1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
3 * Enhancements for sbcl running on the Windows operating system:
4 ** (user-homedir-pathname) and default initialization file
5 locations now know about the user's "Documents and Settings"
6 directory (thanks to Yaroslav Kavenchuk)
7 ** run-program is implemented (thanks to Mike Thomas)
8 ** sockets support (thanks to Timothy Ritchey)
9 ** better backtrace support (thanks to Alastair Bridgewater)
10 ** sb-grovel supported
11 ** asdf-install and sb-posix work somewhat
12 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
13 * minor incompatible change: The reader no longer ignores errors
14 regarding non-existent packages in #+ and #- feature tests.
15 * new feature: command line options --no-sysinit, --no-userinit to
16 inhibit loading the corresponding init files
17 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
18 for sending data through UDP sockets (thanks to François-René Rideau)
19 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
20 error is signalled instead (thanks to François-René Rideau)
21 * minor incompatible change: Error signalling behaviour of lexical
22 operations violating package locks has changed slightly. Refer to
23 documentation on package locks for details.
24 * enhancement: EVAL can process IF-expressions without resorting to the
26 * bug fix: LISTEN sometimes returned T even in cases where no data was
27 immediately available from the stream
28 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
29 were reversed. (reported by Levente Mészáros)
30 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
31 when given a vector displaced to an adjustable array. (reported
33 * bug fix: derivation of float boundaries from numbers outside the
34 appropriate float range (reported by John Wiseman)
35 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
36 allow condition class objects as arguments to DOCUMENTATION and
37 (SETF DOCUMENTATION); only find and set documentation for
38 structure names for the STRUCTURE doc-type. (suggested by Gary
40 * improvements to the Solaris/x86 port:
41 ** works on Solaris 11/Solaris Express
42 ** floating-point exception handling support
43 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
44 * fixed some bugs revealed by Paul Dietz' test suite:
45 ** REMOVE-METHOD returns its generic function argument even when
46 the method is not one of the generic functions' methods.
47 ** objects declared as MEMBER types can be admissible arguments to
49 ** printing characters should simply be printed by the FORMAT ~:C
52 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
53 * new platform: experimental support for SBCL x86/Darwin, including
54 MacOS X 10.4.5 on Intel.
55 * new feature: Unicode character names are now known to the system
56 (through CHAR-NAME and NAME-CHAR).
57 * new feature: the filesystem location of SBCL's core file is
58 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
59 * minor incompatible change: the contrib modules SB-POSIX and
60 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
61 this change is to make it easier to distribute
62 location-independent binaries.
63 * bug fix: as implied by AMOP, standardized classes no longer have
64 slots named by external symbols of public packages. (reported by
66 * optimization: calling functions via constant symbols -- as in
67 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
68 via the function object as in (FUNCALL #'FOO).
69 * optimization: on x86 and x86-64, the calling convention for the
70 case of dealing with an unknown number of values has been altered
71 to be friendlier to the prediction heuristics implemented,
72 particularly with reference to CALL and RET pairing. (thanks to
74 * optimization: on x86, the code for access to thread-local symbol
75 values and binding/unbinding of thread-local symbols is smaller.
76 (thanks to Alastair Bridgewater)
77 * enhancement: CONSTANTP is now able to determine constantness of
78 more complex forms, including calls to constant-foldable standardized
79 functions and some special forms beyond QUOTE.
80 * fixed bug: occasional GC crashes on Solaris/x86
81 * optimization: x86-64 supports stack allocation of results of simple
82 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
83 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
84 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
85 (thanks to James Knight)
86 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
87 by the standard (thanks to Stephen Compall)
89 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
90 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
91 be used for bundling the runtime and the core file into one
92 executable binary. This feature is not currently supported on all SBCL
93 platforms. (thanks to James Bielman and NIIMI Satoshi)
94 * new feature: a generational or ephemeral garbage collector is now
95 the default on the PowerPC platform (both Linux and Darwin). The
96 old Cheney (stop and copy) collector is a build-time option.
97 (thanks to Cyrus Harmon, after Raymond Toy)
98 * minor incompatible change: the method by which SBCL finds its
99 contributed modules has changed; it no longer relies on symbolic
100 links from an $SBCL_HOME/systems directory, but searches directly
101 in the subdirectories of $SBCL_HOME.
102 * enhancement: the dynamic heap size on the Linux/PPC platform is
103 markedly larger, even using the older Cheney garbage collector.
104 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
105 certain complicated string types. (reported by Gary King)
106 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
107 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
108 * fixed bug: DOCUMENTATION on structures no longer signals an
109 error if no structure type of the right name exists. (reported by
111 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
112 applicable non-standard primary or :AROUND method on
113 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
114 no longer cause unbound variable errors. (reported by Kevin Reid)
115 * fixed bug: metacircle resolution in cases where methods have slots
116 added before the slots from STANDARD-METHOD. (reported by Jean
118 * fixed bug: the Power PC assembly code for calling into lisp
119 sometimes computed its return address wrongly (depending on the
120 15th bit of the address, fixed at link time). This used to
121 manifest itself in a segmentation violation while building PCL.
122 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
123 many others over the years)
124 * fixed bug: ROOM no longer reports silly numbers for stack usage if
125 the stack pointer (treated as a signed integer) is negative.
126 (thanks to Peter van Eynde)
128 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
129 * new platform: experimental support for the Windows operating
130 system has been added. (thanks to Alastair Bridgewater)
131 * fixed several bugs in and robustified the PPC FFI (including
132 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
133 * bug fix: allow non-simple string symbol names (reported by Paul
135 * bug fix: interrupt handling on NetBSD (thanks to Richard M
137 * bug fix: saving a core corrupted callbacks on x86/x86-64
138 * bug fix: closed a loophole in metacircularity detection and
139 grounding in the PCL implementation of CLOS.
140 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
141 (reported by tomppa on #lisp)
142 * bug fix: clear the direction flag on Lisp -> C transitions, as
143 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
144 (reported by Andrej Grozin and Hendrik Maryns)
145 * optimization: major improvements to GC efficiency on GENCGC platforms
146 * optimization: faster implementation of EQUAL
147 * optimization: emit more efficient opcodes for some common
148 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
149 * optimization: several other minor code-generation improvements on x86-64
150 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
152 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
153 * minor incompatible change: (SETF CLASS-NAME) and (SETF
154 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
155 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
156 by ANSI as a generic function, and for consistency (SETF
157 GENERIC-FUNCTION-NAME) is treated likewise.
158 * fixed bug #233.b: make constraint propagation notice when a variable
159 value is changed after it is referenced but before it is used
160 * fixed bug #296: no more arbitrary behaviour from filenames with
161 odd characters as --load arguments. (reported by Adam Warner)
162 * fixed bug #390: :CHARACTER-SET pathname components now work as
163 expected. (reported by Tim Daly Jr)
164 * fixed bug #391: complicated :TYPE intersections in slot
165 definitions no longer cause an error in PCL internals.
