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 * minor incompatible change: prevent the user from specializing the
8 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
9 somewhat counter to the intent of the protocol, I (CSR) think, and
10 additionally it just doesn't work in SBCL as currently
11 implemented, thanks to optimizations (that are always valid for
12 the other three SLOT-VALUEish functions, but not for the setter).
13 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
14 failed with a type error. (reported by Pascal Bourguignon)
15 * bug fix: merging pathnames against defaults with :DIRECTORY
16 starting with '(:RELATIVE :BACK) should preserve the :BACK.
17 (reported by James Y Knight)
18 * improved SB-BSD-SOCKETS support on Windows. (thanks to Timothy
20 * bug fix: saving large (>2GB) cores on x86-64 now works
21 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
22 constant index and a (simple-array (signed-byte 32)) array
23 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
24 error (patch by Robert J. Macomber)
25 * feature: TIME now displays the amount of run-time spent in GC
26 * fixed some bugs revealed by Paul Dietz' test suite:
27 ** MISC.641: LET-conversion were not supposed to work in late
30 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
31 * new feature: source path information is generated for macro-expansion
32 errors for use in IDE's like Slime (thanks to Helmut Eller)
33 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
34 * bug fix: compiler does not loop forever on an invalid type in
36 * improvement: compilation of most CLOS applications is significantly
38 * optimization: added a limited bytecode compiler for simple toplevel
39 forms, speeding up compilation and FASL loading
41 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
42 * minor incompatible change: in sbcl-0.9.11 (but not earlier
43 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
44 into subclasses of the finalized class. Now user calls to
45 FINALIZE-INHERITANCE finalize just the one class, and calls by the
46 system return before any subclasses are finalized.
47 * minor incompatible change: The reader no longer ignores errors
48 regarding non-existent packages in #+ and #- feature tests.
49 * new feature: command line options --no-sysinit, --no-userinit to
50 inhibit loading the corresponding init files
51 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
52 for sending data through UDP sockets (thanks to François-René Rideau)
53 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
54 error is signalled instead (thanks to François-René Rideau)
55 * minor incompatible change: Error signalling behaviour of lexical
56 operations violating package locks has changed slightly. Refer to
57 documentation on package locks for details.
58 * enhancement: EVAL can process IF-expressions without resorting to the
60 * bug fix: Some operations involving SIN, COS, or TAN weren't being
61 constant-folded properly due to missing out-of-line functions.
62 (reported by Mika Pihlajamäki)
63 * bug fix: LISTEN sometimes returned T even in cases where no data was
64 immediately available from the stream
65 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
66 were reversed. (reported by Levente Mészáros)
67 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
68 when given a vector displaced to an adjustable array. (reported
70 * bug fix: derivation of float boundaries from numbers outside the
71 appropriate float range (reported by John Wiseman)
72 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
74 * bug fix: precomputing cache entries for generic functions with
75 some subclasses of specializers as yet invalid does not attempt to
76 fill a cache line with a negative offset. (reported by Levente
78 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
79 allow condition class objects as arguments to DOCUMENTATION and
80 (SETF DOCUMENTATION); only find and set documentation for
81 structure names for the STRUCTURE doc-type. (suggested by Gary
83 * improvements to the Win32/x86 port:
84 ** (user-homedir-pathname) and default initialization file
85 locations now know about the user's "Documents and Settings"
86 directory (thanks to Yaroslav Kavenchuk)
87 ** run-program is implemented (thanks to Mike Thomas)
88 ** sockets support (thanks to Timothy Ritchey)
89 ** better backtrace support (thanks to Alastair Bridgewater)
90 ** sb-grovel supported
91 ** asdf-install and sb-posix work somewhat
92 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
93 * improvements to the Solaris/x86 port:
94 ** works on Solaris 11/Solaris Express
95 ** floating-point exception handling support
96 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
97 * fixed some bugs revealed by Paul Dietz' test suite:
98 ** REMOVE-METHOD returns its generic function argument even when
99 the method is not one of the generic functions' methods.
100 ** objects declared as MEMBER types can be admissible arguments to
102 ** printing characters should simply be printed by the FORMAT ~:C
104 ** compiler failure when compiling functions with hairy constant
105 defaults for optional parameters.
106 ** compiler produces wrong code when MAYBE-INLINE-expanding a
107 function, which is already optimized.
109 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
110 * new platform: experimental support for SBCL x86/Darwin, including
111 MacOS X 10.4.5 on Intel.
112 * new feature: Unicode character names are now known to the system
113 (through CHAR-NAME and NAME-CHAR).
114 * new feature: the filesystem location of SBCL's core file is
115 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
116 * minor incompatible change: the contrib modules SB-POSIX and
117 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
118 this change is to make it easier to distribute
119 location-independent binaries.
120 * bug fix: as implied by AMOP, standardized classes no longer have
121 slots named by external symbols of public packages. (reported by
123 * optimization: calling functions via constant symbols -- as in
124 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
125 via the function object as in (FUNCALL #'FOO).
126 * optimization: on x86 and x86-64, the calling convention for the
127 case of dealing with an unknown number of values has been altered
128 to be friendlier to the prediction heuristics implemented,
129 particularly with reference to CALL and RET pairing. (thanks to
130 Alastair Bridgewater)
131 * optimization: on x86, the code for access to thread-local symbol
132 values and binding/unbinding of thread-local symbols is smaller.
133 (thanks to Alastair Bridgewater)
134 * enhancement: CONSTANTP is now able to determine constantness of
135 more complex forms, including calls to constant-foldable standardized
136 functions and some special forms beyond QUOTE.
137 * fixed bug: occasional GC crashes on Solaris/x86
138 * optimization: x86-64 supports stack allocation of results of simple
139 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
140 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
141 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
142 (thanks to James Knight)
143 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
144 by the standard (thanks to Stephen Compall)
146 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
147 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
148 be used for bundling the runtime and the core file into one
149 executable binary. This feature is not currently supported on all SBCL
150 platforms. (thanks to James Bielman and NIIMI Satoshi)
151 * new feature: a generational or ephemeral garbage collector is now
152 the default on the PowerPC platform (both Linux and Darwin). The
153 old Cheney (stop and copy) collector is a build-time option.
154 (thanks to Cyrus Harmon, after Raymond Toy)
155 * minor incompatible change: the method by which SBCL finds its
156 contributed modules has changed; it no longer relies on symbolic
157 links from an $SBCL_HOME/systems directory, but searches directly
158 in the subdirectories of $SBCL_HOME.
159 * enhancement: the dynamic heap size on the Linux/PPC platform is
160 markedly larger, even using the older Cheney garbage collector.
161 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
162 certain complicated string types. (reported by Gary King)
163 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
164 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
165 * fixed bug: DOCUMENTATION on structures no longer signals an
166 error if no structure type of the right name exists. (reported by
168 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
169 applicable non-standard primary or :AROUND method on
170 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
171 no longer cause unbound variable errors. (reported by Kevin Reid)
172 * fixed bug: metacircle resolution in cases where methods have slots
173 added before the slots from STANDARD-METHOD. (reported by Jean
175 * fixed bug: the Power PC assembly code for calling into lisp
176 sometimes computed its return address wrongly (depending on the
177 15th bit of the address, fixed at link time). This used to
178 manifest itself in a segmentation violation while building PCL.
179 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
180 many others over the years)
181 * fixed bug: ROOM no longer reports silly numbers for stack usage if
182 the stack pointer (treated as a signed integer) is negative.
183 (thanks to Peter van Eynde)
185 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
186 * new platform: experimental support for the Windows operating
187 system has been added. (thanks to Alastair Bridgewater)
188 * fixed several bugs in and robustified the PPC FFI (including
189 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
190 * bug fix: allow non-simple string symbol names (reported by Paul
192 * bug fix: interrupt handling on NetBSD (thanks to Richard M
194 * bug fix: saving a core corrupted callbacks on x86/x86-64
195 * bug fix: closed a loophole in metacircularity detection and
196 grounding in the PCL implementation of CLOS.
197 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
198 (reported by tomppa on #lisp)
199 * bug fix: clear the direction flag on Lisp -> C transitions, as
200 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
201 (reported by Andrej Grozin and Hendrik Maryns)
202 * optimization: major improvements to GC efficiency on GENCGC platforms
203 * optimization: faster implementation of EQUAL
204 * optimization: emit more efficient opcodes for some common
205 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
206 * optimization: several other minor code-generation improvements on x86-64
207 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
209 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
210 * minor incompatible change: (SETF CLASS-NAME) and (SETF
211 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
212 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
213 by ANSI as a generic function, and for consistency (SETF
214 GENERIC-FUNCTION-NAME) is treated likewise.
215 * fixed bug #233.b: make constraint propagation notice when a variable
216 value is changed after it is referenced but before it is used
217 * fixed bug #296: no more arbitrary behaviour from filenames with
218 odd characters as --load arguments. (reported by Adam Warner)
219 * fixed bug #390: :CHARACTER-SET pathname components now work as
220 expected. (reported by Tim Daly Jr)
221 * fixed bug #391: complicated :TYPE intersections in slot
222 definitions no longer cause an error in PCL internals.
223 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
224 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
225 returning the number of octets which would be written to the
226 file-stream. (thanks to Robert J. Macomber)
227 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
228 arguments to the internal stream functions. (thanks to David
230 * bug fix: stack exhaustion is now handled on the x86/SunOS
231 platform. (thanks to Daisuke Homma)
232 * bug fix: more accurate ROOM results on GENCGC platforms
233 * optimization: improved type inference for arithmetic-for
234 index variables in LOOP
235 * optimization: faster floating-point SQRT on x86-64
236 * fixed some bugs revealed by Paul Dietz' test suite:
237 ** DOCUMENTATION returns NIL instead of "" for method combinations
238 that don't have a docstring
240 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
241 * minor incompatible change: (SETF CLASS-NAME) and (SETF
242 GENERIC-FUNCTION-NAME) are no longer generic functions, and
243 therefore are not customizeable by user code (as seems to be at
244 least permitted and maybe required by AMOP). As a consolation,
245 however, the SBCL implementation of these functions now calls
246 REINITIALIZE-INSTANCE as specified by AMOP.
247 * bug fix: slot-definition documentation is propagated as per ANSI
248 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
249 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
250 argument, as do eslotd initialization methods. (from Pascal
251 Costanza's "Closer" project)
252 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
253 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
255 * bug fix: it is now possible to have more than one subclass of
256 STANDARD-GENERIC-FUNCTION without causing stack overflow.
257 (reported by Bruno Haible, Pascal Costanza and others)
258 * bug fix: the dependent update protocol now works for generic
259 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
261 * bug fix: condition-class instances corresponding to
262 DEFINE-CONDITION forms are now created eagerly. (reported by
263 Kalle Olavi Niemitalo on comp.lang.lisp)
264 * bug fix: floating point printing is more accurate in some
265 circumstances. (thanks to Simon Alexander)
266 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
267 merged with *DEFAULT-PATHNAME-DEFAULTS*.
268 * bug fix: callbacks on OS X now preserve stack-alignment.
269 (thanks to Cyrus Harmon)
270 * enhancement: the x86-64 disassembler is much better at
271 disassembling SSE instructions. (thanks to Lutz Euler)
272 * enhancement: CHECK-TYPE on a variable now allows type-inference
273 (thanks to Vincent Arkesteijn)
274 * optimization: improved performance of EUC-JP external format.
275 (thanks to NIIMI Satoshi)
276 * optimization: performance improvements to IO on file streams of
277 :ELEMENT-TYPE CHARACTER
278 * optimization: much faster memory allocation on x86-64
279 * optimization: faster garbage collections (latency and throughput)
281 * optimization: faster bignum subtraction and division on x86 and x86-64
282 (thanks to Lutz Euler)
283 * optimization: more accurate type-inference for loops that use a
284 floating point index variable or a negative step.
286 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
287 * bug fix: add a workaround to SBCL looping infinitely at startup on
288 Linux kernels with apparently buggy implementations of personality().
289 (thanks to Svein Ove Aas)
290 * bug fix: Unicode symbols are correctly printed in LDB backtraces
291 (thanks to David Lichteblau)
292 * bug fix: local bindings shadow global symbol macros for
293 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
294 * optimization: non-open coded uses of numeric comparison operators
295 (e.g. >) no longer cons when called with more than one parameter
296 on platforms supporting dynamic-extent allocation.
297 * enhancement: saving cores with foreign code loaded is now supported
298 on MIPS/Linux in addition to the previously supported platforms.
299 * bug fix: FIND-RESTART now tests for activity, not applicability when given
300 a restart object as identifier. (reported by Helmut Eller for CMUCL)
301 * bug fix: division by zero in sb-sprof when no samples were collected
302 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
303 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
304 (thanks to Alexander Kjeldaas)
305 * new feature: ignoring the return values of destructive functions
306 like NREVERSE signals a compile-time style-warning.
307 (thanks to Kevin Reid)
308 * new feature: the alignment of alien structure fields can be
309 explicitly specified. (thanks to Cyrus Harmon)
310 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
311 and dump core on SIGQUIT
313 ** incompatible change: threads do not inherit values of specials
314 from their parents (see manual)
315 ** bug fix: threads stacks belonging to dead threads are freed by the
316 next exiting thread, no need to gc to collect thread stacks anymore
317 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
318 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
319 ** bug fix: dynamic variable and thread start related gc lossage
320 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
322 * fixed some bugs revealed by Paul Dietz' test suite:
323 ** SUBTYPEP is slightly more accurate on heinously complicated
324 CONS types where some of the members have uncertain (in the
325 NIL, NIL sense) type relationships to each other.
327 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
328 no longer purified unless :PURIFY T is explicitly specified.
329 ** Non-purified cores are significantly smaller than before
331 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
332 * new feature: timers based on Zach Beane's excellent timer package
333 * added support for the following external formats: koi8-u,
334 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
335 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
336 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
337 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
338 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
339 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
340 * incompatible change: a threaded SBCL will no longer revert to
341 non-threaded mode on non-NPTL systems, but refuse to start entirely.
342 * bug fix: interrupts are disabled until startup is complete; no
343 more sigsegvs when receiving a signal to soon
344 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
346 * bug fix: add a workaround for the memory randomization features in
347 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
348 workaround will only be in effect on systems with the proc filesystem
350 * bug fix: printing objects of type HASH-TABLE signals a
351 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
353 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
354 have been running for over 50 days. (reported by Gilbert Baumann)
355 * bug fix: the logic for getting names of functions gets less
356 confused when confronded with alternate-metaclass
357 funcallable-instances. (reported by Cyrus Harmon)
358 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
359 consistent, even on internal alternate-metaclass objects.
360 * bug fix: flush closure information collected by physical
361 environment analysis prepass before the main pass. (bug reported
363 * bug fix: compiler pack phase does not modify a hashtable, which is
364 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
365 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
366 may be expanded. (reported by Alexander <malishev> on c.l.l)
367 * bug fix: no more style-warnings when compiling code using LAST
368 under high-SPEED low-SPACE settings. (reported by David Wallin
370 * fixed some incompatibilities between SBCL's MOP and the MOP
372 ** the METAOBJECT class is now implemented;
373 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
374 STANDARD-OBJECT, as required;
375 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
376 now compatible (as required by VALIDATE-SUPERCLASS); there
377 remains a consistency requirement over the presence or absence
378 of the FUNCTION class in the superclass list of finalized
379 classes; see the manual for more details;
380 ** the :AROUND method for COMPUTE-SLOTS on
381 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
382 requested slot ordering.
384 ** bug fix: parent thread now can be gc'ed even with a live
386 ** bug fix: binding a special with PROGV to no value is not
387 the same as that symbol not having been bound (thanks to
389 ** bug fix: binding specials is thread safe (thanks to
391 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
392 and SB-SPROF do not die with 'no handler for signal XX in
393 interrupt_handle_now(..)' anymore
394 ** bug fix: WITH-TIMEOUT works with multiple running threads
395 * fixed some bugs revealed by Paul Dietz' test suite:
396 ** ENSURE-GENERIC-FUNCTION should take a method class object for
397 the :method-class keyword argument.
399 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
400 * new port: the Solaris operating system on x86 processors is now
401 mostly supported, though some rough edges in the environment
402 remain. (thanks to Daisuke Homma)
403 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
404 space for its heap. (thanks to Thiemo Seufer)
405 * optimization: SBCL on MIPS platforms now supports dynamic-extent
406 closures. (thanks to Thiemo Seufer)
407 * minor incompatible change: eof selects abort in the debugger.
408 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
409 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
411 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
412 deprecated, and will go away in a future revision of SBCL.
413 * minor incompatible change: GC-ON and GC-OFF are no longer
414 implemented with a counter, it does not matter how many times gc
415 is switched on or off
416 * bug fix: discriminating functions for generic function classes
417 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
418 make invalid assumptions about method precedence order. (reported
420 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
421 platforms. (thanks to Thiemo Seufer)
422 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
423 * bug fix: correct pathname printing with printer escaping is on.
424 (thanks to Kevin Reid)
425 * bug fix: complex VOP definitions in "user-space" no longer trigger
426 package locks. (reported by Zach Beane)
427 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
428 for funcallable-instance functions, and is no different from
430 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
431 characters in the current readtable. (reported by Nicholas Neuss)
432 * bug fix: SUBTYPEP on various CONS types returns more a more
433 accurate acknowledgment of its certainty.
434 * optimizations: REMOVE-DUPLICATES now runs in linear time on
435 lists in some cases. This partially fixes bug 384.
436 * flush all standard streams before prompting in the REPL and the
438 * bug fix: signal handling and triggering gc do not conflict
439 directly anymore, in particular a high frequency sb-sprof does
440 not prevent gc from running
441 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
442 approximation for timezone and DST information between the
443 universal time epoch and the smallest negative 32-bit time_t.
444 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
445 year 1899 so long as the encoded time is non-negative.
446 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
447 the correct answer for non-constant rotate amounts when performing
448 an inline 32-bit rotation.
450 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
451 there is only one thread in the session
452 ** bug fix: memory leak for streams created in one thread and
453 written to in another
454 ** bug fix: lockup when compiled with gcc4
455 ** bug fix: race that allows the gc to be triggered when gc is
457 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
458 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
460 ** bug fix: finished threads are gc'ed properly
461 * fixed some bugs revealed by Paul Dietz' test suite:
462 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
463 has arguments with a different set of applicable methods from
464 the orignal arguments.
465 ** The type error thrown by MAP now has a correct expected-type
467 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
468 name a compiled function.
469 ** Three MISC tests where a large bignum was improperly coerced to
470 a float that couldn't represent that bignum during type
471 derivation were fixed.
