printer call PRINT-OBJECT for
3735 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3736 methods, giving nearly-ANSI behavior
3737 * converted almost all special variables to use *FOO* naming convention
3738 * deleted PARSE-TIME functionality, since it can be done portably
3739 * moved some files out of cold init into warm init
3740 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3742 * regularized formatting of source files
3743 * added an install.sh script
3744 * fixed ridiculous memory usage of cross-compiler by making
3745 compiler/alloc.lisp not try to do pooling unless it can hook
3746 itself into the GC of the cross-compilation host. Now the system
3747 builds nicely on my old laptop.
3748 * added :SB-ALLOC in target-features.lisp-expr
3749 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3750 was not implemented)
3751 * re-did condition handling and note reporting in the compiler. Notes
3752 are no longer handled by signalling conditions. Style warnings
3753 and warnings are handled more correctly and reported in such a way
3754 that it's easy to find one or the other in your output (so that you
3755 can e.g. figure out which of many problems caused COMPILE-FILE to
3757 * changed the severity of several compiler warnings from full WARNING
3758 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3759 compiler note reporting so that it doesn't use the condition system
3760 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3761 COMPILE-FILE command)
3762 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3763 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3764 no longer supported, since their implementation was deeply intertwingled
3765 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3766 * removed bogus "support" for compiler macros named (SETF FOO), and
3767 removed the compiler macro for SETF INFO (but only after making a fool
3768 of myself on the cmucl-imp mailing list by posting a bogus patch for
3769 DEFINE-COMPILER-MACRO..)
3770 * Compiled files containing forms which have side effects on the Lisp
3771 reader (such as DEFPACKAGE forms) are now handled more correctly.
3772 (Compiler queuing of top level lambdas has been suppressed by setting
3773 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3774 * deleted various currently-unused source files, e.g. gengc.lisp. They
3775 may be added back at some point e.g. when porting to other architectures,
3776 but until they are it's distracting to distribute them and to try to
3778 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3779 there were so many of them they're just distractions, and UNCROSS is
3780 known to be able to handle the current sources
3781 * moved PROFILE functionality into TRACE, so that it will be clear
3782 how the wrapping and unwrapping of functions when you profile them
3783 interacts with the wrapping and unwrapping of functions when you
3784 trace them. (Actually, the functionality isn't there yet, but at least
3785 the interface specification is there. Hopefully, the functionality will
3786 arrive with some future maintenance release.)
3787 * removed host-oops.lisp
3788 * changed signature of QUIT function to allow UNIX-CODE argument
3789 * fixed READ-SEQUENCE bug
3790 * tweaked verbose GC output so that it looks more like the progress
3791 output that ANSI specifies for functions like LOAD
3792 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3793 * added <http://sbcl.sourceforge.com> to the banner information printed by