166 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
167 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
168 returning the number of octets which would be written to the
169 file-stream. (thanks to Robert J. Macomber)
170 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
171 arguments to the internal stream functions. (thanks to David
173 * bug fix: stack exhaustion is now handled on the x86/SunOS
174 platform. (thanks to Daisuke Homma)
175 * bug fix: more accurate ROOM results on GENCGC platforms
176 * optimization: improved type inference for arithmetic-for
177 index variables in LOOP
178 * optimization: faster floating-point SQRT on x86-64
179 * fixed some bugs revealed by Paul Dietz' test suite:
180 ** DOCUMENTATION returns NIL instead of "" for method combinations
181 that don't have a docstring
183 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
184 * minor incompatible change: (SETF CLASS-NAME) and (SETF
185 GENERIC-FUNCTION-NAME) are no longer generic functions, and
186 therefore are not customizeable by user code (as seems to be at
187 least permitted and maybe required by AMOP). As a consolation,
188 however, the SBCL implementation of these functions now calls
189 REINITIALIZE-INSTANCE as specified by AMOP.
190 * bug fix: slot-definition documentation is propagated as per ANSI
191 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
192 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
193 argument, as do eslotd initialization methods. (from Pascal
194 Costanza's "Closer" project)
195 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
196 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
198 * bug fix: it is now possible to have more than one subclass of
199 STANDARD-GENERIC-FUNCTION without causing stack overflow.
200 (reported by Bruno Haible, Pascal Costanza and others)
201 * bug fix: the dependent update protocol now works for generic
202 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
204 * bug fix: condition-class instances corresponding to
205 DEFINE-CONDITION forms are now created eagerly. (reported by
206 Kalle Olavi Niemitalo on comp.lang.lisp)
207 * bug fix: floating point printing is more accurate in some
208 circumstances. (thanks to Simon Alexander)
209 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
210 merged with *DEFAULT-PATHNAME-DEFAULTS*.
211 * bug fix: callbacks on OS X now preserve stack-alignment.
212 (thanks to Cyrus Harmon)
213 * enhancement: the x86-64 disassembler is much better at
214 disassembling SSE instructions. (thanks to Lutz Euler)
215 * enhancement: CHECK-TYPE on a variable now allows type-inference
216 (thanks to Vincent Arkesteijn)
217 * optimization: improved performance of EUC-JP external format.
218 (thanks to NIIMI Satoshi)
219 * optimization: performance improvements to IO on file streams of
220 :ELEMENT-TYPE CHARACTER
221 * optimization: much faster memory allocation on x86-64
222 * optimization: faster garbage collections (latency and throughput)
224 * optimization: faster bignum subtraction and division on x86 and x86-64
225 (thanks to Lutz Euler)
226 * optimization: more accurate type-inference for loops that use a
227 floating point index variable or a negative step.
229 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
230 * bug fix: add a workaround to SBCL looping infinitely at startup on
231 Linux kernels with apparently buggy implementations of personality().
232 (thanks to Svein Ove Aas)
233 * bug fix: Unicode symbols are correctly printed in LDB backtraces
234 (thanks to David Lichteblau)
235 * bug fix: local bindings shadow global symbol macros for
236 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
237 * optimization: non-open coded uses of numeric comparison operators
238 (e.g. >) no longer cons when called with more than one parameter
239 on platforms supporting dynamic-extent allocation.
240 * enhancement: saving cores with foreign code loaded is now supported
241 on MIPS/Linux in addition to the previously supported platforms.
242 * bug fix: FIND-RESTART now tests for activity, not applicability when given
243 a restart object as identifier. (reported by Helmut Eller for CMUCL)
244 * bug fix: division by zero in sb-sprof when no samples were collected
245 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
246 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
247 (thanks to Alexander Kjeldaas)
248 * new feature: ignoring the return values of destructive functions
249 like NREVERSE signals a compile-time style-warning.
250 (thanks to Kevin Reid)
251 * new feature: the alignment of alien structure fields can be
252 explicitly specified. (thanks to Cyrus Harmon)
253 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
254 and dump core on SIGQUIT
256 ** incompatible change: threads do not inherit values of specials
257 from their parents (see manual)
258 ** bug fix: threads stacks belonging to dead threads are freed by the
259 next exiting thread, no need to gc to collect thread stacks anymore
260 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
261 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
262 ** bug fix: dynamic variable and thread start related gc lossage
263 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
265 * fixed some bugs revealed by Paul Dietz' test suite:
266 ** SUBTYPEP is slightly more accurate on heinously complicated
267 CONS types where some of the members have uncertain (in the
268 NIL, NIL sense) type relationships to each other.
270 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
271 no longer purified unless :PURIFY T is explicitly specified.
272 ** Non-purified cores are significantly smaller than before
274 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
275 * new feature: timers based on Zach Beane's excellent timer package
276 * added support for the following external formats: koi8-u,
277 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
278 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
279 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
280 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
281 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
282 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
283 * incompatible change: a threaded SBCL will no longer revert to
284 non-threaded mode on non-NPTL systems, but refuse to start entirely.
285 * bug fix: interrupts are disabled until startup is complete; no
286 more sigsegvs when receiving a signal to soon
287 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
289 * bug fix: add a workaround for the memory randomization features in
290 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
291 workaround will only be in effect on systems with the proc filesystem
293 * bug fix: printing objects of type HASH-TABLE signals a
294 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
296 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
297 have been running for over 50 days. (reported by Gilbert Baumann)
298 * bug fix: the logic for getting names of functions gets less
299 confused when confronded with alternate-metaclass
300 funcallable-instances. (reported by Cyrus Harmon)
301 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
302 consistent, even on internal alternate-metaclass objects.
303 * bug fix: flush closure information collected by physical
304 environment analysis prepass before the main pass. (bug reported
306 * bug fix: compiler pack phase does not modify a hashtable, which is
307 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
308 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
309 may be expanded. (reported by Alexander <malishev> on c.l.l)
310 * bug fix: no more style-warnings when compiling code using LAST
311 under high-SPEED low-SPACE settings. (reported by David Wallin
313 * fixed some incompatibilities between SBCL's MOP and the MOP
315 ** the METAOBJECT class is now implemented;
316 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
317 STANDARD-OBJECT, as required;
318 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
319 now compatible (as required by VALIDATE-SUPERCLASS); there
320 remains a consistency requirement over the presence or absence
321 of the FUNCTION class in the superclass list of finalized
322 classes; see the manual for more details;
323 ** the :AROUND method for COMPUTE-SLOTS on
324 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
325 requested slot ordering.
327 ** bug fix: parent thread now can be gc'ed even with a live
329 ** bug fix: binding a special with PROGV to no value is not
330 the same as that symbol not having been bound (thanks to
332 ** bug fix: binding specials is thread safe (thanks to
334 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
335 and SB-SPROF do not die with 'no handler for signal XX in
336 interrupt_handle_now(..)' anymore
337 ** bug fix: WITH-TIMEOUT works with multiple running threads
338 * fixed some bugs revealed by Paul Dietz' test suite:
339 ** ENSURE-GENERIC-FUNCTION should take a method class object for
340 the :method-class keyword argument.