472 ** SUBTYPEP can now handle types involving the negation of a
473 list-form FUNCTION type.
474 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
475 as (COMPLEX (AND RATIO (NOT FIXNUM))).
476 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
478 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
479 * New feature: Experimental support for bivalent streams: streams
480 opened with :element-type :default now allow character and binary
481 (unsigned-byte 8) I/O
482 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
483 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
485 * bug fix: no more highly sporadic "couldn't check whether ~S is
486 readable" when reading a stream and an interrupt hits in the middle
487 of a select system call
488 * compiler better recognizes complex arrays (reported by Eduardo
490 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
491 for variables on SBCL built with linkage-tables. (reported by Luis
493 * various error reporting improvements.
494 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
495 (Thanks to Thiemo Seufer)
496 * bug fix: nested reader invokations work correctly
497 * bug fix: it is possible to have simultaneous references to foreign
498 code and foreign data with the same name.
500 ** added x86-64 support
501 ** incompatible change: the threading api now works with thread
502 objects instead of thread ids
503 ** bug fix: threads are protected from signals and interruption when
504 starting up or going down
505 ** bug fix: a race where an exiting thread could lose its stack to gc
506 ** bug fix: don't halt on infinite error in threads if possible
507 ** fixed numerous gc deadlocks introduced in the pthread merge
508 ** bug fix: fixed thread safety issues in read and print
509 ** bug fix: debugger doesn't hang on session lock if interrupted at
510 an inappropriate moment
511 ** bug fix: run-program is now thread safe(r)
512 ** bug fix: inner with-recursive-lock no longer releases the mutex
513 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
514 * fixed some bugs revealed by Paul Dietz' test suite:
515 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
516 EXPECTED-TYPE slots filled.
517 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
519 ** MAKE-CONDITION accepts classes as type-designators.
520 ** COMPILE may never return NIL.
521 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
522 range before calling Unix time functions
524 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
525 * numerous signal handling fixes to increase stability
526 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
527 * minor incompatible change: we now correctly canonize default
528 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
529 as per the MOP, rather than the historical (INITARG INITFUNCTION
530 INITFORM). (reported by Bruno Haible)
531 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
532 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
533 for more information.
534 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
535 pathname is a directory pathname.
536 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
537 * dynamic space size on PPC has been increased to 768Mb. (thanks to
539 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
540 :METACLASS argument in addition to a class name. (reported by
541 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
542 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
543 (reported by Bruno Haible; patch from Gerd Moellmann)
544 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
546 * bug fix: more cleanups to the floating point exception handling on
547 x86-64 (thanks to James Knight)
548 * bug fix: the compiler does not try to constant fold calls of
549 COERCE and BIT-* functions when they return freshly constructed
550 objects. (reported by Nikodemus Siivola and Paul Dietz)
551 * optimization: improved the MIPS versions of generic (in the
552 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
553 * optimization: direct conversion of (unsigned-byte 32) to floats on
554 the PowerPC platform.
555 * optimization: structure instances with raw slots now use less
556 memory, and probably show better memory locality. (thanks to
558 * optimization: DYNAMIC-EXTENT declarations for lists and closures
559 are treated as requests for stack allocation on the x86-64,
560 Alpha-32, PPC and SPARC platforms.
561 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
562 the wrong file descriptor; implementation of SOCKET-OPEN-P.
563 (thanks to Tony Martinez)
565 ** gcing a dead thread can no longer lead to lockups
566 ** threads block signals until they are set up properly
567 ** errno is no longer shared by threads
568 ** interrupt-thread restores the eflags register on x86
569 ** fixed some lockups due to gc/thread interaction
570 * fixed some bugs revealed by Paul Dietz' test suite:
571 ** invalid dotted lists no longer raise a read error when
573 ** signal an error if a symbol that names a declaration is used
574 as the name of a type, or vice versa
575 ** allow using the (declare (typespec var*)) abbreviation for
576 (declare (type typespec var*)) with all type specifiers
577 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
578 ** always return NIL from PPRINT-POP when OBJECT is NIL
579 ** don't signal errors when pretty-printing malformed LABELS,
580 FLET or MACROLET forms
581 ** declarations in a DOLIST body are also in scope for the
583 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
585 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
588 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
589 * fixed cross-compiler leakages that prevented building a 32-bit
590 target with a 64-bit host compiler.
591 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
592 opened with :IF-EXISTS :OVERWRITE.
593 * fixed bug 281: error for an invalid qualifier in a short-form method
594 combination method is not signalled until the faulty method is called.
595 * bug fix: iteration variable type inferrer failed to deal with open
596 intervals. (reported by Alan Shields)
597 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
598 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
599 or more runtime options were provided to the sbcl binary.
600 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
601 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
602 * optimization: there's now a fast-path for fixnum arguments in the
603 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
604 * optimization: the code generated on x86-64 is more compact thanks
605 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
606 * bug fix: floating-point exception handling now partly works on
607 x86-64. (Thanks to James Knight)
608 * improvement to the MIPS backend from Thiemo Seufer: C-style
609 64-bit long long arguments and return values to alien functions
611 * fixed some bugs revealed by Paul Dietz' test suite:
612 ** the type-error signalled from WARN has a filled-in DATUM slot.
613 ** the type-error required when a stream is not associated with
614 a file has the stream as its datum.
615 ** type-errors on single-floats on x86-64 no longer have
616 :INVALID-OBJECT as the datum
617 ** the type-errors signalled for invalid function names now have
618 a correct expected type
619 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
620 for structures defined with a :TYPE. Documentation strings for
621 typed structures are no longer immediately discarded
622 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
623 broadcast streams and synonym streams. FILE-LENGTH now also works
624 on broadcast streams.
626 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
627 * incompatible change: the --noprogrammer option, deprecated since
628 version 0.7.5, has been removed. Please use the equivalent
629 --disable-debugger option instead.
630 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
632 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
633 inoperational for a while now) has been completely removed.
634 * null lexical environments are now printed as #<NULL-LEXENV>,
635 significantly reducing the amount of clutter in typical backtraces.
636 * documentation on weak pointers, finalization, and after GC hooks
637 has been added to the manual.
638 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
639 as other specialized array types, is much faster. SUBSEQ and
640 COPY-SEQ on such arrays have also been sped up.
641 * optimization: EQL is now more efficient when at least other argument
642 is known to be of type (OR FIXNUM (NOT NUMBER)).
643 * fixed bug: the runtime is now less vulnerable to changes in the
644 size of the SBCL object on OS X, and virtual memory is reserved for
645 all spaces that need to be at a fixed address.
646 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
647 * fixed bug: finalizers and after GC hooks that cause consing are now
649 * fixed bug: compiler error messages and summaries are now printed to
650 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
651 * fixed inference of the upper bound of an iteration variable.
652 (reported by Rajat Datta).
653 * fixed bug 211e: calling local functions with duplicated constant
654 keyword argument no longer causes a bogus style warning about an
656 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
657 ftype declaration to be disregarded. (reported by Dave Roberts)
658 * fixed bug 373: caused by erronous compilation of references to alien
659 variables in the runtime on ppc/darwin.
660 * fixed bug 376: CONJUGATE type deriver.
661 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
662 (reported by Baughn on #lisp)
663 * a cross-compiler bug on non-x86 platforms has been identified and
664 fixed. (thanks to Bruno Haible)
665 * improvements to the MIPS runtime code for increased stability.
666 (thanks to Thiemo Seufer)
667 * increased the maximimum compact environment size to allow
668 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
669 * partial workaround for bug 135: don't name the function we're
670 calling for hairy cases of EVAL, so as not to accumulate one environment
671 entry per call to EVAL. (reported by Kevin Reid)
672 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
673 * optimization: type testing for non-vector arrays should be faster.
674 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
676 * bug fix: setting 31st element of a bit vector to zero did not work
678 * bug fix: redefining a class definition which failed due to a
679 previous accessor / function clash now works (but see BUGS entry
680 #380 for more problems in this area). (thanks to Zach Beane)
681 * the long form of DEFINE-METHOD-COMBINATION disables method group
682 checking when given a single method group with pattern *. (thanks
684 * on x86 compiler supports stack allocation of results of simple
685 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
686 * support for building and running on Mac OS X 10.4 "Tiger" has been added
687 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
688 * fixed some bugs related to Unicode integration:
689 ** the restarts for recovering from input and output encoding
690 errors only appear when there is in fact such an error to
692 * fixed some bugs revealed by Paul Dietz' test suite:
693 ** MISC.549 and similar: late transformation of unsafe type
694 assertions into derived types caused unexpected code
696 ** SCALE-FLOAT type deriver is less wrong.
697 ** type derivers for EXP, LOG and similar functions compute result
698 types for complex arguments better.
699 ** (MISC.563) CONJUGATE type deriver works for very restricted
701 ** out-of-line type testers for character strings are available.
702 ** EQUAL compiler transform understands specialness of objects
704 ** accessing double-floats stored on the stack now works on x86-64.
705 ** debugger internals could sometimes create invalid lispobjs,
706 resulting in GC crashes.
707 ** MISC.548: type check weakening can convert required type into
709 ** initialization forms for bindings are not in scope of free special
712 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
713 * incompatible change: thread support for non-NPTL systems has
714 been removed - locking is buggy and unreliable. A threaded
715 SBCL build will now warn at startup and refuse to create
716 new threads, unless futex support is detected in the kernel
717 * incompatible change: the top level REPL now has only an ABORT
718 restart associated with it, not TOPLEVEL and ABORT as it used to.
719 TOP and TOPLEVEL are now available as debugger commands for
720 returning to the top level.
721 * incompatible change: forms evaluated in the REPL now use the
722 global optimization policy.
723 * incompatible change: user- and system-initialization files are
724 no longer processed with LOAD, but by READ and EVAL; hence the
725 global optimization policy, startup package, readtable, etc,
727 * The .fasl file format number has been incremented because of
728 various incompatible changes.
729 * internal entry point details and argument counts no longer appear
730 in backtraces unless explicitly requested by setting
731 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
732 * built-in and standard functions no longer have names like "top
733 level local call to FOO".
734 * fixed bug 32: functions defined in non-null lexical environments
735 now have more legible printed representation
736 * fixed bug 33: functions defined in non-null lexical environemnts
737 are now more amenable to inspection by INSPECT.
738 * workaround for bug 354: XEPs no longer appear in backtraces unless
739 explicitly requested.
740 * fixed bug: receiving the signal which results from attempting to
741 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
742 SIGBUS on other *BSDs) on architectures where the C stack is also
743 the Lisp stack (x86 and x86-64) from foreign code no longer leads
744 to debugger confusion or wild instability. (reported by Cheuksan
746 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
747 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
748 (reported by Lutz Euler)
749 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
750 compiled code. (reported by Paul Dietz)
751 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
752 specialized parameter to a method to something that is not TYPEP
753 the specializer is now possible.
754 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
755 face of package deletion.
756 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
757 pathname. (reported by David Tolpin on c.l.l)
758 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
759 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
760 STRING-STREAM and FILE-STREAM, respectively.
761 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
762 than previously; in addition, its test suite is now run on build.
763 (thanks to Luke Gorrie)
764 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
765 * optimization: added a immediate representation for single-floats
767 * fixed some bugs revealed by Paul Dietz' test suite:
768 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
769 ** attempting to create a package with a colliding nickname causes
770 correctable errors to be signalled.
771 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
772 ** several x86-64 backend bugs related to sign-extension of immediate
775 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
776 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
777 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
779 * fixed loading of multiply forward-referenced layouts.
780 (thanks to Cheuksan Wang)
781 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
782 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
783 (see COMPILE-FILE documentation for details of :PRINT option)
784 * fixed bugs 19 and 317: fixed-format floating point printing is
785 more accurate. This also fixes a bug reported by Adam Warner
786 related to the ~@F format directive.
787 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
789 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
790 dispatch table if the from-char is a dispatch macro character.
791 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
792 by the wordsize no longer ignore the last word. (reported by Lutz
794 * fixed bug in type checking of dynamic-extent variables. (reported
796 * optimization: sequence traversal functions use their freedom to
797 coerce function designators to functions.
798 * optimization: code with many calls to CLOS methods specialized on
799 CLOS classes has had redundant type checks removed.
800 * optimization: added declarations to speed up operations that access
801 the internal character database (for example STRING-UPCASE)
802 * optimization: comparison operations between floats and sufficiently small
803 fixnums no longer create extra rationals
804 * fixed some bugs related to Unicode integration:
805 ** portions of multibyte characters at the end of buffers for
806 character-based file input are correctly transferred to the
807 start of the buffer at the next read.
808 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
809 passing it through to OPEN.
810 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
811 argument given to internal calls to OPEN.
812 ** the built-in comment readers (introduced by character sequences
813 ";" and "#|") are more forgiving to encoding errors; they will
814 STYLE-WARN and then attempt to resync the stream at a character
815 boundary. (thanks to Teemu Kalvas)
816 * fixed some bugs in the x86-64 port:
817 ** Negative short int return values from c-calls are sign-extended
819 ** The stack is aligned to 16-bytes for c-calls, as required by
820 the ABI. (thanks to Cheuksan Wang)
821 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
822 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
823 for enums. (thanks to Vincent Arkesteijn)
824 ** Multiple small optimizations and bugfixes for floating point
826 * fixed some bugs revealed by Paul Dietz' test suite:
827 ** Space, Tab, Linefeed, Return and Page have the invalid
828 secondary constituent character trait.
829 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
831 ** WITH-INPUT-FROM-STRING should only update the index place on
833 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
835 ** Bit-array operations (BIT-AND and similar) worked incorrectly
836 with one-dimensional arrays with fill pointers.
837 ** TYPE-OF failed on a complex with an integer realpart and a
839 ** compiler failure during type inference for the code of form
840 (IF (EQL X (THE ...)) ...) (MISC.535).
842 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
843 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
844 platform. The port supports 61-bit fixnums, large memory spaces
845 and reloading shared object files.
846 * enhancement: saving cores with foreign code loaded is now
847 supported on ppc/Darwin in addition to the previously supported
849 * enhancement: the statistical profiler now walks deeper into the
850 call stack for more meaningful call-graphs and accrued time
851 reports (x86/x86-64 only). It also now reports time spent in
853 * enhancement: it is now possible to trace most individual methods
854 of a generic function in addition to tracing the generic function
856 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
858 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
859 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
860 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
861 produces an error. (thanks to Vincent Arkesteijn)
862 * bug fix: NAMESTRING on pathnames with :WILD components in their
863 directories works correctly. (thanks to Artem V. Andreev)
864 * fixed bug 125: compiler preserves identity of closures. (reported
866 * bug fixed: functions with &REST arguments sometimes failed with
867 "Undefined function" when compiled with (DEBUG 3). (reported by
869 * bug fix: overflow during compiling of setting element of a bit
870 vector with constant index and value. (reported by Timmy Douglas)
871 * build fix: fixed the dependence on *LOAD-PATHNAME* and
872 *COMPILE-FILE-PATHNAME* being absolute pathnames.
873 * on x86 compiler partially supports stack allocation of dynamic-extent
875 * GO and RETURN-FROM do not check the extent of their exit points
876 when compiled with SAFETY 0.
877 * fixed some bugs related to Unicode integration:
878 ** encoding and decoding errors are now much more robustly
879 handled; it should now be possible to recover even from invalid
880 input or output to the terminal. (thanks to Teemu Kalvas)
881 ** provided a first cut at implementing STRING-TO-OCTETS and
882 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
883 ** altered the SB-MD5 contributed module slightly, changing the
884 interface just enough for it to be supportable for builds where
885 lisp characters are not eight bits.
886 * fixed some bugs revealed by Paul Dietz' test suite:
887 ** the FORMATTER-generated functions for ~V[ conditionals require
888 the correct number of arguments.
889 ** READ-FROM-STRING returns the mandated second value when applied
890 to displaced strings.
891 ** the #\Rubout and #\Backspace characters are treated as invalid
892 constituent characters by the tokenizer.
894 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
895 * new feature: reloading changed shared object files with
896 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
897 * new feature: references to foreign variables and functions
898 can now be compiled and loaded before the corresponding shared
899 object file is loaded, as long as the foreign definitions are
900 available at runtime.
901 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
902 supported. (thanks to Dan Debertin)
903 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
904 just keywords. (thanks to Vincent Arkesteijn)
905 * fixed bug #331: structure-class instances corresponding to
906 DEFSTRUCT forms are now created eagerly.
907 * fixed bug #345: backtraces from calls to undefined functions work
908 on x86 as well. Related bug #61 is now also partially fixed on x86
909 (backtraces from throws to unknown catch tags.)
910 * bug fix: lambda-list parsing is now stricter vrt. order and number
911 of lambda-list keywords.
912 * bug fix: as specified by AMOP, an error is signalled if a
913 class-option appears multiple times in a DEFCLASS form. (reported
915 * bug fix: RANDOM can be compiled when the compiler derives the type
916 of its numeric argument as a disjoint set of small integers.
917 (reported by Paul Dietz)
918 * bug fix: starting a core saved with shared objects loaded when
919 those objects are not available no longer causes threaded SBCL to
920 hang. (reported by Sean Ross)
921 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
923 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
924 argument, which must always be NIL. (reported by Kalle Niemitalo)
925 * bug fix: printing 1.0d+23 no longer results in an error.
926 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
927 * bug fix: structure slot setters preserve evaluation order. (thanks
929 * fixed some bugs related to Unicode integration:
930 ** RUN-PROGRAM can allow its child to take input from a Lisp
931 stream. (reported by Stefan Scholl)
932 ** ASDF-INSTALL successfully downloads .tar.gz files in all
933 locales. (reported by Ken Causey)
934 * fixed some bugs revealed by Paul Dietz' test suite:
935 ** INCF, DECF and REMF evaluate their place form as specified in
937 ** FORMATTER expands ~{ ~} iteration directives with V or #
938 parameters correctly.
939 ** FORMATTER deals with the ~@[ ~] conditional directive where the
940 consequent uses no arguments correctly.
941 ** the system has a partial understanding of the (COMPLEX RATIO)
943 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
946 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
947 * new feature: a build-time option (controlled by the :SB-UNICODE
948 keyword feature, enabled by default) for building the system with
949 support for the entire 21-bit character space defined by the
951 * new feature: the system now has rudimentary external-format
952 support; the primary user-visible change at this time is that
953 characters with the high bit set (such as lower-case-e-acute) will
954 print correctly to a terminal in a UTF-8 environment.