342 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
343 * new port: the Solaris operating system on x86 processors is now
344 mostly supported, though some rough edges in the environment
345 remain. (thanks to Daisuke Homma)
346 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
347 space for its heap. (thanks to Thiemo Seufer)
348 * optimization: SBCL on MIPS platforms now supports dynamic-extent
349 closures. (thanks to Thiemo Seufer)
350 * minor incompatible change: eof selects abort in the debugger.
351 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
352 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
354 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
355 deprecated, and will go away in a future revision of SBCL.
356 * minor incompatible change: GC-ON and GC-OFF are no longer
357 implemented with a counter, it does not matter how many times gc
358 is switched on or off
359 * bug fix: discriminating functions for generic function classes
360 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
361 make invalid assumptions about method precedence order. (reported
363 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
364 platforms. (thanks to Thiemo Seufer)
365 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
366 * bug fix: correct pathname printing with printer escaping is on.
367 (thanks to Kevin Reid)
368 * bug fix: complex VOP definitions in "user-space" no longer trigger
369 package locks. (reported by Zach Beane)
370 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
371 for funcallable-instance functions, and is no different from
373 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
374 characters in the current readtable. (reported by Nicholas Neuss)
375 * bug fix: SUBTYPEP on various CONS types returns more a more
376 accurate acknowledgment of its certainty.
377 * optimizations: REMOVE-DUPLICATES now runs in linear time on
378 lists in some cases. This partially fixes bug 384.
379 * flush all standard streams before prompting in the REPL and the
381 * bug fix: signal handling and triggering gc do not conflict
382 directly anymore, in particular a high frequency sb-sprof does
383 not prevent gc from running
384 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
385 approximation for timezone and DST information between the
386 universal time epoch and the smallest negative 32-bit time_t.
387 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
388 year 1899 so long as the encoded time is non-negative.
389 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
390 the correct answer for non-constant rotate amounts when performing
391 an inline 32-bit rotation.
393 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
394 there is only one thread in the session
395 ** bug fix: memory leak for streams created in one thread and
396 written to in another
397 ** bug fix: lockup when compiled with gcc4
398 ** bug fix: race that allows the gc to be triggered when gc is
400 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
401 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
403 ** bug fix: finished threads are gc'ed properly
404 * fixed some bugs revealed by Paul Dietz' test suite:
405 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
406 has arguments with a different set of applicable methods from
407 the orignal arguments.
408 ** The type error thrown by MAP now has a correct expected-type
410 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
411 name a compiled function.
412 ** Three MISC tests where a large bignum was improperly coerced to
413 a float that couldn't represent that bignum during type
414 derivation were fixed.
415 ** SUBTYPEP can now handle types involving the negation of a
416 list-form FUNCTION type.
417 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
418 as (COMPLEX (AND RATIO (NOT FIXNUM))).
419 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
421 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
422 * New feature: Experimental support for bivalent streams: streams
423 opened with :element-type :default now allow character and binary
424 (unsigned-byte 8) I/O
425 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
426 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
428 * bug fix: no more highly sporadic "couldn't check whether ~S is
429 readable" when reading a stream and an interrupt hits in the middle
430 of a select system call
431 * compiler better recognizes complex arrays (reported by Eduardo
433 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
434 for variables on SBCL built with linkage-tables. (reported by Luis
436 * various error reporting improvements.
437 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
438 (Thanks to Thiemo Seufer)
439 * bug fix: nested reader invokations work correctly
440 * bug fix: it is possible to have simultaneous references to foreign
441 code and foreign data with the same name.
443 ** added x86-64 support
444 ** incompatible change: the threading api now works with thread
445 objects instead of thread ids
446 ** bug fix: threads are protected from signals and interruption when
447 starting up or going down
448 ** bug fix: a race where an exiting thread could lose its stack to gc
449 ** bug fix: don't halt on infinite error in threads if possible
450 ** fixed numerous gc deadlocks introduced in the pthread merge
451 ** bug fix: fixed thread safety issues in read and print
452 ** bug fix: debugger doesn't hang on session lock if interrupted at
453 an inappropriate moment
454 ** bug fix: run-program is now thread safe(r)
455 ** bug fix: inner with-recursive-lock no longer releases the mutex
456 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
457 * fixed some bugs revealed by Paul Dietz' test suite:
458 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
459 EXPECTED-TYPE slots filled.
460 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
462 ** MAKE-CONDITION accepts classes as type-designators.
463 ** COMPILE may never return NIL.
464 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
465 range before calling Unix time functions
467 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
468 * numerous signal handling fixes to increase stability
469 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
470 * minor incompatible change: we now correctly canonize default
471 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
472 as per the MOP, rather than the historical (INITARG INITFUNCTION
473 INITFORM). (reported by Bruno Haible)
474 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
475 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
476 for more information.
477 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
478 pathname is a directory pathname.
479 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
480 * dynamic space size on PPC has been increased to 768Mb. (thanks to
482 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
483 :METACLASS argument in addition to a class name. (reported by
484 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
485 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
486 (reported by Bruno Haible; patch from Gerd Moellmann)
487 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
489 * bug fix: more cleanups to the floating point exception handling on
490 x86-64 (thanks to James Knight)
491 * bug fix: the compiler does not try to constant fold calls of
492 COERCE and BIT-* functions when they return freshly constructed
493 objects. (reported by Nikodemus Siivola and Paul Dietz)
494 * optimization: improved the MIPS versions of generic (in the
495 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
496 * optimization: direct conversion of (unsigned-byte 32) to floats on
497 the PowerPC platform.
498 * optimization: structure instances with raw slots now use less
499 memory, and probably show better memory locality. (thanks to
501 * optimization: DYNAMIC-EXTENT declarations for lists and closures
502 are treated as requests for stack allocation on the x86-64,
503 Alpha-32, PPC and SPARC platforms.
504 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
505 the wrong file descriptor; implementation of SOCKET-OPEN-P.
506 (thanks to Tony Martinez)
508 ** gcing a dead thread can no longer lead to lockups
509 ** threads block signals until they are set up properly
510 ** errno is no longer shared by threads
511 ** interrupt-thread restores the eflags register on x86
512 ** fixed some lockups due to gc/thread interaction
513 * fixed some bugs revealed by Paul Dietz' test suite:
514 ** invalid dotted lists no longer raise a read error when
516 ** signal an error if a symbol that names a declaration is used
517 as the name of a type, or vice versa
518 ** allow using the (declare (typespec var*)) abbreviation for
519 (declare (type typespec var*)) with all type specifiers
520 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
521 ** always return NIL from PPRINT-POP when OBJECT is NIL
522 ** don't signal errors when pretty-printing malformed LABELS,
523 FLET or MACROLET forms
524 ** declarations in a DOLIST body are also in scope for the
526 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
528 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
531 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
532 * fixed cross-compiler leakages that prevented building a 32-bit
533 target with a 64-bit host compiler.
534 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
535 opened with :IF-EXISTS :OVERWRITE.
536 * fixed bug 281: error for an invalid qualifier in a short-form method
537 combination method is not signalled until the faulty method is called.
538 * bug fix: iteration variable type inferrer failed to deal with open
539 intervals. (reported by Alan Shields)
540 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
541 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
542 or more runtime options were provided to the sbcl binary.
543 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
544 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
545 * optimization: there's now a fast-path for fixnum arguments in the
546 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
547 * optimization: the code generated on x86-64 is more compact thanks
548 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
549 * bug fix: floating-point exception handling now partly works on
550 x86-64. (Thanks to James Knight)
551 * improvement to the MIPS backend from Thiemo Seufer: C-style
552 64-bit long long arguments and return values to alien functions
554 * fixed some bugs revealed by Paul Dietz' test suite:
555 ** the type-error signalled from WARN has a filled-in DATUM slot.
556 ** the type-error required when a stream is not associated with
557 a file has the stream as its datum.
558 ** type-errors on single-floats on x86-64 no longer have
559 :INVALID-OBJECT as the datum
560 ** the type-errors signalled for invalid function names now have
561 a correct expected type
562 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
563 for structures defined with a :TYPE. Documentation strings for
564 typed structures are no longer immediately discarded
565 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
566 broadcast streams and synonym streams. FILE-LENGTH now also works
567 on broadcast streams.
569 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
570 * incompatible change: the --noprogrammer option, deprecated since
571 version 0.7.5, has been removed. Please use the equivalent
572 --disable-debugger option instead.
573 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
575 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
576 inoperational for a while now) has been completely removed.
577 * null lexical environments are now printed as #<NULL-LEXENV>,
578 significantly reducing the amount of clutter in typical backtraces.
579 * documentation on weak pointers, finalization, and after GC hooks
580 has been added to the manual.
581 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
582 as other specialized array types, is much faster. SUBSEQ and
583 COPY-SEQ on such arrays have also been sped up.
584 * optimization: EQL is now more efficient when at least other argument
585 is known to be of type (OR FIXNUM (NOT NUMBER)).
586 * fixed bug: the runtime is now less vulnerable to changes in the
587 size of the SBCL object on OS X, and virtual memory is reserved for
588 all spaces that need to be at a fixed address.
589 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
590 * fixed bug: finalizers and after GC hooks that cause consing are now
592 * fixed bug: compiler error messages and summaries are now printed to
593 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
594 * fixed inference of the upper bound of an iteration variable.
595 (reported by Rajat Datta).
596 * fixed bug 211e: calling local functions with duplicated constant
597 keyword argument no longer causes a bogus style warning about an
599 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
600 ftype declaration to be disregarded. (reported by Dave Roberts)
601 * fixed bug 373: caused by erronous compilation of references to alien
602 variables in the runtime on ppc/darwin.
603 * fixed bug 376: CONJUGATE type deriver.
604 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
605 (reported by Baughn on #lisp)
606 * a cross-compiler bug on non-x86 platforms has been identified and
607 fixed. (thanks to Bruno Haible)
608 * improvements to the MIPS runtime code for increased stability.
609 (thanks to Thiemo Seufer)
610 * increased the maximimum compact environment size to allow
611 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
612 * partial workaround for bug 135: don't name the function we're
613 calling for hairy cases of EVAL, so as not to accumulate one environment
614 entry per call to EVAL. (reported by Kevin Reid)
615 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
616 * optimization: type testing for non-vector arrays should be faster.
617 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
619 * bug fix: setting 31st element of a bit vector to zero did not work
621 * bug fix: redefining a class definition which failed due to a
622 previous accessor / function clash now works (but see BUGS entry
623 #380 for more problems in this area). (thanks to Zach Beane)
624 * the long form of DEFINE-METHOD-COMBINATION disables method group
625 checking when given a single method group with pattern *. (thanks
627 * on x86 compiler supports stack allocation of results of simple
628 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
629 * support for building and running on Mac OS X 10.4 "Tiger" has been added
630 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
631 * fixed some bugs related to Unicode integration:
632 ** the restarts for recovering from input and output encoding
633 errors only appear when there is in fact such an error to
635 * fixed some bugs revealed by Paul Dietz' test suite:
636 ** MISC.549 and similar: late transformation of unsafe type
637 assertions into derived types caused unexpected code
639 ** SCALE-FLOAT type deriver is less wrong.
640 ** type derivers for EXP, LOG and similar functions compute result
641 types for complex arguments better.
642 ** (MISC.563) CONJUGATE type deriver works for very restricted
644 ** out-of-line type testers for character strings are available.
645 ** EQUAL compiler transform understands specialness of objects
647 ** accessing double-floats stored on the stack now works on x86-64.
648 ** debugger internals could sometimes create invalid lispobjs,
649 resulting in GC crashes.
650 ** MISC.548: type check weakening can convert required type into
652 ** initialization forms for bindings are not in scope of free special
655 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
656 * incompatible change: thread support for non-NPTL systems has
657 been removed - locking is buggy and unreliable. A threaded
658 SBCL build will now warn at startup and refuse to create
659 new threads, unless futex support is detected in the kernel
660 * incompatible change: the top level REPL now has only an ABORT
661 restart associated with it, not TOPLEVEL and ABORT as it used to.
662 TOP and TOPLEVEL are now available as debugger commands for
663 returning to the top level.
664 * incompatible change: forms evaluated in the REPL now use the
665 global optimization policy.
666 * incompatible change: user- and system-initialization files are
667 no longer processed with LOAD, but by READ and EVAL; hence the
668 global optimization policy, startup package, readtable, etc,
670 * The .fasl file format number has been incremented because of
671 various incompatible changes.
672 * internal entry point details and argument counts no longer appear
673 in backtraces unless explicitly requested by setting
674 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
675 * built-in and standard functions no longer have names like "top
676 level local call to FOO".
677 * fixed bug 32: functions defined in non-null lexical environments
678 now have more legible printed representation
679 * fixed bug 33: functions defined in non-null lexical environemnts
680 are now more amenable to inspection by INSPECT.
681 * workaround for bug 354: XEPs no longer appear in backtraces unless
682 explicitly requested.
683 * fixed bug: receiving the signal which results from attempting to
684 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
685 SIGBUS on other *BSDs) on architectures where the C stack is also
686 the Lisp stack (x86 and x86-64) from foreign code no longer leads
687 to debugger confusion or wild instability. (reported by Cheuksan
689 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
690 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
691 (reported by Lutz Euler)
692 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
693 compiled code. (reported by Paul Dietz)
694 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
695 specialized parameter to a method to something that is not TYPEP
696 the specializer is now possible.
697 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
698 face of package deletion.
699 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
700 pathname. (reported by David Tolpin on c.l.l)
701 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
702 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
703 STRING-STREAM and FILE-STREAM, respectively.
704 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
705 than previously; in addition, its test suite is now run on build.