955 * minor incompatible change: BASE-CHAR no longer names a class;
956 however, CHARACTER continues to do so, as required by ANSI.
957 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
958 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
959 has been moved to the SB-EXT package (temporarily re-exported from
961 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
962 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
964 * fixed bug #351: better error-handlind and reporting for malformed
966 * fixed bug #350: bignum-printing is now more memory-efficient,
967 allowing printing of very large bignums, eg. (expt 2 10000000).
968 (reported by Bruno Haible)
969 * fixed bug #302: better primitive-type selection for intersection
971 * fixed bug #308: non-graphic characters now all have names, as
972 required. (reported by Bruno Haible)
973 * bug fix: redefining a class with different superclasses now correctly
974 removes it from the direct-subclasses of its previous superclasses.
975 (reported by David Morse)
976 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
977 new value now works. (reported by Bruno Haible)
978 * bug fix: correct canonicalization of multiple non-standard slot
979 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
980 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
981 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
982 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
983 now exists, an signals an error.
984 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
985 during macroexpansion for non-iteration variables as well. (reported
986 by Bruno Haible for CMUCL)
987 * bug fix: Cyclic structures and unprintable objects in compiler
988 messages no longer cause errors. (reported by Bruno Haible)
989 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
990 and READ-FROM-STRING all now return a primary value of NIL if
991 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
992 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
993 (reported by Bruno Haible for CMUCL)
994 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
995 do not designate a condition. (reported by Bruno Haible for
997 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
998 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
999 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1000 permitting resolution in favour of any of the conflicting symbols.
1001 (reported by Bruno Haible for CMUCL)
1002 * bug fix: EQUAL compiler optimizations is less aggressive on
1003 strings which can potentially compare true despite having distinct
1004 specialized array element types.
1005 * bug fix: unit enumerations can be defined without dividing by
1006 zero. (thanks to Vincent Arkesteijn)
1007 * FORMAT compile-time argument count checking has been enhanced.
1008 (report from Bruno Haible for CMUCL)
1009 * a partial workaround for the bug 262: the compiler does not try to
1010 inline-expand a local function doing RETURN-FROM from a deleted
1011 BLOCK. (thanks to Peter Denno for the bug report and to David
1012 Wragg for the simple test case)
1013 * fixed some bugs revealed by Paul Dietz' test suite:
1014 ** RENAME-PACKAGE allows all package designators as new package
1016 ** constraint propagation and lambda variable substitution are
1017 more cautious in dealing with partially deleted code.
1018 ** compiler sometimes lost reoptimization passes.
1019 ** CERROR, given a condition as condition designator, passes its
1020 remaining arguments to the continue format control without
1022 ** Case-altering FORMAT directives work correctly on non-ASCII
1024 ** The REAL type specifier handles bounds outside the floating
1025 point ranges without signalling FLOATING-POINT-OVERFLOW.
1026 ** Functions with IR1-transformations can create intercomponent
1027 references to global functions.
1028 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1030 ** FORMAT ~R treats a nil value for its first parameter correctly.
1032 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1033 * enhancement: saving cores with foreign code loaded is now
1034 supported on x86/NetBSD and sparc/Linux in addition to the previously
1035 supported platforms.
1036 * bug fix: on some platforms repeated installations caused multiple
1037 copies of HTML documentation to be installed -- should not happen
1038 any more. (reported by Stefan Scholl)
1039 * bug fix: parsing self-recursive alien record types multiple times
1040 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1041 original patch by Helmut Eller for CMUCL)
1042 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1043 (thanks to Richard Kreuter)
1044 * bug fix: defining classes whose accessors are methods on existing
1045 generic functions in other (locked) packages no longer signals
1046 bogus package lock violations. (reported by François-René Rideau)
1047 * bug fix: special variables as DEFMETHOD parameters no longer have
1048 associated bogus type declarations. (reported by David Wragg and
1050 * bug fix: read-write consistency on streams of element-type
1051 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1052 * bug fix: redefiniton of the only method of a generic function with
1053 no DEFGENERIC no longer emits a full WARNING. In addition,
1054 redefinition of generic functions with no DEFGENERIC to an
1055 incompatible lambda list now signals an error. (thanks to Zach
1057 * bug fix: DEFGENERIC now works even when there's a function of the
1058 same name in an enclosing lexical environment. (thanks to Zach
1060 * fixed compiler failure, caused by instrumenting code during
1061 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1062 * optimization: added loop analysis and improved register allocation
1063 * fixed some bugs revealed by Paul Dietz' test suite:
1064 ** POSITION on displaced vectors with non-zero displacement
1065 returns the right answer.
1066 ** (SIMPLE-STRING) is a valid type specifier for sequence
1068 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1070 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1071 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1073 ** The justification version of the FORMAT ~< directive treats
1074 non-zero minpad parameter correctly.
1076 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1077 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1078 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1079 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1080 the supported interface.
1081 * new feature: Single-stepping of code compiled with DEBUG 2 or
1082 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1083 * new feature: saving cores with foreign code loaded is now
1084 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1085 Timothy Moore's work for CMUCL)
1086 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1087 parameters to * instead of NIL if no initform is supplied.
1088 (reported by Johan Bockgaard)
1089 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1090 correctly when FUNCALL appears as the car of the form. Note:
1091 despite this FUNCALL forms are not currently subject to
1092 compiler-macro expansion. (port of Raymond Toy's fix for the
1093 same from CMUCL, reported by Johan Bockgaard)
1094 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1095 (thanks for Teemu Kalvas)
1096 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1097 case instead of terminating the iteration (thanks for Julian
1098 Squires, Sean Champ and Raymond Toy)
1099 * bug fix: incorrect expansion of defgeneric that caused a style
1100 warning. (thanks for Zach Beane)
1101 * bug fix: slot accessor effective method computation works properly
1102 for classes with multiple non-standard applicable methods on
1103 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1104 * on x86 compiler supports stack allocation of results of LIST and
1105 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1106 CMUCL implementation by Gerd Moellmann)
1107 * fixed some bugs revealed by Paul Dietz' test suite:
1108 ** FORMAT strings with both the ~<~:;~> form of the justification
1109 directive and pretty-printing directives cause an error.
1111 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1112 * incompatible change: the internal functions
1113 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1114 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1115 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1116 instead of the old functions.
1117 * new feature: on platforms where "dladdr" is available foreign
1118 function names now appear in backtraces. (based on Helmut Eller's
1120 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1121 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1123 * documentation: ASDF manual is now distributed with SBCL.
1124 * bug fix: SBCL can now load its contributed modules with REQUIRE
1125 even if the system-provided entries have been removed from
1126 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1128 * bug fix: backtraces involving undefined functions or assembly
1129 routines are more informative. (thanks to Brian Downing)
1130 * bug fix: mutually referent alien structures now work correctly.
1131 (reported by Rick Taube)
1132 * bug fix: structures defined by WITH-ALIEN can be referred to
1133 within other definitions in the same WITH-ALIEN.
1134 * bug fix: division operators (MOD, TRUNCATE and the like) with
1135 constant zero divisors and integer dividends no longer generate
1137 * bug fix: provide default methods for INPUT-STREAM-P and
1138 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1139 * bug fix: improve the detection and resolution of MOP metacycles
1140 (where effective method computation depends on the generic function
1141 having its effective method computed). (reported by Bruno Haible)
1142 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1143 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1144 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1145 * bug fix: pathnames with relative directory components can be
1146 represented relative to default pathnames.
1147 * optimization: in taking the GCD of bignums, reduce the two bignums
1148 to approximately the same size (using Euclid's algorithm) before
1149 applying the more sophisticated binary GCD. (thanks to Juho
1151 * optimization: COUNT on bitvectors now operates word-at-a-time.
1152 * optimization: ASH with a positive, but not necessarily constant,
1153 (leftwards) shift, when in a modular context, is compiled to a
1155 * fixed some bugs revealed by Paul Dietz' test suite:
1156 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1157 if the corresponding argument is NIL.
1158 ** FORMAT directives accepting a minpad parameter treat negative
1160 ** literal commas as character parameters to FORMAT directives are
1162 ** literal spaces directly after ~<Newline> directives within a
1163 format-logical-block (~:< ~@:>) do not induce :FILL-style
1164 conditional newlines.
1165 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1166 as specified: it no longer includes conditional newlines).
1167 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1169 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1170 :PER-LINE-PREFIX argument does not evaluate to a string.
1172 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1173 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1174 the manual for details; for now, package locks can be disabled by
1175 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1176 if no major problems are found then it is likely that they will be
1177 compiled in unconditionally.
1178 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1179 unsupported operators on all platforms. To load a shared library
1180 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1181 object file, link it into a shared library outside of SBCL and
1182 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1184 * bug workaround: ROOM T and the GC were not working together
1185 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1186 reporting facility aren't preserved. That reporting has been
1187 disabled, so now until and unless someone figures out how to make it
1188 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1189 * minor incompatible change: as threatened around sbcl-0.8.0, the
1190 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1191 The symbols are also exported from SB-PCL for backwards
1192 compatibility, but more so than before SB-PCL should be treated as
1193 an implementation-internal package.
1194 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1196 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1197 It is known as of this release not to work on the Alpha, however.
1198 * fixed bug #167: errors signalled due to illegal syntax in method
1199 bodies are now more legible.
1200 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1201 designators and can hence be used with TYPEP.
1202 * fixed bug #333: CHECK-TYPE now ensures that the type error
1203 signalled, if any, has the right object to be accessed by
1204 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1205 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1206 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1208 * fixed bug #334: programmatic addition of slots using specialized
1209 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1210 and :ALLOCATION :CLASS slots.
1211 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1212 float exponents. (rereported by Peter Seibel)
1213 * fixed a bug: #\Space (and other whitespace characters) are no
1214 longer considered to be macro characters in standard syntax by
1215 GET-MACRO-CHARACTER.
1216 * fixed bug: initialization of condition class metaobjects no longer
1217 causes an instance of the condition to be created. (reported by
1219 * fixed bug: it is now possible to have slots such that
1220 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1221 description is neither :INSTANCE nor :CLASS.
1222 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1223 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1224 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1225 non-standard SLOT-DEFINITION-ALLOCATION values.
1226 * fixed bugs: various race conditions handling exiting threads.
1227 CL-PPCRE's thread torture test now passes.
1228 * fixed arguably-a-bug: GC time proportional to number of threads in
1229 system even when most of them are idle
1230 * optimization: improved performance of BIT and SBIT on bit-vectors.
1231 * .fasl file incompatibility: The fasl file version number has
1232 been incremented because of changes associated with package locks.
1234 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1235 * minor incompatible change: the system no longer provides
1236 optimization hints (conditions which are TYPEP
1237 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1238 mode; these hints are still emitted when higher SPEED optimization
1240 * new contrib module: a sampling profiler (profiling by statistical
1241 sampling, rather than by instrumenting functions) is available as
1242 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1243 * the behaviour of the standard function ED is now customizeable by
1244 third parties through a hook variable: see ED's documentation
1245 string for information on the protocol.
1246 * the compiler no longer emits efficiency notes for (FUNCALL X)
1247 when the type of X is uncertain under default optimization
1249 * fixed bug 276: mutating a binding of a specialized parameter to a
1250 method to something that is not TYPEP the specializer is now
1252 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1253 exhibit the required behaviour on the x86 platform. (thanks to
1254 Peter van Eynde, Eric Marsden and Bruno Haible)
1255 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1256 even for difficult arguments. (reported by Peter Graves)
1257 * fixed bug 141a: the backquote printer now descends quoted
1259 * fixed another bug in backquote printing: no more destructive
1260 modification of the form's list structure. (reported by Brian
1262 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1263 move between its address being taken and the call to
1264 interrupt_thread, fixing a crashing race condition.
1265 * the SB-POSIX contrib implementation has been adjusted so that it
1266 no longer exhibits ridiculously poor performance when constructing
1267 instances corresponding to C structs.
1269 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1270 * minor incompatible change: the sb-grovel contrib now treats C
1271 structures as alien (in the sense of SB-ALIEN) objects rather than
1272 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1273 has implications for memory management of client code
1274 (sb-grovel-returned objects must now be manually managed) and for
1275 type safety (alien objects now have full types).
1276 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1277 used to control emission of compiler diagnostics, rather than the
1278 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1279 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1280 quality should be considered deprecated.
1281 * install.sh now installs the user manual as well
1282 * (not quite a new documentable feature, but worth considering in
1283 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1284 beginnings of a semantically meaningful condition hierarchy is
1285 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1287 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1288 designator as the defaults argument.
1289 * fixed bug: Displaced arrays whose displaced-to array has become
1290 too small now cause ARRAY-DIMENSION to signal an error, providing
1291 for safer bounds-checking. (reported by Bruno Haible)
1292 * fixed bug: DEFCLASS slot definitions with identical :READER and
1293 :WRITER names now signal a reasonable error. (reported by Thomas
1295 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1297 * fixed bug: Invalid slot specification errors now print correctly.
1298 (thanks to Zach Beane)
1299 * fixed bug 320: Shared to local slot value transfers in class
1300 redefinitions now happen corrently from superclasses as
1301 well. (reported by Bruno Haible)
1302 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1304 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1305 improper lists correctly. (reported by Bruno Haible)
1306 * fixed bug 313: source-transform for <fun-name> was erroneously
1307 applied to a call of a value of a variable with name <fun-name>.
1308 (reported by Antonio Menezes Leitao)
1309 * fixed bug 307: The obsolete instance protocol ensures that
1310 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1311 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1312 remove dead unknown-values globs from the middle of the stack.
1313 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1314 easier to resume long computations after using *BREAK-ON-SIGNALS*
1315 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1316 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1317 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1318 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1319 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1320 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1322 * fixed a (fairly theoretical) bug in string printing: if
1323 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1324 does not have array-element-type equal to the most general string
1326 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1327 function name. (thanks to Nikodemus Siivola)
1328 * fixed bug: compile-time format string checker failed on
1329 non-closed ~{. (reported by Thomas F Burdick)
1330 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1331 don't warn on legal loop code involving more than one aggregate
1332 boolean. (thanks to Nikodemus Siivola)
1333 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1334 in keyword position to MAKE-PACKAGE was not being reported as
1336 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1337 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1339 * fixed bugs 280 and 312: the checking for multiple definitions in a
1340 file is less likely to become confused by uses of inline
1342 * fixed bug: the #S reader macro performs the keyword coercion
1343 specified for slot names. (reported by Kalle Niemitalo)
1344 * fixed bug: lambda lists may contain symbols whose names start with
1345 & but are not lambda-list-keywords; their occurrence triggers a
1347 * fixed bug 321: define-method-combination argument lambda lists do
1348 not cause invalid code to be generated when &optional or &aux
1349 variables have default values. (reported by Bruno Haible)
1350 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1351 have CLOS classes; this appears also to have fixed reported
1352 instability in stack exhaustion detection.
1353 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1354 MUFFLE-WARNING when no associated restart is present is now
1356 * optimization: rearranged the expansion of various defining macros
1357 so that each expands into only one top-level form in a
1358 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1360 * optimization: used a previously little-used slot in symbols to
1361 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1363 * fixed some bugs revealed by Paul Dietz' test suite:
1364 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1366 ** Fixed an optimization bug related to inheritance of initforms
1367 from local to shared slots.
1368 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1369 ** CAST optimizer forgot to flush argument derived type.
1370 ** print/read consistency on floats is now orders of magnitude
1371 more likely. (thanks also to Bruno Haible for a similar report
1373 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1374 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1375 for objects of type REAL. Make it so.
1376 ** SXHASH is specified (infelicitously) to respect similarity,
1377 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1378 it so. (thanks to Markus Ziegler)
1379 ** on the Alpha, the compiler succeeds in compiling functions
1380 returning a known number of arguments greater than 63.
1381 ** fixed handling of invalid NIL arguments in keyword position
1383 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1385 ** provided a readably-printable representation for RANDOM-STATE
1387 ** ensured that pathnames, where they have a namestring, always
1388 print using #P"..." syntax.
1390 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1391 * Support for the forthcoming 2.0 version of the NetBSD kernel
1392 running on x86 hardware has been added. (thanks to Perry
1393 E. Metzger most immediately, and others for their past work)
1394 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1395 current release version; SBCL's OpenBSD support had been broken
1396 since about the time of OpenBSD's switch to ELF binary format.)
1397 * [placeholder for DX summary]
1398 ** user code with &REST lists declared dynamic-extent, under high
1399 speed or space and low safety and debug optimization policy.
1400 * The manual has been converted to Texinfo format and the debugger
1401 chapter from the cmucl manual has been added.
1402 * A facility has been added to extract documentation strings from
1403 sbcl and store them as Texinfo-formatted snippets for inclusion in
1404 the manual (via Texinfo's @include directive)
1405 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1406 the test case to Dave Roberts)
1407 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1408 pointers. (reported by Sean Ross)
1409 * bug fix: PROFILE output is printed nicely even for large numerical
1410 values. (thanks to Zach Beane)
1411 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1412 greater than 32 handle EOF correctly.
1413 * bug fix: on X86 an immediate argument of the IMUL instruction is
1414 correctly printed in disassembly. (thanks to Lutz Euler)
1415 * bug fix: class slots in redefined classes preserve their old
1416 values. (thanks to Bruno Haible and Nikodemus Siivola)
1417 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1418 succeeds. (reported by Marco Baringer)
1419 * fixed some bugs revealed by Paul Dietz' test suite:
1420 ** READ-SEQUENCE now works on ECHO-STREAMs.
1421 ** RATIONALIZE works more according to its specification. (thanks
1424 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1425 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1426 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1427 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1428 should matter to you only if you rebind the printer control
1429 variables and then find you want different bindings in the
1430 debugger than in the ordinary execution of your program.)
1431 * The runtime build system has been tweaked to support building
1432 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1433 assembler and linker. (thanks to Nikodemus Siivola)
1434 * Unbound, undefined, undeclared variables now trigger full
1435 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1436 more often programmer error than deliberate exploitation of undefined
1438 * optimization: the hash algorithm for strings has changed to one
1439 that is less vulnerable to spurious collisions. (thanks to Juho
1441 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1442 needless bounds checking. (thanks to Juho Snellman)
1443 * optimization: implemented multiplication as a modular
1444 (UNSIGNED-BYTE 32) operation on the PPC backend.
1445 * fixed some bugs revealed by Paul Dietz' test suite:
1446 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1448 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1449 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1450 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1452 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1454 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1456 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1457 their output stream on EOF from read.
1458 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1459 have been read to end-of-file.