706 (thanks to Luke Gorrie)
707 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
708 * optimization: added a immediate representation for single-floats
710 * fixed some bugs revealed by Paul Dietz' test suite:
711 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
712 ** attempting to create a package with a colliding nickname causes
713 correctable errors to be signalled.
714 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
715 ** several x86-64 backend bugs related to sign-extension of immediate
718 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
719 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
720 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
722 * fixed loading of multiply forward-referenced layouts.
723 (thanks to Cheuksan Wang)
724 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
725 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
726 (see COMPILE-FILE documentation for details of :PRINT option)
727 * fixed bugs 19 and 317: fixed-format floating point printing is
728 more accurate. This also fixes a bug reported by Adam Warner
729 related to the ~@F format directive.
730 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
732 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
733 dispatch table if the from-char is a dispatch macro character.
734 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
735 by the wordsize no longer ignore the last word. (reported by Lutz
737 * fixed bug in type checking of dynamic-extent variables. (reported
739 * optimization: sequence traversal functions use their freedom to
740 coerce function designators to functions.
741 * optimization: code with many calls to CLOS methods specialized on
742 CLOS classes has had redundant type checks removed.
743 * optimization: added declarations to speed up operations that access
744 the internal character database (for example STRING-UPCASE)
745 * optimization: comparison operations between floats and sufficiently small
746 fixnums no longer create extra rationals
747 * fixed some bugs related to Unicode integration:
748 ** portions of multibyte characters at the end of buffers for
749 character-based file input are correctly transferred to the
750 start of the buffer at the next read.
751 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
752 passing it through to OPEN.
753 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
754 argument given to internal calls to OPEN.
755 ** the built-in comment readers (introduced by character sequences
756 ";" and "#|") are more forgiving to encoding errors; they will
757 STYLE-WARN and then attempt to resync the stream at a character
758 boundary. (thanks to Teemu Kalvas)
759 * fixed some bugs in the x86-64 port:
760 ** Negative short int return values from c-calls are sign-extended
762 ** The stack is aligned to 16-bytes for c-calls, as required by
763 the ABI. (thanks to Cheuksan Wang)
764 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
765 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
766 for enums. (thanks to Vincent Arkesteijn)
767 ** Multiple small optimizations and bugfixes for floating point
769 * fixed some bugs revealed by Paul Dietz' test suite:
770 ** Space, Tab, Linefeed, Return and Page have the invalid
771 secondary constituent character trait.
772 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
774 ** WITH-INPUT-FROM-STRING should only update the index place on
776 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
778 ** Bit-array operations (BIT-AND and similar) worked incorrectly
779 with one-dimensional arrays with fill pointers.
780 ** TYPE-OF failed on a complex with an integer realpart and a
782 ** compiler failure during type inference for the code of form
783 (IF (EQL X (THE ...)) ...) (MISC.535).
785 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
786 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
787 platform. The port supports 61-bit fixnums, large memory spaces
788 and reloading shared object files.
789 * enhancement: saving cores with foreign code loaded is now
790 supported on ppc/Darwin in addition to the previously supported
792 * enhancement: the statistical profiler now walks deeper into the
793 call stack for more meaningful call-graphs and accrued time
794 reports (x86/x86-64 only). It also now reports time spent in
796 * enhancement: it is now possible to trace most individual methods
797 of a generic function in addition to tracing the generic function
799 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
801 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
802 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
803 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
804 produces an error. (thanks to Vincent Arkesteijn)
805 * bug fix: NAMESTRING on pathnames with :WILD components in their
806 directories works correctly. (thanks to Artem V. Andreev)
807 * fixed bug 125: compiler preserves identity of closures. (reported
809 * bug fixed: functions with &REST arguments sometimes failed with
810 "Undefined function" when compiled with (DEBUG 3). (reported by
812 * bug fix: overflow during compiling of setting element of a bit
813 vector with constant index and value. (reported by Timmy Douglas)
814 * build fix: fixed the dependence on *LOAD-PATHNAME* and
815 *COMPILE-FILE-PATHNAME* being absolute pathnames.
816 * on x86 compiler partially supports stack allocation of dynamic-extent
818 * GO and RETURN-FROM do not check the extent of their exit points
819 when compiled with SAFETY 0.
820 * fixed some bugs related to Unicode integration:
821 ** encoding and decoding errors are now much more robustly
822 handled; it should now be possible to recover even from invalid
823 input or output to the terminal. (thanks to Teemu Kalvas)
824 ** provided a first cut at implementing STRING-TO-OCTETS and
825 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
826 ** altered the SB-MD5 contributed module slightly, changing the
827 interface just enough for it to be supportable for builds where
828 lisp characters are not eight bits.
829 * fixed some bugs revealed by Paul Dietz' test suite:
830 ** the FORMATTER-generated functions for ~V[ conditionals require
831 the correct number of arguments.
832 ** READ-FROM-STRING returns the mandated second value when applied
833 to displaced strings.
834 ** the #\Rubout and #\Backspace characters are treated as invalid
835 constituent characters by the tokenizer.
837 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
838 * new feature: reloading changed shared object files with
839 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
840 * new feature: references to foreign variables and functions
841 can now be compiled and loaded before the corresponding shared
842 object file is loaded, as long as the foreign definitions are
843 available at runtime.
844 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
845 supported. (thanks to Dan Debertin)
846 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
847 just keywords. (thanks to Vincent Arkesteijn)
848 * fixed bug #331: structure-class instances corresponding to
849 DEFSTRUCT forms are now created eagerly.
850 * fixed bug #345: backtraces from calls to undefined functions work
851 on x86 as well. Related bug #61 is now also partially fixed on x86
852 (backtraces from throws to unknown catch tags.)
853 * bug fix: lambda-list parsing is now stricter vrt. order and number
854 of lambda-list keywords.
855 * bug fix: as specified by AMOP, an error is signalled if a
856 class-option appears multiple times in a DEFCLASS form. (reported
858 * bug fix: RANDOM can be compiled when the compiler derives the type
859 of its numeric argument as a disjoint set of small integers.
860 (reported by Paul Dietz)
861 * bug fix: starting a core saved with shared objects loaded when
862 those objects are not available no longer causes threaded SBCL to
863 hang. (reported by Sean Ross)
864 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
866 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
867 argument, which must always be NIL. (reported by Kalle Niemitalo)
868 * bug fix: printing 1.0d+23 no longer results in an error.
869 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
870 * bug fix: structure slot setters preserve evaluation order. (thanks
872 * fixed some bugs related to Unicode integration:
873 ** RUN-PROGRAM can allow its child to take input from a Lisp
874 stream. (reported by Stefan Scholl)
875 ** ASDF-INSTALL successfully downloads .tar.gz files in all
876 locales. (reported by Ken Causey)
877 * fixed some bugs revealed by Paul Dietz' test suite:
878 ** INCF, DECF and REMF evaluate their place form as specified in
880 ** FORMATTER expands ~{ ~} iteration directives with V or #
881 parameters correctly.
882 ** FORMATTER deals with the ~@[ ~] conditional directive where the
883 consequent uses no arguments correctly.