1460 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1462 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1463 description of determination of which consecutive characters
1465 ** Printing the "Space" character with escaping on now yields "#\\ ",
1466 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1467 ** Reading floating-point numbers with *READ-BASE* set to a number
1468 less than 10 works correctly.
1469 ** Reading floating-point numbers with *READ-BASE* set to a number
1470 more than 10 works correctly.
1471 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1472 the readtable currently in effect.
1474 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1475 * minor incompatible change: parsing of namestrings on a physical
1476 (Unix) host has changed; numbers after the final #\. in a
1477 namestring are no longer interpreted as a version field. This is
1478 intented to be largely invisible to the user, except that the
1479 meaning of the namestring "*.*.*" has changed: it now refers to a
1480 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1481 should usually be replaced by
1482 (make-pathname :name :wild :type :wild :version :wild)
1483 with the added benefit that this is more likely to be portable.
1484 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1485 option to OPEN now signals an error if the file being opened
1486 exists; this may have an impact on existing code.
1487 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1488 signals received, on the PowerPC platforms (both Linux and
1489 Darwin). (thanks to Pierre Mai for pointing out the location of
1491 * several fixes on OS X: The system now builds and runs cleanly on
1492 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1493 (thanks to Brian Mastenbrook)
1494 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1495 second-resolution: integer multiples of 1/3600 between -24 and 24.
1496 (thanks to Vincent Arkesteijn)
1497 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1498 type when called with 1 argument; PEEK-CHAR checked type of
1499 PEEK-TYPE only after having read first character from a
1500 stream. (reported by Peter Graves)
1501 * bug fix: the garbage collector now has much better locality
1502 behaviour, and in particular no longer treats all memory as being
1503 exhausted when many small objects point to each other in a deeply
1505 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1506 recognized as being TYPEP their class.
1507 * bug fix: the PUSHNEW documentation string has been corrected.
1508 (thanks to Vincent Arkesteijn)
1509 * bug fix: defaulting of the value for the last of an atypically
1510 large number of multiple values being bound was not being
1511 performed correctly on the Alpha or PPC platforms
1512 * optimization: implemented multiplication as a modular
1513 (UNSIGNED-BYTE 32) operation on the x86 backend.
1514 * optimization: SEARCH on simple-base-strings can now be open-coded.
1515 (see also contrib/compiler-extras.lisp for inspiration for
1516 teaching the compiler about the Boyer-Moore algorithm).
1517 * value, returned by MAX (and MIN) called with several EQUALP, but
1518 not EQL, arguments now does not depend on compiler settings.
1519 * fixed some bugs revealed by Paul Dietz' test suite:
1520 ** in stack analysis liveness information is propagated from
1521 non-local entry points.
1522 ** pathwise CAST removing failed when the CAST node did not start
1524 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1525 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1527 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1528 host is already defined.
1529 ** RENAME-FILE works on streams instead of signalling an internal
1531 ** PEEK-CHAR uses the current readtable when determining whether
1532 or not a character is whitespace.
1533 ** MERGE-PATHNAMES handles the case when the pathname does not
1534 specify a name while the default-pathname specifies a version.
1535 ** Pathnames now stand a better chance of respecting print/read
1537 ** Attempting to use standardized file system operators with a
1538 pathname with invalid :DIRECTORY components signals a
1540 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1542 ** DIRECTORY on logical pathnames is more correct.
1543 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1544 signal a TYPE-ERROR if their argument is not a stream.
1545 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1546 designator argument does not designate a stream.
1547 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1548 examining the synonym.
1549 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1551 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1552 element-type larger than ([UN]SIGNED-BYTE 32).
1554 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1555 * When built with the :SB-FUTEX feature, threaded builds now take
1556 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1557 for faster/more reliable mutex and condition variable support.
1558 * Incompatible change (but one you probably shouldn't have been using
1559 anyway): the interface and code for arbitrating between multiple
1560 threads in the same user session has been redesigned.
1561 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1562 error when called without an explicit environment argument.
1563 (thanks to Nikodemus Siivola)
1564 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1565 confused about their position. (thanks to Adam Warner and Gerd
1567 * bug fix: Pretty printing backquoted forms with unquotations in the
1568 argument list position of various code constructs such as LAMBDA
1569 now works correctly. (reported by Paul Dietz)
1570 * bug fix: Pretty printing unquotations no longer loses all
1571 stream position information.
1572 * optimization: performance of string output streams is now less
1573 poor for multiple small sequence writes.
1574 * optimization: performance of CSUBTYPEP in the presence of complex
1575 expressions involving CONS and NOT many times has been improved.
1576 (reported by Paul Dietz)
1577 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1579 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1580 streams for the REPL, for the debugger, and so on. (thanks to
1582 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1583 * fixed some bugs revealed by Paul Dietz' test suite:
1584 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1585 is ignored if it is too small, rather than propagating through
1586 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1587 ** extremely complex negations of CONS types were not being
1588 sufficiently canonicalized, leading to inconsistencies in
1590 ** VALUES tranformer lost derived type.
1592 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1593 * fixed a bootstrapping bug: the build process no longer assumes
1594 that the various BOOLE-related constants have the same value in
1595 host and target lisps. (noted by Paul Dietz' test suite on an
1596 SBCL binary built from CLISP)
1597 * The system can now be dynamically linked on the MIPS platform,
1598 which enables dynamic loading of foreign code from Lisp. (thanks
1599 to Ralf Baechle for discussions on the MIPS ABI)
1600 * The system now records debugging information for its own source
1601 files in a filesystem-position-independent manner, relative to
1602 the "SYS" logical host.
1603 * fixed a compiler bug: MV-LET convertion did not check references
1604 to the "max args" entry point. (reported by Brian Downing)
1605 * tweaked disassembly notes to be less confident about proclaiming
1606 some instruction as an LRA. (thanks to Brian Downing)
1607 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1608 now each have their own history, command character, and other
1609 characteristics. (thanks to David Lichteblau)
1610 * fixed some bugs revealed by Paul Dietz' test suite:
1611 ** compiler failure in compiling LOGAND expressions including a
1613 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1614 shift greater than 32.
1615 ** FLUSH-DEST did not mark blocks for type check regeneration.
1616 ** HANDLER-CASE failed to accept declarations in handler clauses
1617 in some circumstances.
1619 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1620 * New code in contrib/sb-introspect (still probably not entirely
1621 stable yet) provides some support for smart Lisp development
1622 environments like SLIME.
1623 * The conditions signalled for errors occurring when loading .fasl
1624 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1625 in a way which should help ASDF recover gracefully.
1626 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1627 stuff has been cleaned up. If you code contrib/ stuff, this might
1628 affect you, and you can look at contrib/README, contrib/STANDARDS,
1629 and/or the 0.8.4.27 diff to check.
1630 * In full calls the compiler now does not generate checks for declared
1631 argument types for all arguments.
1632 * various threading fixes
1633 ** and some experimental patches which didn't make it into
1634 the main tree for this release, but which are shipped in
1635 contrib/experimental-thread.patch as a possible fix for some
1636 failures (deadlock, spinning...) in GC-intensive multithreaded
1638 * fixed PPC build problem (source code incompatibility of different
1639 library versions): added offsetof() hackery which attempts to divine
1640 where glibc maintainers put uc_mcontext today
1641 * fixed bug 282: compiler does not trust type assertions while passing
1642 arguments to a full call.
1643 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1644 &OPTIONAL VALUES type parameter.
1645 * fix bug 214: algorithm for noting rejected templates is now more
1646 similar to that of template seletion. (also reported by rydis on
1648 * fixed bug 141b: printing backquoted information readably and prettily
1649 inserts a space where necessary.
1650 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1651 (CHARACTER 10) are now reported as errors, rather than propagated
1652 as unknown types. (reported by piso on #lisp)
1653 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1654 with values NIL and :ERROR. (thanks to Milan Zamazal)
1655 * fixed bug 191c: CLOS now does proper keyword argument checking as
1656 described in CLHS 7.6.5 and 7.6.5.1.
1657 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1658 longer raise an error; further, using a list as a for-as-arithmetic
1659 counter now raises a meaningful error.
1660 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1661 now understood by sequence creation functions such as MAKE-SEQUENCE
1663 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1664 a STRING-INPUT-STREAM.
1665 * compiler enhancement: SIGNUM is now better able to derive the type
1667 * type declarations inside WITH-SLOTS are checked. (reported by
1669 * fixed some bugs revealed by Paul Dietz' test suite:
1670 ** incorrect optimization of TRUNCATE for a positive first
1671 argument and negative second.
1672 ** compiler failure in let-convertion during flushing dead code.
1673 ** compiler failure while deriving type of TRUNCATE on an
1674 interval, containing 0.
1675 ** ASH of a negative bignum by a negative bignum count now returns
1677 ** intersection of CONS types now canonicalizes properly, fixing
1678 inconsistencies in SUBTYPEP.
1680 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1681 * incompatible change: The --disable-debugger command line
1682 option now clobbers the debugger at a more fundamental
1683 level, by redefining #'INVOKE-DEBUGGER instead of by
1684 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1685 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1686 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1687 enter the debugger REPL and then suffer recursive errors
1688 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1689 while under the new system, BREAK in --disable-debugger mode
1690 terminates the system just as an unhandled error would.
1691 * fixed compiler performance when processing loops with a step >1;
1692 * bug fix: DOCUMENTATION now retrieves generic function
1693 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1694 support has been systematized, and now supports the methods
1695 specified by ANSI, along with a default method and a method for
1696 slot documentation. (reported by Nathan Froyd)
1697 * bug fix: effective methods associated with a generic function are
1698 no longer cached over a change of that generic function's method
1699 combination. (reported by Andreas Fuchs)
1700 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1701 names a non-existent directory. (thanks to Andreas Fuchs)
1702 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1703 the PPC platform, lead to stack corruption; this has been fixed.
1704 (reported by Rainer Joswig)
1705 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1706 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1707 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1708 arithmetic in general, is now much more reliable on the PPC
1710 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1711 platform now returns the right answer.
1712 * optimization: restored some effective method precomputation in
1713 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1714 precomputation is now tunable.
1715 * optimization: compiler-internal data structure use has been
1716 reviewed, and changes have been made that should improve the
1717 performance of the compiler by about 20%.
1718 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1719 simple-base-strings and simple-bit-vectors is improved.
1720 * optimization: the optimization of 32-bit logical and arithmetic
1721 functions introduced in version 0.8.3 on the x86 has been
1722 implemented on the mips, ppc and sparc platforms; an
1723 implementation of the same facility, but for 64-bit arithmetic,
1724 has been added for the alpha.
1725 * microoptimization: the compiler is better able to make use of the
1726 x86 LEA instruction for multiplication by constants.
1727 * bug fix: in some situations compiler did not report usage of
1728 generic arithmetic in (SPEED 3) policy.
1729 * bug 145b fix: compiler used wrong type specifier while converting
1730 MEMBER-types to numeric.
1731 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1733 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1734 index argument now works properly on the Alpha platform.
1735 * bug fix: floating point exception treatment on the Alpha platform
1737 * bug fix: FILE-POSITION works much better on string input and
1738 output streams. (thanks to Nikodemus Siivola)
1739 * bug fix: many threading/garbage collection symptoms sorted.
1740 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1741 might be pseudo-atomic.
1742 * internal change: Stopping for GC is now done with signals not
1743 ptrace. GC is now done in whichever thread wanted it, instead of
1745 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1747 * bug fix: result form in DO is not contained in the implicit
1749 * incompatible change: ICR structure is changed; the value part of
1750 CONTINUATION is now called LVAR; corresponding functions are
1751 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1752 * added type deriver for ISQRT (thanks to Robert E. Brown).
1753 * bug fix: better support for loading from the command line when an
1754 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1756 * fixed some bugs revealed by Paul Dietz' test suite:
1757 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1758 ** ROUND and FROUND now give the right answer when given very
1759 small float arguments.
1760 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1762 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1763 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1764 ** LCM should return a non-negative integer.
1765 ** PARSE-INTEGER returned the index of a terminator instead of the
1766 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1767 ** PARSE-INTEGER returned an incorrect index being applied to a
1769 ** LCM with two arguments of 0 returns 0 rather than signalling
1771 ** unsigned addition of a 32-bit constant with the high bit set no
1772 longer causes an internal compiler error on the x86.
1773 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1774 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1775 with negative last argument.
1776 ** byte specifiers with zero size and position no longer cause
1777 an error during type derivation.
1778 ** bignum multiplication on the Alpha platform now returns the
1780 * porting: The system now builds on SuSE AMD64, although it still
1781 generates a 32-bit binary.
1782 * .fasl file incompatibility: The fasl file version number has
1783 been incremented (because of the changes to internal compiler
1784 data structures referred to above).
1786 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1787 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1788 more accurately, on the Darwin kernel running on PowerPC hardware.
1789 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1790 * Compiler code deletion notes now signal a condition of type
1791 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1792 an associated MUFFLE-WARNING restart.
1793 * The compiler now performs limited argument count validation of
1794 constant format strings in FORMAT, and where appropriate in ERROR,
1795 CERROR and WARN. (thanks to Gerd Moellmann)
1796 * New ASDF-INSTALL contrib can be used for automatic download and
1797 installation of third-party Lisp code from CCLAN or other sites
1799 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1800 which forces another thread to execute a function supplied by the
1802 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1803 now accept and act upon their :ELEMENT-TYPE keyword argument.
1804 (reported by Martin Atzmueller, Edi Weitz)
1805 * bug fix: FILE-POSITION now accepts position designators up to
1806 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1807 is the greater. (thanks to Patrik Nordebo)
1808 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1809 Kalle Olavi Niemitalo)
1810 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1811 printable. (reported by Eric Marsden)
1812 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1813 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1814 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1815 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1816 (thanks to Stig Sandoe)
1817 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1818 UNREAD-CHAR now correctly includes the unread character in the
1819 target sequence. (thanks to Gerd Moellmann)
1820 * bug fix in threaded builds: the system can now be suspended and
1821 resumed by shell job control with minimal disruption.
1822 * bug fixes in times and timezones >2038AD
1823 * better handling of "where is GNU make?" problem in build scripts
1824 (thanks to Nikodemus Siivola)
1825 * new optimization: inside a named function any reference to a
1826 function with the same name is considered to be a self-reference;
1827 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1828 optimization quality.
1829 * new optimization on x86: logical functions and + now have
1830 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1831 used when the result is truncated to 32 bits.
1832 * VALUES declaration is partially enabled.
1833 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1834 * fixed some bugs revealed by Paul Dietz' test suite:
1835 ** The system now obeys the constraint imposed by
1836 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1837 types form a lattice under type intersection.
1838 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1839 ** ASSOC now ignores NIL elements in an alist.
1840 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1841 and (1+ MOST-POSITIVE-FIXNUM) answers.
1842 ** The addition of a method with invalid qualifiers to a generic
1843 function does not cause an error to be signalled immediately;
1844 a warning is signalled, and the error is generated only on
1845 calling the generic function.
1846 * changed .fasl file version number, in part to add type codes for
1847 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1848 obscure ANSI requirements
1850 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1851 * fixed bug 148: failure to inline-expand a local function left
1852 garbage, confusing the compiler.
1853 * fixed bugs 3cd: structure slot readers perform type check if the
1854 slot can have an invalid value (i.e. it is either not initialized
1855 or can be written with a less specific slot writer).
1856 * bug fix: the compiler now traps array references to elements off
1857 the end of an array; previously, the bounds checking in some
1858 circumstances could go off-by-one.
1859 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1861 * type declarations for array element types now obey the description
1862 on the CLHS page "Declaration TYPE", as per discussions on
1863 sbcl-help around 2003-05-08. This means that a declaration
1864 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1865 declaration, all references to BAR will be asserted or assumed
1866 (with THE, so dependent on compiler policy) to involve objects of
1867 type FOO. Note that no such declaration is implied in
1868 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1869 * declared types of functions from the "Conditions"
1870 chapter. (reported by Paul Dietz)
1871 * bug fix: CERROR accepts a function as its first argument.
1872 * bug fix: NTH an NTHCDR accept a bignum as index
1873 arguments. (reported by Adam Warner)
1874 * optimization: character compare routines now optimize comparing
1875 against a constant character. (reported by Gilbert Baumann)
1876 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1877 argument now works properly on the MIPS platform.
1878 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1879 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1880 declared to have a functional type, does not cause an error
1882 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1883 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1884 * fixed bug: if last continuation of a deleted block has a
1885 destination, this destination should be deleted too. (reported by
1887 * fixed a bug in the bootstrap process: the host compiler's values
1888 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1889 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1890 test case from Patrik Nordebo)
1891 * improved the ability of the disassembler on the PPC platform to
1892 provide helpful disassembly notes.
1893 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1894 the class in more cases than previously.
1895 * bug fix: FILE-POSITION now understands :START and :END for
1896 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1897 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1898 8) data. (thanks to David Lichteblau)
1899 * bug fix: it is possible to add a method to a generic function
1900 without lambda list.
1901 * bug fix: reader failed to signal END-OF-FILE inside an
1902 object representation. (reported by Nikodemus Siivola)
1903 * fixed some bugs revealed by Paul Dietz' test suite:
1904 ** LAST and [N]BUTLAST should accept a bignum.
1905 ** condition slot accessors are methods.
1906 ** (VECTOR NIL) is a subtype of STRING.
1908 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1909 * minor incompatible change: some nonsensical specialized lambda
1910 lists (used in DEFMETHOD) which were previously ignored now signal
1912 * minor incompatible change: the system is now aware of the types of
1913 variables in the COMMON-LISP package, and will signal errors for
1914 most violations of these type constraints (where previously they
1915 were silently accepted).
1916 * minor incompatible change: COMPILE-FILE now uses the freedom
1917 afforded (ANSI 3.2.2.3) to use derived function types for
1918 functions defined in the same file. This also permits the system
1919 to warn on static type mismatches and function
1920 redefinition. (Currently it does not work with high DEBUG level.)
1921 * minor incompatible change: VALUES declaration is disabled.
1922 * When issuing notes, the compiler now signals a condition of type
1923 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1924 restart for use in user handlers. It is expected that the
1925 COMPILER-NOTE condition will eventually become a condition
1926 supertype to a hierarchy of note types, which will then be
1927 handleable in a similar fashion. However, at the moment, no such
1928 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1929 but it's an implementation detail, not a classification for the
1931 * Changes in type checking closed the following bugs:
1932 ** type checking of unused values (192b, 194d, 203);
1933 ** template selection based on unsafe type assertions (192c, 236);
1934 ** type checking in branches (194bc).
1935 * A short form of VALUES type specifier has ANSI meaning (it has
1936 increased the number of situations when SBCL cannot perform type
1938 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1939 keywords or constants is permissible.
1940 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1941 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1942 classes. (thanks to Antonio Martinez)
1943 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1944 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1945 argument to a pretty-print stream. Instead, it leaves any such
1946 operations to DESCRIBE-OBJECT methods.
1947 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1949 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1950 lambda lists. (thanks to David Lichteblau)
1951 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1952 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1953 (thanks to Gerd Moellmann)
1954 * fixed bug 246: increased compilation speed of long
1955 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1957 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1959 * DEFCONSTANT now throws a condition of type
1960 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1961 constant to a non-EQL value; CONTINUE and ABORT restarts
1962 respectively change and preserve the value.
1963 * fixed bug 63: The code walker, part of the implementation of CLOS,
1964 is now better at handling symbol macros.
1965 * bug fix: There is no longer an internal implementation type named
1966 CL:LENGTH. (reported by Raymond Toy)
1967 * bug fix: In macro-like defining macros/special operators the
1968 implicit block does not enclose the lambda list.
1969 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1970 enclosed in parentheses) are not suitable as type specifiers, and
1971 their use properly signals an error now.
1972 * bug fix: An argument count mismatch for a type specifier in code
1973 being compiled no longer causes an unhandled error at compile
1974 time, but signals a compile-time warning.
1975 * fixed simple vector readable printing
1976 * bug fix: DESCRIBE takes more care over whether the class
1977 precedence list slot of a class is bound before accessing it.
1978 (reported by Markus Krummenacker)
1979 * bug fix: FORMATTER can successfully compile pretty-printer format
1980 strings which use variants of the ~* directive inside.
1981 * bug fix: SEARCH now applies its TEST predicate to the elements of
1982 the arguments in the correct order. (thanks to Wolfhard Buss)
1983 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1984 in type inference. (thanks to Robert E. Brown)
1985 * bug fix: Reading in symbols with an explicit package name of ""
1986 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1987 * fixed some bugs revealed by Paul Dietz' test suite:
1988 ** NIL is now allowed as a structure slot name.
1989 ** Arbitrary numbers, not just REALs, are allowed in certain
1990 circumstances in LOOP for-as-arithmetic clauses.
1991 ** Multiple class redefinitions before slot access no longer
1992 causes a type error.
1993 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1994 association between the name and a class.
1995 ** Generic functions with non-standard method-combination and over
1996 five methods all of which return constants no longer return NIL
1997 after the first few invocations. (thanks to Gerd Moellmann)
1998 ** CALL-NEXT-METHOD with no arguments now passes the original
1999 values of the arguments, even in the presence of assignment.
2000 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2002 ** DISASSEMBLE works with closures and funcallable instances.
2003 ** ADD-METHOD now returns the generic function, not the new method.
2004 ** FIND-METHOD signals an error if the lengths of the specializers
2005 is incompatible with the generic function, even if the ERRORP
2007 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2008 which its argument is a member.
2009 ** DEFCLASS only redefines the class named by its class-name
2010 argument if that name is the proper name of the class;
2011 otherwise, it creates a new class.
2012 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2013 of the UNBOUND-SLOT condition to the name of the slot.
2014 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2015 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2016 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2017 SLOT-MAKUNBOUND in the specified fashion.
2019 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2020 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2021 cross-compilation host. As a consequence, we can now bootstrap our
2022 way up to SBCL starting with a bare gcc toolchain and human-readable
2023 source code (first the source to CLISP, then the source to SBCL).
2024 * A contributed module containing a partial implementation of the
2025 simple-streams interface has been included. (thanks to Rudi
2027 * A contributed module implementing the RFC1321 Message Digest
2028 Algorithm, known as MD5, has been included.
2029 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2030 no longer has any effect, as the code controlled by this feature
2031 has been deleted. (As far as we know, no-one has ever built using
2032 this feature, and its semantics were confused in any case).
2033 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2034 work on conditions (as required by the ANSI specification),
2035 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2036 expected behaviour on conditions. Users should note, however,
2037 that such behaviour is not required by the ANSI specification,
2038 and so use of this behaviour may render their code unportable.
2039 * fixed some bugs revealed by Paul Dietz' test suite:
2040 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2042 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2043 ** SXHASH on condition objects no longer returns NIL.
2044 ** :ALLOCATION :CLASS slots are better treated; their values are
2045 updated on class redefinition, and initforms inherited from
2046 superclasses are applied.
2047 ** REMOVE-METHOD returns its generic function argument even when
2048 no method was removed.
2049 ** SHARED-INITIALIZE now initializes the values of the requested
2050 slots, including those with :ALLOCATION :CLASS.
2051 ** ALLOCATE-INSTANCE now works on structure classes defined via
2052 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2054 ** SLOT-EXISTS-P now works on conditions, as well as structures
2056 ** MAKE-LOAD-FORM now has the required methods on
2057 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2058 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2059 keyword, and now implements the SLOT-NAMES argument.
2060 ** methods with &OPTIONAL arguments no longer allow too many
2061 arguments to be passed in the call without error.
2062 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2063 option is consistent with the required arguments of the generic
2064 function lambda list.
2065 * bug fix: REQUIRE accepts a string designator. (Thanks to
2067 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2068 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2069 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2070 * bug fix: The compiler now checks for duplicated variables in macro
2072 * bug fix: SETQ on globals returns the correct value.
2073 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2074 not a COMPILER-ERROR (followed by some other strange error on
2075 choosing the CONTINUE restart).
2076 * bug fix: make.sh and friends are now more consistent in the way that
2077 they look for GNU "make".
2079 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2080 * experimental native threads support (on x86 Linux >=2.4 only).
2081 This is not compiled in by default: you need to add :SB-THREAD to
2082 the target features. See the "Beyond ANSI" chapter of the manual
2084 * fix for longstanding nonANSIism: The old distinction between
2085 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2086 The return value from CL:FIND-CLASS is now a CLOS class, and
2087 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2088 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2089 * An interface to the MetaObject Protocol, as described in Kiczales,
2090 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2091 MIT Press, 1991, is available from the SB-MOP package.
2092 * incompatible change: the SB-PCL package should now be considered
2093 a private implementation detail, and no longer a semi-private MOP
2095 * minor incompatible change: due to rearrangement for threads, the
2096 control stack and binding stack are now allocated at arbitrary
2097 addresses instead of being hardcoded per-port. Users affected by
2098 this probably have to be doing advanced things with shared
2099 libraries, and will know who they are.
2100 * minor incompatible change: Previously, all --eval forms used were
2101 processed with READ before any of them were processed with EVAL.
2102 Now each --eval form is processed with both READ and EVAL before
2103 the next --eval form is processed. (Thus package operations like
2104 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2105 work as the user might reasonably expect.)
2106 * minor incompatible change: *STANDARD-INPUT* is now only an
2107 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2109 * minor incompatible change: Y-OR-N-P is now character-oriented,
2110 not line oriented. Also, YES-OR-NO-P now works without errors.
2111 (thanks to Antonio Martinez)
2112 * sb-aclrepl module improvements: an integrated inspector, added
2113 repl features, and a bug fix to :trace command.
2114 * Known functions, which cannot be open coded by the backend, are
2115 considered to be able to check types of their arguments. (fixing
2116 a bug report by Nathan J. Froyd)
2117 * fixed a bug in computing method discriminating functions: It is
2118 now possible to define methods specialized on classes which have
2119 forward-referenced superclasses. (thanks to Gerd Moellmann)
2120 * fixed evaluation order in optional entries (reported by Gilbert
2122 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2123 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2124 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2125 required-by-AMOP NAME argument, as well as CLASS and
2126 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2127 * fixed bug 20: DEFMETHOD can define methods using names that are
2128 not the proper names of classes to designate class specializers.
2129 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2130 Unix file descriptors, instead of blowing up. (thanks to Antonio
2132 * Garbage collection refactoring: user-visible change is that a
2133 call to the GC function during WITHOUT-GCING will not do garbage
2134 collection until the end of the WITHOUT-GCING. If you were doing
2135 this you were probably losing anyway.
2136 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2137 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2138 TYPEP the latter but not the former.
2139 * The compiler issues a full WARNING for calls to undefined functions
2140 with names from the CL package.
2141 * MAP-INTO for a vector destination is open coded. (reported by
2142 Brian Downing on c.l.l)
2143 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2144 documentation string.
2145 * fixed some bugs revealed by Paul Dietz' test suite:
2146 ** COPY-ALIST now signals an error if its argument is a dotted
2148 ** Condition slots are now accessed more correctly in the presence
2149 of multiple initargs for a given slot.
2150 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2151 exclude restarts of the same name associated with a different
2153 ** DEFCLASS of forward-referenced classes with another
2154 forward-referenced class in the superclasses list no longer
2156 ** Condition slots are now initialized once each, not multiple
2157 times. (thanks to Gerd Moellmann)
2158 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2160 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2161 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2162 arguments contain duplicated elements.
2163 ** RESTART-CASE understands local macros.
2164 ** RESTART-CASE associates exactly its own restarts with a condition.
2165 ** ENDP in safe mode checks its argument to be of type LIST.
2166 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2167 in question is unbound.
2168 ** Optimized MAKE-INSTANCE functions no longer cause internal
2169 assertion failures in the presence of duplicate initargs.
2170 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2171 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2172 propagated even in the ctor optimized implementation of
2174 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2176 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2177 * a better implementation of SXHASH on (simple) bit vectors,
2178 measured both in execution speed and in distribution of results
2179 over the positive fixnums, has been installed. Likewise, a better
2180 implementation of EQUAL for simple bit vectors is now available.
2181 * fixed CEILING optimization for a divisor of form 2^k.
2182 * fixed bug 240 (emitting extra style warnings "using the lexical
2183 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2184 by Antonio Martinez)
2185 * fixed SXHASH, giving different results for NIL depending on type
2186 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2187 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2188 variable. (found by Rolf Wester)
2189 * SBCL does not ignore type declarations for special
2190 variables. (reported by rif on c.l.l 2003-03-05)
2191 * some bug fixes in contrib/sb-aclrepl/
2192 * fixed some bugs revealed by Paul Dietz' test suite:
2193 ** a bug in the CONS type specifier, whereby the CAR and CDR
2194 types got intertwined, has been fixed;
2195 ** the type system is now able to reason about the interaction
2196 between INTEGER and RATIO types more completely;
2197 ** APPEND, [N]REVERSE and NRECONC check that those their
2198 arguments, which must be proper lists, are really so;
2199 ** An array specialized to be unable to hold elements has been
2200 implemented, as required -- yes, really -- by ANSI;
2201 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2202 on malformed property lists;
2204 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2205 * incompatible packaging change: in line with Unix convention,
2206 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2207 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2208 * REQUIRE and PROVIDE are now optionally capable of doing something
2209 useful. See the documentation string for REQUIRE.
2210 * infrastructure for a managed SBCL contrib system: contributed
2211 modules in this release include:
2212 ** the ASDF system definition facility;
2213 ** an interface to the BSD Sockets API;
2214 ** an ACL-like convenience interface to the repl;
2215 (thanks to Kevin Rosenberg)
2216 ** an implementation of ROTATE-BYTE, with efficient implementation
2218 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2219 gives a closer approximation to the right answer than previously.
2220 (thanks to Raymond Toy)
2221 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2222 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2223 optional environment arguments, as required by ANSI.
2224 * fixed bugs in other functions taking environment objects, allowing
2225 calls with an explicit NIL environment argument to be compiled
2227 * fixed bug 228: primary return values from
2228 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2229 COMPILE or FUNCTION.
2230 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2231 :TYPE will no longer signal errors on innocuous objects.
2232 * fixed bug 231b: SETQ is better at respecting type declarations in
2233 the lexical environment.
2234 * fixed a bug in DEFCLASS: classes named by symbols with no or
2235 unprintable packages can now be defined.
2236 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2237 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2238 * fixed compiler failure related to checking types of functions.
2239 (reported by Robert E. Brown)
2240 * the compiler is now much more consistent in its error-checking
2241 treatment of bounding index arguments to sequence functions: in
2242 (SAFETY 3) code, errors will be signalled in almost all cases if
2243 invalid sequence bounding indices are passed to functions defined
2244 by ANSI to operate on sequences.
2245 * fixed a bug in the build procedure: documentation of SBCL-specific
2246 packages is now preserved and available in the final Lisp image.
2247 * lifted FDEFINITION lookup out of loops in the implementation of
2248 many list operations. (thanks to Robert E. Brown)
2249 * fixed a bug in the reader: the #n# reader macro now works for
2250 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2251 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2252 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2253 E. Brown and Tony Martinez respectively)
2254 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2255 SLOT-BOUNDP: in particular, it is now able to optimize them much
2256 better, and is now not vulnerable to having packages renamed.
2257 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2258 always called when a slot is not present in an instance. (thanks
2260 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2261 longer signal an error on unreading a character following EOF on
2262 the previous constituent stream. (thanks to Tony Martinez)
2263 * fixed some bugs revealed by Paul Dietz' test suite:
2264 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2265 not just nonnegative fixnums;
2266 ** the logical bit-array operators such as BIT-AND now accept an
2267 explicit NIL for their "opt-arg" argument (to indicate a
2268 freshly-consed result bit-array);
2269 ** ELT now signals an error on an invalid sequence index in safe
2271 ** the type system is now cleverer about negations of numeric
2272 types, and consequently understands the BIGNUM and RATIO types
2274 ** the type system is now cleverer about the interaction between
2275 INTEGER and RATIO types: while bugs still remain, many more
2276 cases are accurately computed;
2277 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2278 if it is in the last clause;
2279 ** CONSTANTLY now correctly returns a side-effect-free function in
2281 ** DECLARE is no longer treated as a special-operator; in
2282 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2283 * incremented fasl file version number due to the change in the
2284 DEFSTRUCT-SLOT-DESCRIPTION structure.
2286 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2287 * minor incompatible change: code processed by the "interpreter" or
2288 EVAL now has a compilation optimization policy of (DEBUG 2)
2289 (changed from (DEBUG 1)) to improve debuggability of interactive
2290 development, and to allow the use of the debug RETURN command in
2292 * an experimental implementation of the RETURN command for the
2293 debugger has been included. (thanks to Frederik Kuivinen)
2294 * fixed bug 62: constraints were not propagated into a loop.
2295 * fixed bug in embedded calls of SORT (reported and investigated by
2297 * fixed some bugs revealed by Paul Dietz' test suite:
2298 ** printing and reading of arrays with some dimensions having
2299 length 0 (thanks to Gerd Moellmann);
2300 ** BOA constructor with &AUX argument without a default value does
2301 not cause a type error;
2302 ** CONSTANTP now returns true for all self-evaluating objects.
2304 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2305 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2306 accessors that are related by inheritance, as specified in the
2307 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2308 Valtteri Vuorikoski)
2309 * The compiler is now able to inline functions that were defined in
2310 a complex lexical environment (e.g. inside a MACROLET).
2311 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2313 * Support for the upcoming FreeBSD-5.0 release has been included.
2314 (thanks to Dag-Erling Smorgrav)
2315 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2316 effect when it is not in a toplevel context.
2317 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2318 stand a better chance of being correct. (thanks to Gerd
2320 * fixed bug in COERCE, which now signals an error on coercing a
2321 rational to a bounded real type which excludes the expected
2323 * The compiler is now able to derive types more accurately from the
2324 COERCE and COMPILE functions.
2325 * fixed bug 223: functional binding is considered to be constant
2326 only for symbols in the CL package.
2327 * fixed bug 231: SETQ did not check the type of a variable being set
2328 (reported by Robert E. Brown)
2329 * A new optimization for MAKE-INSTANCE has been included, fixing
2330 various bugs (including relating to :ALLOCATION :CLASS slots and
2331 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2333 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2334 ** As required by ANSI, LOOP now disallows anonymous collection
2335 clauses such as COLLECT I in conjunction with aggregate boolean
2336 clauses such as THEREIS (= I 1);
2337 ** LOOP now signals an error when any variable is reused in the
2338 same loop (including the potentially useful construct analogous
2339 to WITH A = 1 WITH A = (1+ A);
2340 ** IT is only a special loop symbol within the first clause of a
2341 conditional loop clause;
2342 ** LOOP with a typed iteration variable over a hashtable now
2343 signals a type error iff it should.
2344 * fixed some other bugs revealed by Paul Dietz' test suite:
2345 ** FILE-STREAM now names the class previously known as FD-STREAM;
2346 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2347 argument) no longer signals an error;
2348 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2349 of the slot symbol, rather than using the current package
2350 ((:CONC-NAME "") continues to intern the slot's name in the
2352 * incremented fasl file version number, because of the incompatible
2353 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2354 of the new implementation of DEFINE-COMPILER-MACRO.
2356 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2357 * Support for building SBCL for MIPS platforms running in
2358 little-endian mode has now been checked in, and basic
2359 functionality on said platforms verified.
2360 * minor incompatible change: PCL now records the pathname of a file
2361 in which methods and the like are defined, rather than its
2363 * minor incompatible change: TRUENAME now considers the truename of
2364 a file naming a directory to be the pathname with :DIRECTORY
2365 component indicating that directory.
2366 * minor incompatible change: a NAMED clause in the extended form of
2367 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2368 reason for the previous behaviour is unclear.
2369 * more systematization and improvement of CLOS and MOP conformance
2370 in PCL (thanks to Gerd Moellman and Pierre Mai):
2371 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2373 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2374 primary methods with no specializers;
2375 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2377 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2378 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2379 FORWARD-REFERENCED-CLASSes; error reporting on
2380 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2382 ** SXHASH on CLOS instances now uses a slot internal to the
2383 instance to return different numbers on distinct instances,
2384 while preserving the same return value through invocations of
2386 ** DEFMETHOD signals errors when methods with longer incongruent
2387 lambda lists are added to generic functions;
2388 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2389 CLASS, as specified in AMOP;
2390 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2391 on the order returned by the primary method for classes of
2392 class STANDARD-CLASS;
2393 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2394 * fixed some bugs shown by Paul Dietz' test suite:
2395 ** DOLIST puts its body in TAGBODY;
2396 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2398 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2399 value producing form;
2400 ** if more variables are given to PROGV than values, extra
2401 variables are bound and made to have no value;
2402 ** NSUBSTITUTE on list arguments gets the right answer with
2404 ** ELT signals an error of type TYPE-ERROR when the index argument
2405 is not a valid sequence index;
2406 ** LOOP signals (at macroexpansion time) an error of type
2407 PROGRAM-ERROR when duplicate variable names are found;
2408 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2409 ** FUNCALL of special-operators now cause an error of type
2411 ** PSETQ now works as required in the presence of side-effecting
2412 symbol-macro places;
2413 ** NCONC accepts any object as its last argument;
2414 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2416 ** loop-for-as-package does not require a package to be explicitely
2418 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2420 * fixed bug 166: compiler preserves "there is a way to go"
2421 invariant when deleting code.