884 ** the system has a partial understanding of the (COMPLEX RATIO)
886 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
889 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
890 * new feature: a build-time option (controlled by the :SB-UNICODE
891 keyword feature, enabled by default) for building the system with
892 support for the entire 21-bit character space defined by the
894 * new feature: the system now has rudimentary external-format
895 support; the primary user-visible change at this time is that
896 characters with the high bit set (such as lower-case-e-acute) will
897 print correctly to a terminal in a UTF-8 environment.
898 * minor incompatible change: BASE-CHAR no longer names a class;
899 however, CHARACTER continues to do so, as required by ANSI.
900 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
901 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
902 has been moved to the SB-EXT package (temporarily re-exported from
904 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
905 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
907 * fixed bug #351: better error-handlind and reporting for malformed
909 * fixed bug #350: bignum-printing is now more memory-efficient,
910 allowing printing of very large bignums, eg. (expt 2 10000000).
911 (reported by Bruno Haible)
912 * fixed bug #302: better primitive-type selection for intersection
914 * fixed bug #308: non-graphic characters now all have names, as
915 required. (reported by Bruno Haible)
916 * bug fix: redefining a class with different superclasses now correctly
917 removes it from the direct-subclasses of its previous superclasses.
918 (reported by David Morse)
919 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
920 new value now works. (reported by Bruno Haible)
921 * bug fix: correct canonicalization of multiple non-standard slot
922 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
923 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
924 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
925 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
926 now exists, an signals an error.
927 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
928 during macroexpansion for non-iteration variables as well. (reported
929 by Bruno Haible for CMUCL)
930 * bug fix: Cyclic structures and unprintable objects in compiler
931 messages no longer cause errors. (reported by Bruno Haible)
932 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
933 and READ-FROM-STRING all now return a primary value of NIL if
934 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
935 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
936 (reported by Bruno Haible for CMUCL)
937 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
938 do not designate a condition. (reported by Bruno Haible for
940 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
941 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
942 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
943 permitting resolution in favour of any of the conflicting symbols.
944 (reported by Bruno Haible for CMUCL)
945 * bug fix: EQUAL compiler optimizations is less aggressive on
946 strings which can potentially compare true despite having distinct
947 specialized array element types.
948 * bug fix: unit enumerations can be defined without dividing by
949 zero. (thanks to Vincent Arkesteijn)
950 * FORMAT compile-time argument count checking has been enhanced.
951 (report from Bruno Haible for CMUCL)
952 * a partial workaround for the bug 262: the compiler does not try to
953 inline-expand a local function doing RETURN-FROM from a deleted
954 BLOCK. (thanks to Peter Denno for the bug report and to David
955 Wragg for the simple test case)
956 * fixed some bugs revealed by Paul Dietz' test suite:
957 ** RENAME-PACKAGE allows all package designators as new package
959 ** constraint propagation and lambda variable substitution are
960 more cautious in dealing with partially deleted code.
961 ** compiler sometimes lost reoptimization passes.
962 ** CERROR, given a condition as condition designator, passes its
963 remaining arguments to the continue format control without
965 ** Case-altering FORMAT directives work correctly on non-ASCII
967 ** The REAL type specifier handles bounds outside the floating
968 point ranges without signalling FLOATING-POINT-OVERFLOW.
969 ** Functions with IR1-transformations can create intercomponent
970 references to global functions.
971 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
973 ** FORMAT ~R treats a nil value for its first parameter correctly.
975 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
976 * enhancement: saving cores with foreign code loaded is now
977 supported on x86/NetBSD and sparc/Linux in addition to the previously
979 * bug fix: on some platforms repeated installations caused multiple
980 copies of HTML documentation to be installed -- should not happen
981 any more. (reported by Stefan Scholl)
982 * bug fix: parsing self-recursive alien record types multiple times
983 no longer causes infinite recursion. (reported by Thomas F. Burdick,
984 original patch by Helmut Eller for CMUCL)
985 * bug fix: stack-exhaustion detection works now on NetBSD as well.
986 (thanks to Richard Kreuter)
987 * bug fix: defining classes whose accessors are methods on existing
988 generic functions in other (locked) packages no longer signals
989 bogus package lock violations. (reported by François-René Rideau)
990 * bug fix: special variables as DEFMETHOD parameters no longer have
991 associated bogus type declarations. (reported by David Wragg and
993 * bug fix: read-write consistency on streams of element-type
994 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
995 * bug fix: redefiniton of the only method of a generic function with
996 no DEFGENERIC no longer emits a full WARNING. In addition,
997 redefinition of generic functions with no DEFGENERIC to an
998 incompatible lambda list now signals an error. (thanks to Zach
1000 * bug fix: DEFGENERIC now works even when there's a function of the
1001 same name in an enclosing lexical environment. (thanks to Zach
1003 * fixed compiler failure, caused by instrumenting code during
1004 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1005 * optimization: added loop analysis and improved register allocation
1006 * fixed some bugs revealed by Paul Dietz' test suite:
1007 ** POSITION on displaced vectors with non-zero displacement
1008 returns the right answer.
1009 ** (SIMPLE-STRING) is a valid type specifier for sequence
1011 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1013 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1014 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1016 ** The justification version of the FORMAT ~< directive treats
1017 non-zero minpad parameter correctly.
1019 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1020 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1021 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1022 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1023 the supported interface.
1024 * new feature: Single-stepping of code compiled with DEBUG 2 or
1025 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1026 * new feature: saving cores with foreign code loaded is now
1027 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1028 Timothy Moore's work for CMUCL)
1029 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1030 parameters to * instead of NIL if no initform is supplied.
1031 (reported by Johan Bockgaard)
1032 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1033 correctly when FUNCALL appears as the car of the form. Note:
1034 despite this FUNCALL forms are not currently subject to
1035 compiler-macro expansion. (port of Raymond Toy's fix for the
1036 same from CMUCL, reported by Johan Bockgaard)
1037 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1038 (thanks for Teemu Kalvas)
1039 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1040 case instead of terminating the iteration (thanks for Julian
1041 Squires, Sean Champ and Raymond Toy)
1042 * bug fix: incorrect expansion of defgeneric that caused a style
1043 warning. (thanks for Zach Beane)
1044 * bug fix: slot accessor effective method computation works properly
1045 for classes with multiple non-standard applicable methods on
1046 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1047 * on x86 compiler supports stack allocation of results of LIST and
1048 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1049 CMUCL implementation by Gerd Moellmann)
1050 * fixed some bugs revealed by Paul Dietz' test suite:
1051 ** FORMAT strings with both the ~<~:;~> form of the justification
1052 directive and pretty-printing directives cause an error.
1054 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1055 * incompatible change: the internal functions
1056 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1057 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1058 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1059 instead of the old functions.
1060 * new feature: on platforms where "dladdr" is available foreign
1061 function names now appear in backtraces. (based on Helmut Eller's
1063 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1064 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1066 * documentation: ASDF manual is now distributed with SBCL.
1067 * bug fix: SBCL can now load its contributed modules with REQUIRE
1068 even if the system-provided entries have been removed from
1069 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1071 * bug fix: backtraces involving undefined functions or assembly
1072 routines are more informative. (thanks to Brian Downing)
1073 * bug fix: mutually referent alien structures now work correctly.
1074 (reported by Rick Taube)
1075 * bug fix: structures defined by WITH-ALIEN can be referred to
1076 within other definitions in the same WITH-ALIEN.
1077 * bug fix: division operators (MOD, TRUNCATE and the like) with
1078 constant zero divisors and integer dividends no longer generate
1080 * bug fix: provide default methods for INPUT-STREAM-P and
1081 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1082 * bug fix: improve the detection and resolution of MOP metacycles
1083 (where effective method computation depends on the generic function
1084 having its effective method computed). (reported by Bruno Haible)
1085 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1086 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1087 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1088 * bug fix: pathnames with relative directory components can be
1089 represented relative to default pathnames.
1090 * optimization: in taking the GCD of bignums, reduce the two bignums
1091 to approximately the same size (using Euclid's algorithm) before
1092 applying the more sophisticated binary GCD. (thanks to Juho
1094 * optimization: COUNT on bitvectors now operates word-at-a-time.
1095 * optimization: ASH with a positive, but not necessarily constant,
1096 (leftwards) shift, when in a modular context, is compiled to a
1098 * fixed some bugs revealed by Paul Dietz' test suite:
1099 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1100 if the corresponding argument is NIL.
1101 ** FORMAT directives accepting a minpad parameter treat negative
1103 ** literal commas as character parameters to FORMAT directives are
1105 ** literal spaces directly after ~<Newline> directives within a
1106 format-logical-block (~:< ~@:>) do not induce :FILL-style
1107 conditional newlines.
1108 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1109 as specified: it no longer includes conditional newlines).
1110 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1112 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1113 :PER-LINE-PREFIX argument does not evaluate to a string.
1115 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1116 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1117 the manual for details; for now, package locks can be disabled by
1118 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1119 if no major problems are found then it is likely that they will be
1120 compiled in unconditionally.
1121 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1122 unsupported operators on all platforms. To load a shared library
1123 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1124 object file, link it into a shared library outside of SBCL and
1125 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1127 * bug workaround: ROOM T and the GC were not working together
1128 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1129 reporting facility aren't preserved. That reporting has been
1130 disabled, so now until and unless someone figures out how to make it
1131 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1132 * minor incompatible change: as threatened around sbcl-0.8.0, the
1133 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1134 The symbols are also exported from SB-PCL for backwards
1135 compatibility, but more so than before SB-PCL should be treated as
1136 an implementation-internal package.
1137 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1139 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1140 It is known as of this release not to work on the Alpha, however.
1141 * fixed bug #167: errors signalled due to illegal syntax in method
1142 bodies are now more legible.
1143 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1144 designators and can hence be used with TYPEP.
1145 * fixed bug #333: CHECK-TYPE now ensures that the type error
1146 signalled, if any, has the right object to be accessed by
1147 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1148 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1149 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1151 * fixed bug #334: programmatic addition of slots using specialized
1152 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1153 and :ALLOCATION :CLASS slots.
1154 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1155 float exponents. (rereported by Peter Seibel)
1156 * fixed a bug: #\Space (and other whitespace characters) are no
1157 longer considered to be macro characters in standard syntax by
1158 GET-MACRO-CHARACTER.
1159 * fixed bug: initialization of condition class metaobjects no longer
1160 causes an instance of the condition to be created. (reported by
1162 * fixed bug: it is now possible to have slots such that
1163 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1164 description is neither :INSTANCE nor :CLASS.
1165 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1166 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1167 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1168 non-standard SLOT-DEFINITION-ALLOCATION values.
1169 * fixed bugs: various race conditions handling exiting threads.
1170 CL-PPCRE's thread torture test now passes.
1171 * fixed arguably-a-bug: GC time proportional to number of threads in
1172 system even when most of them are idle
1173 * optimization: improved performance of BIT and SBIT on bit-vectors.
1174 * .fasl file incompatibility: The fasl file version number has
1175 been incremented because of changes associated with package locks.
1177 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1178 * minor incompatible change: the system no longer provides
1179 optimization hints (conditions which are TYPEP
1180 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1181 mode; these hints are still emitted when higher SPEED optimization
1183 * new contrib module: a sampling profiler (profiling by statistical
1184 sampling, rather than by instrumenting functions) is available as
1185 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1186 * the behaviour of the standard function ED is now customizeable by
1187 third parties through a hook variable: see ED's documentation
1188 string for information on the protocol.
1189 * the compiler no longer emits efficiency notes for (FUNCALL X)
1190 when the type of X is uncertain under default optimization
1192 * fixed bug 276: mutating a binding of a specialized parameter to a
1193 method to something that is not TYPEP the specializer is now
1195 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1196 exhibit the required behaviour on the x86 platform. (thanks to
1197 Peter van Eynde, Eric Marsden and Bruno Haible)
1198 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1199 even for difficult arguments. (reported by Peter Graves)
1200 * fixed bug 141a: the backquote printer now descends quoted
1202 * fixed another bug in backquote printing: no more destructive
1203 modification of the form's list structure. (reported by Brian
1205 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1206 move between its address being taken and the call to
1207 interrupt_thread, fixing a crashing race condition.
1208 * the SB-POSIX contrib implementation has been adjusted so that it
1209 no longer exhibits ridiculously poor performance when constructing
1210 instances corresponding to C structs.
1212 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1213 * minor incompatible change: the sb-grovel contrib now treats C
1214 structures as alien (in the sense of SB-ALIEN) objects rather than
1215 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1216 has implications for memory management of client code
1217 (sb-grovel-returned objects must now be manually managed) and for
1218 type safety (alien objects now have full types).
1219 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1220 used to control emission of compiler diagnostics, rather than the
1221 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1222 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1223 quality should be considered deprecated.
1224 * install.sh now installs the user manual as well
1225 * (not quite a new documentable feature, but worth considering in
1226 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1227 beginnings of a semantically meaningful condition hierarchy is
1228 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1230 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1231 designator as the defaults argument.
1232 * fixed bug: Displaced arrays whose displaced-to array has become
1233 too small now cause ARRAY-DIMENSION to signal an error, providing
1234 for safer bounds-checking. (reported by Bruno Haible)
1235 * fixed bug: DEFCLASS slot definitions with identical :READER and
1236 :WRITER names now signal a reasonable error. (reported by Thomas
1238 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1240 * fixed bug: Invalid slot specification errors now print correctly.
1241 (thanks to Zach Beane)
1242 * fixed bug 320: Shared to local slot value transfers in class
1243 redefinitions now happen corrently from superclasses as
1244 well. (reported by Bruno Haible)
1245 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1247 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1248 improper lists correctly. (reported by Bruno Haible)
1249 * fixed bug 313: source-transform for <fun-name> was erroneously
1250 applied to a call of a value of a variable with name <fun-name>.