2422 * fixed bug 172: macro lambda lists with required arguments after
2423 &REST arguments now cause an error to be signalled. (thanks to
2425 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2426 bug. (thanks to Matthew Danish)
2427 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2429 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2430 arguments are lexically rebound. (thanks to Gerd Moellmann and
2432 * fixed bug 194: error messages are now more informative when there
2433 is no primary method applicable in a call to a generic
2434 function. (thanks to Gerd Moellmann)
2435 * fixed bug in command line argument checking (thanks to Julian
2437 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2438 (thanks to Matthew Danish)
2439 * incremented fasl file version number, because of the
2440 SXHASH-related changes in the layout of CLOS data structures
2442 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2443 * minor incompatible change: The runtime (the Unix executable named
2444 "sbcl") is now much pickier about the .core files it will load.
2445 Essentially it now requires .core files to descend from the same
2446 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2447 as the runtime does. (The intent is to prevent the crashes which
2448 can occur, and which can even be reported as mysterious failures,
2449 when people patch the sources or change the build parameters
2450 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2451 sbcl and .core files.)
2452 * fixed bug: VALUES-LIST is no longer optimized away.
2453 * fixed bug 142: The FFI conversion of C string values to Lisp
2454 string values no longer conses excessively. (thanks to Nathan
2455 Froyd porting Raymond Toy's fix to CMU CL)
2456 * began to systematize and improve MOP conformance in PCL (thanks to
2457 Nathan Froyd, Gerd Moellman and Pierre Mai):
2458 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2460 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2461 ** FINALIZE-INHERITANCE is now called on class finalization;
2462 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2463 argument precedence order.
2464 * fixed bug 202: The compiler no longer fails on functions whose
2465 derived types contradict their declared type.
2466 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2467 so it can be non-toplevel.
2468 * The fasl file version number has changed (because of the new
2469 implementation of DEFMACRO).
2470 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2471 safe code, that any length requirement by their type-specifier
2472 argument is valid. The exceptions to this are described in bug
2474 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2475 functions now check the types of their inputs as required by ANSI.
2476 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2477 introduced symbol is DECLAREd to be SPECIAL.
2478 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2479 * fixed bug: :COUNT argument to sequence functions may be negative
2480 * fixed bug: body of DO-SYMBOLS may contain declarations
2481 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2482 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2483 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2484 symbol macro only once
2485 * fixed printing of call frame when argument list is unavailable
2486 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2487 * compiler no longer signals WARNING on unknown keyword
2490 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2491 * A beta-quality port to the mips architecture running Linux,
2492 based on the old CMUCL backend, has been made. It has been tested
2493 on a big-endian kernel, and works sufficiently well to be able to
2494 rebuild itself; it has not been tested in little-endian mode.
2495 * fixed an inconsistency between gencgc.c and purify.c which made
2496 dumping/loading .core files unreliable
2497 * fixed bug 120a: The compiler now deals correctly with IFs where
2498 the consequent is the same as the alternative, instead of
2499 misderiving the return type. (thanks to Alexey Dejneka)
2500 * fixed bug 113: Logical pathnames are now dumpable (the logical
2501 host is resolved at load-time, throwing an error if it is not
2503 * fixed bug 174: FORMAT's error message is slightly clearer when a
2504 non-printing character is used in a format directive.
2505 * fixed several bugs in compiler checking of type declarations, i.e.
2506 violations of the Python "declarations are assertions" principle
2507 (thanks to Alexey Dejneka)
2508 * fixed several bugs in PCL's error checking (thanks to Gerd
2510 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2512 * fixed bug in compilation of functions as first class values
2513 (thanks to Antonio Martinez-Shotton)
2514 * The compiler's handling TYPE-ERRORs which it can prove will
2515 inevitably happen at runtime has been cleaned up and corrected
2516 in several ways. (thanks to Alexey Dejneka)
2517 * improved argument type checking for various basic arithmetic
2518 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2519 lavished on them in the past that they can be compiled in many
2520 ways in different special cases
2521 * fixed bug 181: compiler checks validity of user supplied type
2523 * cleaned up code flushing in optimization: Function calls which
2524 should signal errors for safety purposes (e.g. which ANSI says
2525 should signal errors when their arguments are of incorrect type)
2526 are no longer optimized away.
2527 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2528 * incremented fasl file version number, because changes in the
2529 implementation of sequence functions like COERCE caused
2530 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2531 in old inline expansions) to become undefined. (Actually these
2532 changes were later undone, so we might very well be binary
2533 compatible with 0.7.7 after all, but leaving the version number
2534 incremented seemed like the simplest and most conservative
2537 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2538 * An alpha-quality port to the parisc architecture running Linux,
2539 based on the old CMUCL backend, has been made. This, even more so
2540 than the other backends, should be considered still a work in
2541 progress; known problems include that the Linux kernel in 64-bit
2542 mode does not propagate the correct sigcontext structure to
2543 userspace, and consequently SBCL on a parisc64 kernel will not
2545 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2546 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2547 suppressed.) Also now that the compiler is looking at declarations
2548 in the environment, it checks optimization declarations as well,
2549 and suppresses inlining when (> DEBUG SPEED).
2550 * More fixes have been made to treatment of floating point exception
2551 treatment and other Unix signals. In particular, floating point
2552 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2553 * The detection and handling of control stack exhaustion (infinite
2554 or very deeply nested recursion) has changed. Stack exhaustion
2555 detection is now done by write-protecting pages at the OS level
2556 and applies at all optimization settings; when found, a
2557 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2558 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2559 be caught using IGNORE-ERRORS.
2560 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2561 correctly when transforming recursive LABELS functions to LETs.
2562 (thanks to Alexey Dejneka)
2563 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2564 that are names of constants or global variables.
2565 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2566 alien routines with docstrings.
2567 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2568 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2570 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2571 (thanks to Eric Marsden)
2572 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2573 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2574 * Bug fix: The SPARC backend can now compile functions involving
2575 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2576 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2577 to a functions expecting a PATHNAME-DESIGNATOR.
2578 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2579 lambda lists. (thanks to Alexey Dejneka)
2580 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2581 functions so that EXPORTing the name of the class doesn't cause
2582 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2583 to Antonio Martinez for reporting this)
2584 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2585 constant-folded without failing an assertion. (thanks to Einar
2586 Floystad Dorum for reporting this)
2587 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2588 types are now dealt with more correctly by the compiler.
2589 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2590 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2591 behaviour analogous to RENAME-FILE. This puts its behaviour more
2592 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2594 * The fasl file version number has changed again. (because of the
2595 bug fix involving the names of PCL MAKE-INSTANCE functions)
2597 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2598 * bug fix: Floating point exceptions are treated much more
2599 consistently on the x86/Linux and PPC/Linux platforms.
2600 * Array initialization with :INITIAL-ELEMENT is now much faster for
2601 cases when the compiler cannot open code the array creation, but
2602 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2603 array accesses have also seen a speed increase.
2604 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2605 specified. (This was at the root of some bad interactions between
2606 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2608 * bug fix: Internal error arguments for undefined functions are now
2609 computed correctly on the PPC/Linux platform.
2610 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2611 Raymond Toy's patch for CMU CL)
2612 * Support for the Solaris 9 operating environment has been included
2613 (thanks to Daniel Merritt)
2614 * A very ugly but hopefully complete draft of the missing FFI chapter
2615 of the manual has been created by reformatting the corresponding
2616 CMU CL manual chapter into (currently very ugly and incoherent)
2617 DocBook and bringing it up to date for SBCL behavior. Thus, the
2618 manual is now essentially complete, at least by my extreme
2619 once-and-only-once standards, whereby it's acceptable to refer to
2620 the doc strings of SB-EXT functions as the primary documentation.
2621 * The fasl file version number has changed again, due to cleanup of
2622 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2623 is no longer a static symbol.)
2625 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2626 * SBCL now builds with OpenMCL (version 0.12) as the
2627 cross-compilation host; also, more progress has been made toward
2628 bootstrapping under CLISP.
2629 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2631 * bug 158 fixed: The compiler can now deal with integer loop
2632 increments different from 1; fixing this turned out also to fix
2634 * bug 169 fixed: no more bogus warnings about using lexical bindings
2635 despite the presence of perfectly good SPECIAL declarations (thanks
2636 to David Lichteblau)
2637 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2638 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2639 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2641 * bug 180 fixed: Method combination specifications no longer ignore
2642 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2643 * bug fix: Structure type predicate functions now check their argument
2644 count as they should.
2645 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2646 correctly. (thanks to Pierre Mai)
2647 * minor incompatible change: The --noprogrammer option is deprecated
2648 in favor of the new --disable-debugger option, which is very similar.
2649 (The major difference is that it takes effect at a slightly different
2650 time at startup, causing handling of errors in --sysinit and
2651 --userinit files will be affected differently.) The
2652 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2653 been added to allow this functionality to be controlled from ordinary
2654 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2655 maintainers, who might want to run non-interactive scripts to
2656 build SBCL cores which will later be used interactively.)
2657 * minor incompatible change: The LOAD function no longer, when given
2658 a wild pathname to load, loads all files matching that pathname.
2659 Instead, an error of type FILE-ERROR is signalled.
2661 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2662 * bug 147 fixed: The compiler preserves its block link/count
2663 invariants more correctly now so that it doesn't crash. (thanks
2665 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2667 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2668 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2669 no-value-for-this-character as NIL (as specified by ANSI).
2670 * HOST-NAMESTRING on physical pathnames now returns a string that is
2671 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2672 (thanks to Christophe Rhodes)
2673 * The Alpha port handles icache flushing more correctly. (thanks to
2675 * More progress has been made toward bootstrapping under CLISP. (thanks
2676 to Christophe Rhodes)
2677 * The fasl file format has changed again, because dynamic loading
2678 on OpenBSD (which has non-ELF object files) motivated some cleanups
2679 in the way that foreign symbols are transformed and passed around.
2680 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2681 is no longer treated as whitespace by the reader, but instead as
2682 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2683 |A
\7fB|, instead of A as it used to.
2685 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2686 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2687 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2688 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2689 this as of 0.7.1.45, but was left out of the previous news section
2690 (thanks to Dan Barlow)
2691 * SBCL now runs on the Solaris operating system on SPARC architectures
2692 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2693 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2694 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2695 * SPARC backend cleanups, allowing builds of cores optimized for V8
2696 and V9 SPARCS, and also emission of code targeted to a particular
2697 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2699 * SBCL is closer to bootstrapping under CLISP, thanks to various
2700 fixes by Christophe Rhodes.
2701 * The fasl file format has changed again, to allow the compiler's
2702 INFO database to support symbol macros.
2703 * The user manual (in doc/) is formatted into HTML more nicely.
2704 (thanks to coreythomas)
2705 * The system is smarter about SUBTYPEP relationships, especially
2706 those involving NOT types (including types such as ATOM which are
2707 represented internally using NOT types). Thus SUBTYPEP is less
2708 likely to return (VALUES NIL NIL) in general, and in particular
2709 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2711 * The fasl file format has changed again, because the internal
2712 representation of types now includes a new slot to support the new
2713 SUBTYPEP-of-NOT-types logic.
2714 * (not a change in the main branch of SBCL, but a related prototype
2715 which can hopefully be merged into the main branch of SBCL in the
2716 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2717 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2718 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2719 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2720 dispatch macro character combinations. (thanks to Alexey Dejneka)
2721 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2722 that superficially look like logical namestrings correctly.
2723 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2725 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2726 default supplied-p) for &optional and &key arguments. (thanks to
2729 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2730 * incompatible change: The compiler is now less aggressive about
2731 tail call optimization, doing it only when (> SPACE DEBUG) or
2732 (> SPEED DEBUG). (This is an incompatible change because there are
2733 programs which relied on the old CMU-CL-style behavior to optimize
2734 away their unbounded recursion which will now die of stack overflow.)
2735 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2736 for non-GENCGC systems has been increased to 20M (since that
2737 seems much closer to the likely performance optimum for modern
2738 systems than the old 4M value was)
2739 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2740 and *DEBUG-PRINT-LEVEL*
2741 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2742 of CMU CL's support for SPARC, and various endianness and other
2743 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2744 * new syntactic sugar for the Unix command line: --load foo.bar is now
2745 an alternate notation for --eval '(load "foo.bar")'.
2747 ** The system now detects stack overflow and handles it gracefully,
2748 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2749 optimization settings. (This is a good thing in general, and
2750 its introduction in this version should be particularly timely
2751 for anyone whose code fails because of suppression of tail
2753 ** The system now hunts for the C variable "environ" in a more
2754 devious way, to avoid segfaults when the C library version differs
2755 between compile time and run time. (thanks to Christophe Rhodes)
2756 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2757 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2758 ** The compiler no longer issues bogus style warnings for undefined
2759 classes in the same source file as the DEFCLASSes which defined
2760 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2762 ** fixes in CONDITION class precedence list for undefined function
2763 errors (thanks to Alexei Dejneka)
2764 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2765 correctly. (thanks to Dan Barlow)
2766 ** portability fixes aiming at bootstrapping under CLISP (thanks
2767 to Dave McDonald and Christophe Rhodes)
2768 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2769 ** fixes in type translation and and type inference (thanks to
2771 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2772 ** various fixes in the new ports (thanks to Dan Barlow)
2773 * several changes related to debugging:
2774 ** suppression of tail recursion, as noted above
2775 ** stack overflow detection, as noted above
2776 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2777 is now the default. (For some time encapsulation has been more
2778 reliable than the breakpoint-based :ENCAPSULATE NIL
2779 implementation, at least on X86 systems; and I just noticed that
2780 encapsulation also seems closer to the spirit of the ANSI
2783 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2785 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2786 up properly again. (There was a packaging bug in 0.7.0 which
2787 left their definitions in SB-SYS::LOAD-FOREIGN and
2788 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2789 vital for most things which interface to C-level interfaces,
2790 like extensions working with sockets or databases or
2791 Perl-compatible regexes or whatever, and the need to fix
2792 this bug is the main reason that 0.7.1 was released so
2794 ** DEFGENERIC is now choosier about the methods it redefines, so that
2795 reLOADing a previously-LOADed file containing DEFGENERICs does
2796 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2797 cycle now works as it should. (thanks to Alexey Dejneka)
2798 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2799 fixed by Christophe Rhodes. (He actually submitted this patch
2800 months ago, and I delayed until after 0.7.0.)
2801 ** Bug 111 (internal compiler confusion about runtime checks on
2802 FUNCTION types) was fixed by Alexey Dejneka.
2803 * Some internal cleanups (getting rid of variables which aren't
2804 needed now that the byte interpreter is gone) caused the fasl
2805 file format number to change again.
2807 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2808 * major incompatible change: The default fasl file extension, i.e. the
2809 default extension for files produced by COMPILE-FILE, has changed
2810 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2812 ** There are many changes in the implementation of the compiler.
2813 SBCL is now essentially a compiler-only implementation of ANSI
2814 Common Lisp. EVAL still "interprets" a few special cases, but
2815 almost all the interesting cases are handled by creating
2816 a LAMBDA expression, calling COMPILE on it, then calling
2817 FUNCALL on the result.
2818 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2819 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2820 Since the code is newer, there might still be some new bugs
2821 (though not as many as before Martin Atzmueller's fixes:-). But
2822 the new code is substantially simpler and clearer, and hopefully
2823 any remaining bugs will be simpler, less fundamental, and more
2824 fixable then the bugs in the old code.
2825 ** The revised compiler is still a little unsteady on its feet.
2827 *** The debugging information it produces (particularly the names
2828 of FUNCTION objects) is sometimes much less useful than what
2829 the old compiler produced.
2830 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2831 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2832 within a MACROLET) has been temporarily weakened.
2833 ** There are new compiler optimizations for various functions:
2834 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2835 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2836 *** the math functions TRUNCATE, FLOOR, and CEILING
2837 *** the function-of-all-trades COERCE
2838 Mostly these should be transparent, but there's one
2839 potentially-annoying problem (bug 117): when the compiler
2840 inline-expands a function and does type analysis on the result,
2841 it can create control paths which have type mismatches, and
2842 when it can't prove that those control paths aren't taken,
2843 it will issue WARNINGs about the type mismatches. This is
2844 a particular problem in practice for the new sequence functions.
2845 It's not clear how this should be fixed, and for now, a
2846 workaround is given in the entry for 117 in the BUGS file.
2847 ** (Because of the interaction between the two previous items --
2848 occasional inlining problems and new inline expansions -- some
2849 of the new sequence function optimizations won't really kick in
2850 completely until debugging information, and then inlining, are
2851 straightened out in some future version.)
2852 * minor incompatible changes:
2853 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2854 DIRECTORY no longer implicitly promotes NIL slots of its
2855 pathname argument to :WILD. In particular, when you ask for the
2856 contents of a directory (which you used to be able to do without
2857 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2858 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2859 ** changes in behavior that ANSI explicitly defines to be
2860 implementation dependent:
2861 *** The new compiler-only implementation still conforms with ANSI,
2862 but acts a little different than before. Besides the obvious
2863 changes in performance tradeoffs (that the cost per form passed
2864 to EVAL has gone up, and the cost per form executed by EVAL
2865 has gone down), the behavior of the system changes a little
2866 because there are no longer any interpreted function objects.
2867 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2868 e.g. doing COMPILE on the output of interactive DEFUN is
2870 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2872 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2873 has changed from (:CL) to NIL.
2874 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2875 e.g. #\Newline and #\Tab, don't have names specified in the
2876 ANSI Common Lisp standard, is now based on their ASCII symbolic
2877 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2878 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2879 are no longer used for output.
2880 ** changes in internal implementation constants:
2881 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2882 4 million. (If your application spends a lot of time GCing and
2883 you have a lot of RAM, you might want to experiment with
2884 increasing it even more.)
2885 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2886 However, almost all old code should still continue to work without
2887 immediate update, as SB-C-CALL is now a (deprecated) nickname
2889 ** Old operator names in the style DEF-FOO are now deprecated in
2890 favor of new corresponding names DEFINE-FOO, for consistency with
2891 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2892 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2893 internal symbols, but a few supported extensions like
2894 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2895 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2896 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2897 etc. as you get deeper into recursive calls to the debugger
2898 command loop, instead of the old "5]", "5]]", "5]]]"
2899 sequence. (I was motivated to do this when squabbles between
2900 ILISP and SBCL left me very deeply nested in the debugger. In the
2901 short term, this change will probably provoke more ILISP/SBCL
2902 squabbles, but hopefully it will be an improvement in the long run.)
2903 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2904 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2905 function, since declaiming return types involving aliens is
2906 (1) annoyingly messy to do by hand and (2) vital to efficient
2907 compilation of code which calls such functions.
2908 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2909 longer reexported by the SB-EXT package. They're solely useful
2910 for alien code, so it seems more logical that you should get
2911 them from the SB-ALIEN package, not in SB-EXT.
2912 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2913 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2914 features. Instead, the code that they used to control is always
2915 built into the system.
2916 * many other bug fixes
2917 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2918 advantage of the new EVAL-WHEN stuff and to clean them up in
2919 general, and they are now more ANSI-compliant in a number of
2920 ways. Martin Atzmueller is responsible for a lot of this.
2921 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2923 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2924 *** correct ERROR type for various file operations
2925 *** some fixes for Lisp streams
2926 *** DEFMETHOD syntax checking
2927 *** changing old weird representation of debug information as
2928 strings (which, among their other deficiencies, don't transform
2929 correctly when you rename packages, and don't change their
2930 print representation when you change things like *PACKAGE*
2931 and *PRINT-LENGTH*) to symbols and lists of symbols
2932 He also made several improvements and fixed several bugs in DESCRIBE.
2933 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2935 *** misbehavior of WRITE-STRING/WRITE-LINE
2936 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2937 and several other LOOP problems as well
2938 *** DIRECTORY when similar filenames are present
2939 *** DEFGENERIC with :METHOD options
2940 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2941 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2942 *** argument ordering in FIND with :TEST option
2943 *** mishandled package designator argument in APROPOS-LIST
2944 *** various problems in the backquote readmacro
2945 *** a bug in APROPOS
2946 *** probably some others that I'm not describing very well here,
2947 since the CVS log documents them by reference to sbcl-devel
2948 messages, and the SourceForge archives aren't working well.:-(
2949 ** Dan Barlow improved the Alpha port (and is making progress on the
2950 PPC port, for those of you who think different).
2951 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2952 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2953 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2954 continues to work on the SPARC port (for those of you in a position
2955 to look down upon our little PC-compatible boxes from a great height).
2956 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2957 and uses that copy, rather than the current dynamic value, when
2958 it's trying to decide whether to truncate output. Thus e.g.
2959 (let ((*print-lines* 50))
2960 (pprint-logical-block (stream nil)
2962 (let ((*print-lines* 8))
2963 (print (aref possiblybigthings i) stream)))))
2964 should now truncate the logical block only at 50 lines, instead of
2965 often truncating it at 8 lines, as it did before.
2966 * The doc/cmucl/ directory, containing old CMU CL documentation
2967 from the time of the fork, is no longer part of the base system.
2968 SourceForge has shut down its anonymous FTP service, and with it
2969 my original plan for distributing the old CMU CL documentation
2970 there. For now, if you need these files you can download an old
2971 SBCL source release and extract them from it.
2972 * The fasl file version number changed again, for dozens of reasons,
2973 some of which are apparent above.
2975 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2976 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2977 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2978 patch, so that DISASSEMBLE output is much nicer.
2979 * The code in the SB-PROFILE package now seems reasonably stable.
2980 I still haven't decided what the final interface should look like
2981 (I'd like PROFILE to interact cleanly with TRACE, since both
2982 facilities use function encapsulation) but if you have a need
2983 for profiling now, you can probably use it successfully with
2984 the current CMU-CL-style interface.
2985 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2986 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2987 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2988 link now returns the dangling link itself, and for similar
2989 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2990 link itself. (In these cases the old code signalled an error and
2991 looped endlessly, respectively.) Thus, DIRECTORY now works even
2992 in the presence of dangling and cyclic symbolic links.
2993 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2994 is now a supported extension again, since the consensus on
2995 sbcl-devel was that it can be useful for ordinary development
2996 work, not just for debugging SBCL itself.
2997 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2998 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2999 that currently-defined functions might be redefined later with
3000 different return types.
3001 * Hash tables can be printed readably, as inspired by CMU CL code
3002 of Eric Marsden and SBCL code of Martin Atzmueller.
3003 * better error handling in CLOS method combination, thanks to
3004 Martin Atzmueller porting Pierre Mai's CMU CL patches
3005 * more overflow fixes for >16Mbyte I/O buffers
3006 * A bug in READ has been fixed, so that now a single Ctrl-D
3007 character suffices to cause end-of-file on character streams.
3008 In particular, now you only need one Ctrl-D at the command
3009 line (not two) to exit SBCL.
3010 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3012 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3013 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3014 does the right thing.
3015 * The implementation of some type tests, especially for CONDITION
3016 types, is now tidier and maybe faster, due to CMU CL code
3017 originally by Douglas Crosher, ported by Martin Atzmueller.
3018 * Some math functions have been fixed, and there are new
3019 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3020 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3021 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3022 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3023 likely to be built into sbcl-0.7.0.)
3024 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3025 CL, ported to SBCL by Martin Atzmueller)
3026 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3027 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3028 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3029 * There's a new slam.sh hack to shorten the edit/compile/debug
3030 cycle for low-level changes to SBCL itself, and a new
3031 :SB-AFTER-XC-CORE target feature to control the generation of
3032 the after-xc.core file needed by slam.sh.
3033 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3034 COMPILE-FILE is no longer supported, so that now every function
3035 gets an entry point, so that block compilation looks a little
3036 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3037 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3038 returns the number of bytes consed since the system started,
3039 rather than the number consed since the first time the function
3040 was called. (The new definition parallels ANSI functions like
3041 CL:GET-INTERNAL-RUN-TIME.)
3042 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3043 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3044 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3045 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3046 * Fasl file version numbers are now independent of the target CPU,
3047 since historically most system changes which required version
3048 number changes have affected all CPUs equally. Similarly,
3049 the byte fasl file version is now equal to the ordinary
3052 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3053 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3054 is no longer recognized. I apologize for this, because it was
3055 listed in SB-EXT as a supported extension, but I found that
3056 its existing behavior was poorly specified, as well as incorrectly
3057 specified, and it looked like too much of a mess to straighten it
3058 out. I have enough on my hands trying to get ANSI stuff to work..
3059 * many patches ported from CMU CL by Martin Atzmueller, with
3060 half a dozen bug fixes in pretty-printing and the debugger, and
3061 half a dozen others elsewhere
3062 * fixed bug 13: Floating point infinities are now supported again.
3063 They might still be a little bit flaky, but thanks to bug reports
3064 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3065 as flaky as they were.
3066 * The --noprogrammer command line option is now supported. (Its
3067 behavior is slightly different in detail from what the old man
3068 page claimed it would do, but it's still appropriate under the
3069 same circumstances that the man page talks about.)
3070 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3071 are now supported, and enabled by default. Thus, the compiler can
3072 handle many floating point and complex operations much less
3073 inefficiently. (Thus e.g. you can implement a complex FFT
3075 * The compiler now detects type mismatches between DECLAIM FTYPE
3076 and DEFUN better, and implements CHECK-TYPE more correctly, and
3077 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3078 thanks to patches from Martin Atzmueller.
3079 * various fixes to make the cross-compiler more portable to
3080 ANSI-conforming-but-different cross-compilation hosts (notably
3081 Lispworks for Windows, following bug reports from Arthur Lemmens)
3082 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3083 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3084 fixed thanks to Pierre Mai's CMU CL patches.
3085 * fixes to keep the system from overflowing internal counters when
3086 it tries to use i/o buffers larger than 16M bytes
3087 * fixed bug 45a: Various internal functions required to support
3088 complex special functions have been merged from CMU CL sources.
3089 (When I was first setting up SBCL, I misunderstood a compile-time
3090 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3091 * improved support for type intersection and union, fixing bug 12
3092 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3093 more obscure bugs as well
3094 * some steps toward byte-compiling non-performance-critical
3095 parts of the system, courtesy of patches from Martin Atzmueller
3096 * Christophe Rhodes has made some debian packages of sbcl at
3097 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3098 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3099 stable, but are nonetheless usable. When he's ready, I'd be happy
3100 to add them to the SourceForge "File Releases" section. (And if
3101 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3102 * new fasl file format version number (because of changes in
3103 internal representation of (OR ..) types to accommodate the new
3104 support for (AND ..) types, among other things)
3106 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3107 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3109 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3111 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3112 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3113 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3114 * DESCRIBE now works on structure objects again.
3115 * Most function call argument type mismatches are now handled as
3116 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3117 know whether the function will be redefined before the call is
3118 executed. (The compiler could flag local calls with full WARNINGs,
3119 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3120 it doesn't keep track of enough information to know whether calls
3121 are local in this sense.)
3122 * Compiler output is now more verbose, with messages truncated
3123 later than before. (There should be some supported way for users
3124 to override the default verbosity, but I haven't decided how to
3125 provide it yet, so this behavior is still controlled by the internal
3126 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3127 src/compiler/ir1util.lisp.)
3128 * Fasl file format version numbers have increased again, because
3129 support for the Gray streams extension changes the layout of the
3130 system's STREAM objects.
3131 * The Gray subclassable streams extension now works, thanks to a
3132 patch from Martin Atzmueller.
3133 * The full LOAD-FOREIGN extension (not just the primitive
3134 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3135 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3136 but like most other programs, it defaults to copying the Unix
3137 environment from the original process instead of starting the
3138 new process in an empty environment.
3139 * Extensions which manipulate the Unix environment now support
3140 an :ENVIRONMENT keyword option which doesn't smash case or
3141 do other bad things. The CMU-CL-style :ENV option is retained
3142 for porting convenience.
3143 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3144 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3146 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3148 * A patch from Martin Atzmueller seems to have solved the SIGINT
3149 problem, and as far as we know, signal-handling now works cleanly.
3150 (If you find any new bugs, please report them!)
3151 * The system no longer defaults Lisp source file names to types
3152 ".l", ".cl", or ".lsp", but only to ".lisp".
3153 * The compiler no longer uses special default file extensions for
3154 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3155 seems to expect a single default extension for all compiled code,
3156 and there's no compelling reason to try to stretch the standard
3157 to allow two different extensions.) Instead, byte-compiled files
3158 default to the same extension as native-compiled files.
3159 * Fasl file format version numbers have increased again, because
3160 a rearrangement of internal implementation packages made some
3161 dumped symbols in old fasl files unreadable in new cores.
3162 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3163 many fewer weird special cases.
3164 * Bug #17 (differing COMPILE-FILE behavior between logical and
3165 physical pathnames) has been fixed, and some related misbehavior too,
3166 thanks to a patch from Martin Atzmueller.
3167 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3168 by Tim Moore, ported to SBCL by Martin Atzmueller.
3169 * Martin Atzmueller fixed several filesystem-related problems,
3170 including bug #36, in part by porting CMU CL patches, which were
3171 written in part by Paul Werkowski.
3172 * More compiler warnings in src/runtime/ are gone, thanks to
3173 more patches from Martin Atzmueller.
3174 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3177 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3179 * DESCRIBE now works on CONDITION objects.
3180 * The debugger now handles errors which arise when trying to print
3181 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3183 * The build system now uses an additional file, customize-target-features.lisp,
3184 to allow local modifications to the target *FEATURES* list. (The point of
3185 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3186 debugging features enabled, without having to worry about propagating it
3187 into everyone's system when I do a "cvs update".) When no
3188 customize-target-features.lisp file exists, the target *FEATURES* list
3189 should be constructed the same way as before.
3190 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3191 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3192 warn about behavior which is undefined under the ANSI spec. Note: This
3193 is specified by ANSI, but it's not very popular with programmers.
3194 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3195 macro in the SBCL sources for an example of a workaround which you
3196 might use to make portable ANSI-standard code which does what you want.
3197 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3198 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3199 when it's used as a non-toplevel form. (This is required in order
3200 to implement the DEFCONSTANT-EQX macro.)
3201 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3202 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3203 funny things, due to bugs in the implementation of EVAL-WHEN
3204 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3205 non-toplevel DEFCONSTANTs are uncommon.)
3206 * The core file version number and fasl file version number have been
3207 incremented, because the old noncompliant DEFCONSTANT behavior involved
3208 calling functions which no longer exist, and because I also took the
3209 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3210 * fixed bug 1 (error handling before read-eval-print loop starts), and
3211 redid debugger restarts and related debugger commands somewhat while
3213 ** The QUIT debugger command is gone, since it did something
3214 rather different than the SB-EXT:QUIT command, and since it never
3215 worked properly outside the main toplevel read/eval/print loop.
3216 Invoking the new TOPLEVEL restart provides the same functionality.
3217 ** The GO debugger command is also gone, since you can just invoke
3218 the CONTINUE restart directly instead.
3219 ** The TOP debugger command is also gone, since it's redundant with the
3220 FRAME 0 command, and since it interfered with abbreviations for the
3222 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3224 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3225 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3226 some porting work by Martin Atzmueller)
3227 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3228 with a boatload of patches which he ported to SBCL. Now that those
3230 ** The system tries to make sure that its low-priority messages
3231 are prefixed by semicolons, to help people who like to use
3232 syntax highlighting in their ILISP buffer. (This patch
3233 was originally due to Raymond Toy.)
3234 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3235 patches originally written by Raymond Toy.
3236 ** The compiler understands coercion between single-value and
3237 multiple-VALUES type expressions better, getting rid of some very
3238 weird behavior, thanks to patches originally by Robert MacLachlan
3239 and Douglas Crosher.
3240 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3241 lambda lists, thanks to a patch originally by Pierre Mai.
3242 ** The system no longer bogusly warns about "abbreviated type
3244 ** The compiler gets less confused by inlining and RETURN-FROM,
3245 thanks to some patches originally by Tim Moore.
3246 ** The system no longer hangs when dumping circular lists to fasl
3247 files, thanks to a patch originally from Douglas Crosher.
3248 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3249 undefined function error.
3250 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3251 in structure slot initforms) for now, documented workaround instead:-|
3252 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3253 * fixed bug 5: added stubs for various Gray stream functions called
3254 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3255 installed, at least appropriate type errors are generated
3256 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3257 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3258 consistently in DEFMETHOD forms.
3259 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3260 it doesn't seem to affect SBCL after all
3261 * The C runtime system now builds with better optimization and many
3262 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3264 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3266 * The system is now under CVS at SourceForge (instead of the
3267 CVS repository on my home machine).
3268 * The new signal handling code has been tweaked to treat register
3269 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3270 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3271 downstream, e.g. in debug-int.lisp, has implicit dependencies
3272 on the unsignedness of integer representation of machine words,
3273 and that caused the system to bomb out with infinite regress
3274 when trying to recover from type errors involving signed values,
3275 e.g. (BUTLAST '(1 2 3) -1).)
3276 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3277 (This was one of the bugs Peter Van Eynde reported back in July.)
3278 * The system now uses code inspired by Colin Walters' O(N)
3279 implementation of MAP (from the cmucl-imp@cons.org mailing
3280 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3281 inline the MAP operation, and there is more than one
3282 sequence argument to the MAP call (so that it can't just
3283 do ETYPECASE once and for all based on the type of the
3284 single sequence argument). (The old non-inline implementation
3285 of the general M-argument sequence-of-length-N case required
3286 O(M*N*N) time when any of the sequence arguments were LISTs.)
3287 * The QUIT :UNIX-CODE keyword argument has been renamed to
3288 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3289 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3291 * Raymond Wiker's patches to port dynamic loading from Linux to
3292 FreeBSD have been added.
3293 * The BUGS file is now more nearly up to date, thanks in large part
3294 to Martin Atzmueller's review of it.
3295 * The debugger now flushes standard output streams before it begins
3296 its output ("debugger invoked" and so forth).
3297 * The core version number and fasl file version number have both
3298 been incremented, because of incompatible changes in the layout
3300 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3301 used to not be called for a saved Lisp image.)
3302 * Martin Atzmueller's version of a patch to fix a compiler crash,
3303 as posted on sbcl-devel 13 September 2000, has been installed.
3304 * Instead of installing Martin Atzmueller's patch for the
3305 compiler transform for SUBSEQ, I deleted the compiler transform,
3306 and transforms for some similar consing operations.
3307 * A bug in signal handling which kept TRACE from working on OpenBSD
3309 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3310 away by constant folding
3311 * The system now defines its address space constants in one place
3312 (in the Lisp sources), and propagates them automatically elsewhere
3313 (through GENESIS and the sbcl.h file). Therefore, patching the
3314 address map is less unnecessarily tedious and error-prone. The
3315 Lisp names of address space constants have also been systematized.
3316 * CVS tags like dollar-Header-dollar have been removed from
3317 the sources, because they have never saved me trouble and
3318 they've been source of trouble working with patches and other
3319 diff-related operations.
3320 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3321 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3323 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3325 * The system has been ported to OpenBSD.
3326 * The system now compiles with a simple "sh make.sh" on the systems
3327 that it's supported on. I.e., now you no longer need to tweak
3328 text in the target-features.lisp-expr and symlinks in src/runtime/
3329 by hand, the make.sh takes care of it for you.
3330 * The system is no longer so grossly inefficient when compiling code
3331 involving vectors implemented as general (not simple) vectors (VECTOR T),
3332 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3333 which dares to use the various sequence functions on non-simple
3334 vectors, takes less of a performance hit.
3335 * There is now a primitive type predicate VECTOR-T-P
3336 to test for the (VECTOR T) type, so that e.g.
3337 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3338 can now be compiled with some semblance of efficiency. (The old code
3339 turned the type declaration into a full call to %TYPEP at runtime!)
3340 * AREF on (VECTOR T) is still not fast, since it's still compiled
3341 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3342 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3343 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3344 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3345 without making full calls to SUBTYPEP at runtime.
3346 (Some analogous efficiency issues for non-simple vectors specialized to
3347 element types other than T, or for non-simple multidimensional arrays,
3348 have not been addressed. They could almost certainly be handled the
3349 same way if anyone is motivated to do so.)
3350 * The changes in array handling break binary compatibility, so
3351 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3352 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3353 instead of (VALUES T T).
3354 * By following the instructions that Dan Barlow posted to sbcl-devel
3355 on 2 July 2000, I was able to enable primitive dynamic object
3356 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3357 functionality is not implemented (since it calls ld to resolve
3358 library references automatically, requiring RUN-PROGRAM for its
3359 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3360 try to resolve library references) is now supported.
3361 * The system now flushes the standard output streams when it terminates,
3362 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3363 them at several other probably-convenient times, e.g. in each pass of
3364 the toplevel read-eval-print loop, and after evaluating a form given
3365 as an "--eval" command-line option. (These changes were motivated by a
3366 discussion of stream flushing issues on cmucl-imp in August 2000.)
3367 * The source transform for TYPEP of array types no longer assumes
3368 that an array whose element type is a not-yet-defined type
3369 is implemented as an array of T, but instead punts, so that the
3370 type will be interpreted at runtime.
3371 * There is now some support for cross-compiling in make.sh: each of
3372 the phases of make.sh has its own script. (This should be transparent
3373 to people doing ordinary, non-cross-compile builds.)
3374 * Since my laptop doesn't have hundreds of megabytes of memory like
3375 my desktop machine, I became more motivated to do some items on
3376 my to-do list in order to reduce the size of the system a little:
3377 ** Arrange for various needed-only-at-cold-init things to be
3378 uninterned after cold init. To support this, those things have
3379 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3380 symbols with such names are now uninterned after cold init).
3381 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3382 fasl files for cold load.
3383 ** Remove the old compiler structure pooling code (which used to
3384 be conditional on the target feature :SB-ALLOC) completely.
3385 ** Redo the representation of some data in cold init to be more compact.
3386 (I also looked into supporting byte compiled code at bootstrap time,
3387 which would probably reduce the size of the system a lot, but that
3388 looked too complicated, so I punted for now.)
3389 * The maximum signal nesting depth in the src/runtime/ support code has
3390 been reduced from 4096 to 256. (I don't know any reason for the very
3391 large old value. If the new smaller value turns out to break something,
3392 I'll probably just bump it back up.)
3393 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3395 * Many, many bugs reported by Peter Van Eynde have been added to
3396 the BUGS list; some have even been fixed.
3397 * While enabling dynamic object file loading, I tried to make the
3398 code easier to understand, renaming various functions and variables
3399 with less ambiguous names, and changing some function calling
3400 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3401 * While trying to figure out how to do the OpenBSD port, I tried to
3402 clean up some of the code in src/runtime/. In particular, I dropped
3403 support for non-POSIX signal handling, added various comments,
3404 tweaked the code to reduce the number of compilation warnings, and
3405 renamed some files to increase consistency.
3406 * To support the new automatic configuration functionality in make.sh,
3407 the source file target-features.lisp-expr has been replaced with the
3408 source file base-target-features.lisp-expr and the machine-generated
3409 file local-target-features.lisp-expr.
3410 * fixed a stupid quoting error in make.sh so that using CMU CL
3411 "lisp -batch" as cross-compilation host works again
3413 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3415 * DESCRIBE no longer tries to call itself recursively to describe
3416 bound/fbound values, so that it no longer fails on symbols which are
3417 bound to themselves (like keywords, T, and NIL).
3418 * DESCRIBE now works on generic functions.
3419 * The printer now prints less-screwed-up representations of closures
3420 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3421 * A private symbol is used instead of the :EMPTY keyword previously
3422 used to mark empty slots in hash tables. Thus
3423 (DEFVAR *HT* (MAKE-HASH-TABLE))
3424 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3425 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3426 now does what ANSI says that it should. (You can still get
3427 similar noncompliant behavior if bang on the hash table
3428 implementation with all the symbols you get back from
3429 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3430 This breaks binary compatibility, since tests for equality to
3431 :EMPTY are wired into things like the macroexpansion of
3432 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3434 * There's now a minimal placeholder implementation for CL:STEP,
3435 as required by ANSI.
3436 * An obscure bug in the interaction of the normal compiler, the byte
3437 compiler, inlining, and structure predicates has been patched
3438 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3439 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3441 * Missing ordinary arguments in a macro call are now detected even
3442 when the macro lambda list contains &KEY or &REST.
3443 * The debugger no longer complains about encountering the top of the
3444 stack when you type "FRAME 0" to explicitly instruct it to go to
3445 the top of the stack. And it now prints the frame you request even
3446 if it's the current frame (instead of saying "You are here.").
3447 * As specified by ANSI, the system now always prints keywords
3448 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3450 * The default initial SIZE of HASH-TABLEs is now smaller.
3451 * Type information from CLOS class dispatch is now propagated
3452 into DEFMETHOD bodies, so that e.g.
3453 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3455 is now basically equivalent to
3456 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3457 (DECLARE (TYPE SINGLE-FLOAT X))
3459 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3460 operation, without having to do run-time type dispatch.
3461 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3462 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3463 * Since I don't seem to be making big file reorganizations very often
3464 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3465 is overflowing my ability to conveniently back them up), I've finally
3466 checked the system into CVS. (The CVS repository is on my home system,
3467 not at SourceForge -- putting it on SourceForge might come later.)
3468 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3469 high-level GC-NOTIFY-FOO functions send their output. (There's
3470 still very little control of where low-level verbose GC functions
3471 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3472 less than it used to -- the GC-NOTIFY-FOO functions are now under
3473 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3474 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3475 in only one place in the source code, and propagates it automatically
3476 everywhere that it's needed. Thus e.g. when I bump the version from
3477 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3478 * The C source files now include boilerplate legalese and documentation
3479 at the head of each file (just as the Lisp source files already did).
3480 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3481 to his page will be replaced with a link to his new CLiki service.
3483 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3485 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3486 * The build process now looks for GNU make under the default name "gmake",
3487 instead of "make" as it used to. If GNU make is not available as "gmake"
3488 on your system, you can change this default behavior by setting the
3489 GNUMAKE environment variable.
3490 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3491 can build without error under CMU CL.
3493 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3495 * There is now a partial SBCL user manual (with some new text and some
3496 text cribbed from the CMU CL manual).
3497 * The beginnings of a profiler have been added (starting with the
3498 CMU CL profiler and simplifying and cleaning up). Eventually the
3499 main interface should be through the TRACE macro, but for now,
3500 it's still accessed through vaguely CMU-CL-style functions and macros
3501 exported from the package SB-PROFILE.
3502 * Some problems left over from porting CMU CL to the new
3503 cross-compilation bootstrap process have been cleaned up:
3504 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3505 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3506 part when getting rid of the compiler *BACKEND* stuff.)
3507 ** The value of *NULL-TYPE* was screwed up, because it was
3508 being initialized before the type system knew the final
3509 definition of the 'NULL type. This screwed up several key
3510 optimizations in the compiler, causing inefficiency in all sorts
3511 of places. (I found it because I wanted to understand why
3512 GET-INTERNAL-RUN-TIME was consing.)
3513 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3514 PROCLAIM FTYPE information. Unfortunately this broke binary
3515 compatibility again, since now the forms output by DEFGENERIC
3516 to refer to functions which didn't exist in 0.6.3.
3517 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3518 can use the new (as of 0.6.3) transform for SOME into MAP into
3520 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3521 values to (INTEGER 0 1000000), so that the time code will no longer
3522 occasionally get blown up by Linux returning 1000000 microseconds
3523 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3524 its output conform to the ANSI spec. (Alas, this makes its output
3525 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3526 * A full call to MAP NIL with a single sequence argument no longer conses.
3527 * fixes to problems pointed out by Martin Atzmueller:
3528 * The manual page no longer talks about multiprocessing as though
3529 it were currently supported.
3530 * The ILISP support patches have been removed from the distribution,
3531 because as of version 5.10.1, ILISP now supports SBCL without us
3532 having to maintain patches.
3533 * added a modified version of Raymond Toy's recent CMU CL patch for
3534 EQUALP comparison of HASH-TABLE
3536 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3538 * The system still can't cross-compile itself with
3539 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3540 entails), but at least it can compile more of itself that way
3541 than it used to be able to, and various buglets which were uncovered
3542 by trying to cross-compile itself that way have now been fixed.
3543 * This release breaks binary compatibility again. This time
3544 at least I've incremented the FASL file format version to 2, so that the
3545 problem can be detected reliably instead of just causing weird errors.
3546 * various new style warnings:
3547 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3548 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3549 ** using the lexical binding of a variable named in the *FOO* style
3550 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3551 as specified by ANSI.
3552 * *RANDOM-STATE* is no longer automatically initialized from
3553 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3554 default behavior of the system is to repeat its behavior every time
3555 it's run. If you'd like to change this behavior, you can always
3556 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3557 old convention there was no comparably easy way to get the system to
3558 repeat its behavior every time it was run.
3559 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3560 FTYPE declarations has been removed, in favor of their ANSI
3561 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3562 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3563 the types of their sequence arguments can be determined at compile time.
3564 This is done through a new open code expansion for MAP which eliminates
3565 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3566 argument types can be determined at compile time.
3567 * The optimizer now transforms COERCE into an identity operation when it
3568 can prove that the coerced object is already of the correct type. (This
3569 can be a win for machine generated code, including the output of other
3570 optimization transforms, such as the MAP transform above.)
3571 * Credit information has been moved from source file headers into CREDITS.
3572 * Source file headers have been made more standard.
3573 * The CASE macro now compiles without complaining even when it has
3576 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3578 * (Note that the way that the PCL macroexpansions were rewritten
3579 to accommodate the change in DEFGENERIC below breaks binary
3580 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3581 not run under sbcl-0.6.2. Once we get out of alpha releases,
3582 i.e. hit release 1.0.0, we'll probably try to maintain binary
3583 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3584 and sbcl-1.4.4. Until then, however, it might be fairly common
3585 for maintenance releases to break binary compatibility.)
3586 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3588 * The reporting of unhandled signals has been changed to print some
3589 explanatory text as well as the report form. (Previously only
3590 the report form was printed.)
3591 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3592 it defines, so that the compiler no longer issues undefined function
3593 warnings for compiled-but-not-yet-loaded generic functions.
3594 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3595 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3596 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3597 section of the ANSI spec are supported.
3598 * The "" nickname for the "KEYWORD" package has been removed.
3599 The reader still handles symbol tokens which begin with a package marker
3600 as keywords, but it doesn't expose its mechanism for doing so in the
3601 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3602 * The system now issues STYLE-WARNINGs for contradictory TYPE
3603 proclamations. (Warnings for contradictory FTYPE proclamations would
3604 be nice too, but those can't be done usefully unless the type system
3605 is made smarter about FUNCTION types.)
3606 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3607 systematized, so that "*target-*.lisp is supposed to exist only on the
3608 target and imply that there's a related file which exists on the
3609 host, and *host-*.lisp is supposed to exist only on the host and imply
3610 that there's a related file which exists on the target. This involves a
3611 lot of renaming. Hopefully the acute confusion caused by the renaming
3612 will be justified by the reduction in chronic confusion..
3613 ** runtime-type.lisp -> early-target-type.lisp
3614 ** target-type.lisp -> late-target-type.lisp
3615 ** early-host-format.lisp -> early-format.lisp
3616 ** late-host-format.lisp -> late-format.lisp
3617 ** host-error.lisp -> misc-error.lisp
3618 ** early-error.lisp -> early-target-error.lisp
3619 ** late-error.lisp -> late-target-error.lisp
3620 ** host-defboot.lisp -> early-defboot.lisp
3621 ** code/misc.lisp -> code/target-misc.lisp
3622 ** code/host-misc.lisp -> code/misc.lisp
3623 ** code/numbers.lisp -> code/target-numbers.lisp
3624 ** code/early-numbers.lisp -> numbers.lisp
3625 ** early-host-type.lisp -> early-type.lisp
3626 ** late-host-type.lisp -> late-type.lisp
3627 ** host-typep.lisp -> typep.lisp
3628 ** load.lisp -> target-load.lisp
3629 ** host-load.lisp -> load.lisp
3630 ** host-disassem.lisp -> disassem.lisp
3631 ** host-insts.lisp -> insts.lisp
3632 ** byte-comp.lisp -> target-byte-comp.lisp
3633 ** host-byte-comp.lisp -> byte-comp.lisp
3634 ** host-signal.lisp -> signal.lisp
3635 ** host-defstruct.lisp -> defstruct.lisp
3636 ** late-target-type.lisp -> deftypes-for-target.lisp
3637 Furthermore, several other previously target-only files foo.lisp (e.g.
3638 hash-table.lisp and random.lisp) have been split into a target-and-host
3639 foo.lisp file and a target-only target-foo.lisp file, with their key type
3640 definitions in the target-and-host part, so that the cross-compiler will
3641 know more about target types.
3642 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3643 gone away. In their place are various *BACKEND-FOO* variables
3644 corresponding to the slots of the old structure.
3645 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3646 into the target SBCL has been fixed.
3647 * The chill.lisp system for loading cold code into a running SBCL
3649 * Support for the CMU CL "scavenger hook" extension has been removed.
3650 (It was undocumented and unused in the CMU CL sources that SBCL was
3651 derived from, and stale in sbcl-0.6.1.)
3652 * Various errors in the cross-compiler type system were detected
3653 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3654 (enabling various consistency checks). Many of them were fixed,
3655 but some hard problems remain, so the compiler is back to
3656 running without *TYPE-SYSTEM-INITIALIZED* for now.
3657 * As part of the cross-compiler type system cleanup, I implemented
3658 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3659 * I have started adding UNCROSS calls throughout the type system
3660 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3661 on cross-compiler input files will be able to go away, and various
3663 * CONSTANTP now returns true for quoted forms (as explicitly required
3666 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3668 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3669 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3670 (MAKE-LIST -1). In the longer term, it ought to become true
3671 that declarations are assertions even at SAFETY 1. For now, it's not
3672 quite true even at SAFETY 3, but it's at least more nearly true..
3673 (Note that this change seems to increases the size of the system by
3674 O(5%) and to decrease the speed of the compiler by 20% or more.)
3675 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3676 for the problem of printing dozens of lines of distracting information
3677 about low-level system machinery as part of the top stack frame
3678 on entry to the debugger when an undefined function was called.
3679 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3680 is not reset to COMMON-LISP-USER.
3681 * Compilation of stuff related to dyncount.lisp has been made conditional
3682 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3683 smaller. The various dyncount-related symbols have been moved into
3684 a new "SB-DYNCOUNT" package.
3685 * tty-inspect.lisp has been renamed to inspect.lisp.
3686 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3687 feature has gone away. (When we eventually port to other flavors of
3688 libc and/or Unix, we'll try to make the differences between flavors
3689 invisible at the user level.)
3690 * Various other *FEATURES* tags, and/or their associated conditionals,
3691 have been removed if obsolescent, or given better documentation, or
3692 sometimes given more-mnemonic names.
3694 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3696 * tidied up "make.sh" script
3697 * tidied up system directory structure
3698 * better "clean.sh" behavior
3699 * added doc/FOR-CMUCL-DEVELOPERS
3700 * many many small tweaks to output format, e.g. removing possibly-confusing
3701 trailing #\. character in DESCRIBE-INSTANCE
3702 * (EQUALP #\A 'A) no longer signals an error.
3703 * new hashing code, including EQUALP hashing
3704 * tidied up Lisp initialization and toplevel
3705 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3706 * command line argument processing
3707 * added POSIX-GETENV function to deal with Unix-ish environment variables
3708 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3709 terminating SBCL on EOF
3710 * non-verbose GC by default
3711 * There is no more "sbcl" shell script; the sbcl file is now the C
3712 runtime executable (just like CMU CL).
3713 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3715 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3716 symbol and package information as Lisp native symbol and package objects
3717 instead of strings naming symbols and strings naming packages. This way,
3718 whenever packages are renamed (as in warm init), debug information is
3719 transformed along with everything else.
3720 * tweaked the optimization policy declarations which control the building
3721 of SBCL itself. Now, among other things, the system no longer saves
3722 source location debugging information. (This helps two problems at once
3723 by reducing SBCL size and by keeping SBCL from trying to look for its
3724 sources -- which may not exist -- when reporting errors.)
3725 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3726 debugging and testing purposes
3727 * cleaned up printing, making the printer call PRINT-OBJECT for
3728 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3729 methods, giving nearly-ANSI behavior
3730 * converted almost all special variables to use *FOO* naming convention
3731 * deleted PARSE-TIME functionality, since it can be done portably
3732 * moved some files out of cold init into warm init
3733 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3735 * regularized formatting of source files
3736 * added an install.sh script
3737 * fixed ridiculous memory usage of cross-compiler by making
3738 compiler/alloc.lisp not try to do pooling unless it can hook
3739 itself into the GC of the cross-compilation host. Now the system
3740 builds nicely on my old laptop.
3741 * added :SB-ALLOC in target-features.lisp-expr
3742 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3743 was not implemented)
3744 * re-did condition handling and note reporting in the compiler. Notes
3745 are no longer handled by signalling conditions. Style warnings
3746 and warnings are handled more correctly and reported in such a way
3747 that it's easy to find one or the other in your output (so that you
3748 can e.g. figure out which of many problems caused COMPILE-FILE to
3750 * changed the severity of several compiler warnings from full WARNING
3751 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3752 compiler note reporting so that it doesn't use the condition system
3753 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3754 COMPILE-FILE command)
3755 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3756 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3757 no longer supported, since their implementation was deeply intertwingled
3758 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3759 * removed bogus "support" for compiler macros named (SETF FOO), and
3760 removed the compiler macro for SETF INFO (but only after making a fool
3761 of myself on the cmucl-imp mailing list by posting a bogus patch for
3762 DEFINE-COMPILER-MACRO..)
3763 * Compiled files containing forms which have side effects on the Lisp
3764 reader (such as DEFPACKAGE forms) are now handled more correctly.
3765 (Compiler queuing of top level lambdas has been suppressed by setting
3766 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3767 * deleted various currently-unused source files, e.g. gengc.lisp. They
3768 may be added back at some point e.g. when porting to other architectures,
3769 but until they are it's distracting to distribute them and to try to
3771 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3772 there were so many of them they're just distractions, and UNCROSS is
3773 known to be able to handle the current sources
3774 * moved PROFILE functionality into TRACE, so that it will be clear
3775 how the wrapping and unwrapping of functions when you profile them
3776 interacts with the wrapping and unwrapping of functions when you
3777 trace them. (Actually, the functionality isn't there yet, but at least
3778 the interface specification is there. Hopefully, the functionality will
3779 arrive with some future maintenance release.)
3780 * removed host-oops.lisp
3781 * changed signature of QUIT function to allow UNIX-CODE argument
3782 * fixed READ-SEQUENCE bug
3783 * tweaked verbose GC output so that it looks more like the progress
3784 output that ANSI specifies for functions like LOAD
3785 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3786 * added <http://sbcl.sourceforge.com> to the banner information printed by