1251 (reported by Antonio Menezes Leitao)
1252 * fixed bug 307: The obsolete instance protocol ensures that
1253 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1254 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1255 remove dead unknown-values globs from the middle of the stack.
1256 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1257 easier to resume long computations after using *BREAK-ON-SIGNALS*
1258 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1259 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1260 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1261 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1262 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1263 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1265 * fixed a (fairly theoretical) bug in string printing: if
1266 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1267 does not have array-element-type equal to the most general string
1269 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1270 function name. (thanks to Nikodemus Siivola)
1271 * fixed bug: compile-time format string checker failed on
1272 non-closed ~{. (reported by Thomas F Burdick)
1273 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1274 don't warn on legal loop code involving more than one aggregate
1275 boolean. (thanks to Nikodemus Siivola)
1276 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1277 in keyword position to MAKE-PACKAGE was not being reported as
1279 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1280 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1282 * fixed bugs 280 and 312: the checking for multiple definitions in a
1283 file is less likely to become confused by uses of inline
1285 * fixed bug: the #S reader macro performs the keyword coercion
1286 specified for slot names. (reported by Kalle Niemitalo)
1287 * fixed bug: lambda lists may contain symbols whose names start with
1288 & but are not lambda-list-keywords; their occurrence triggers a
1290 * fixed bug 321: define-method-combination argument lambda lists do
1291 not cause invalid code to be generated when &optional or &aux
1292 variables have default values. (reported by Bruno Haible)
1293 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1294 have CLOS classes; this appears also to have fixed reported
1295 instability in stack exhaustion detection.
1296 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1297 MUFFLE-WARNING when no associated restart is present is now
1299 * optimization: rearranged the expansion of various defining macros
1300 so that each expands into only one top-level form in a
1301 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1303 * optimization: used a previously little-used slot in symbols to
1304 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1306 * fixed some bugs revealed by Paul Dietz' test suite:
1307 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1309 ** Fixed an optimization bug related to inheritance of initforms
1310 from local to shared slots.
1311 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1312 ** CAST optimizer forgot to flush argument derived type.
1313 ** print/read consistency on floats is now orders of magnitude
1314 more likely. (thanks also to Bruno Haible for a similar report
1316 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1317 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1318 for objects of type REAL. Make it so.
1319 ** SXHASH is specified (infelicitously) to respect similarity,
1320 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1321 it so. (thanks to Markus Ziegler)
1322 ** on the Alpha, the compiler succeeds in compiling functions
1323 returning a known number of arguments greater than 63.
1324 ** fixed handling of invalid NIL arguments in keyword position
1326 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1328 ** provided a readably-printable representation for RANDOM-STATE
1330 ** ensured that pathnames, where they have a namestring, always
1331 print using #P"..." syntax.
1333 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1334 * Support for the forthcoming 2.0 version of the NetBSD kernel
1335 running on x86 hardware has been added. (thanks to Perry
1336 E. Metzger most immediately, and others for their past work)
1337 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1338 current release version; SBCL's OpenBSD support had been broken
1339 since about the time of OpenBSD's switch to ELF binary format.)
1340 * [placeholder for DX summary]
1341 ** user code with &REST lists declared dynamic-extent, under high
1342 speed or space and low safety and debug optimization policy.
1343 * The manual has been converted to Texinfo format and the debugger
1344 chapter from the cmucl manual has been added.
1345 * A facility has been added to extract documentation strings from
1346 sbcl and store them as Texinfo-formatted snippets for inclusion in
1347 the manual (via Texinfo's @include directive)
1348 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1349 the test case to Dave Roberts)
1350 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1351 pointers. (reported by Sean Ross)
1352 * bug fix: PROFILE output is printed nicely even for large numerical
1353 values. (thanks to Zach Beane)
1354 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1355 greater than 32 handle EOF correctly.
1356 * bug fix: on X86 an immediate argument of the IMUL instruction is
1357 correctly printed in disassembly. (thanks to Lutz Euler)
1358 * bug fix: class slots in redefined classes preserve their old
1359 values. (thanks to Bruno Haible and Nikodemus Siivola)
1360 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1361 succeeds. (reported by Marco Baringer)
1362 * fixed some bugs revealed by Paul Dietz' test suite:
1363 ** READ-SEQUENCE now works on ECHO-STREAMs.
1364 ** RATIONALIZE works more according to its specification. (thanks
1367 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1368 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1369 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1370 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1371 should matter to you only if you rebind the printer control
1372 variables and then find you want different bindings in the
1373 debugger than in the ordinary execution of your program.)
1374 * The runtime build system has been tweaked to support building
1375 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1376 assembler and linker. (thanks to Nikodemus Siivola)
1377 * Unbound, undefined, undeclared variables now trigger full
1378 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1379 more often programmer error than deliberate exploitation of undefined
1381 * optimization: the hash algorithm for strings has changed to one
1382 that is less vulnerable to spurious collisions. (thanks to Juho
1384 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1385 needless bounds checking. (thanks to Juho Snellman)
1386 * optimization: implemented multiplication as a modular
1387 (UNSIGNED-BYTE 32) operation on the PPC backend.
1388 * fixed some bugs revealed by Paul Dietz' test suite:
1389 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1391 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1392 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1393 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1395 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1397 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1399 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1400 their output stream on EOF from read.
1401 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1402 have been read to end-of-file.
1403 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1405 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1406 description of determination of which consecutive characters
1408 ** Printing the "Space" character with escaping on now yields "#\\ ",
1409 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1410 ** Reading floating-point numbers with *READ-BASE* set to a number
1411 less than 10 works correctly.
1412 ** Reading floating-point numbers with *READ-BASE* set to a number
1413 more than 10 works correctly.
1414 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1415 the readtable currently in effect.
1417 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1418 * minor incompatible change: parsing of namestrings on a physical
1419 (Unix) host has changed; numbers after the final #\. in a
1420 namestring are no longer interpreted as a version field. This is
1421 intented to be largely invisible to the user, except that the
1422 meaning of the namestring "*.*.*" has changed: it now refers to a
1423 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1424 should usually be replaced by
1425 (make-pathname :name :wild :type :wild :version :wild)
1426 with the added benefit that this is more likely to be portable.
1427 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1428 option to OPEN now signals an error if the file being opened
1429 exists; this may have an impact on existing code.
1430 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1431 signals received, on the PowerPC platforms (both Linux and
1432 Darwin). (thanks to Pierre Mai for pointing out the location of
1434 * several fixes on OS X: The system now builds and runs cleanly on
1435 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1436 (thanks to Brian Mastenbrook)
1437 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1438 second-resolution: integer multiples of 1/3600 between -24 and 24.
1439 (thanks to Vincent Arkesteijn)
1440 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1441 type when called with 1 argument; PEEK-CHAR checked type of
1442 PEEK-TYPE only after having read first character from a
1443 stream. (reported by Peter Graves)
1444 * bug fix: the garbage collector now has much better locality
1445 behaviour, and in particular no longer treats all memory as being
1446 exhausted when many small objects point to each other in a deeply
1448 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1449 recognized as being TYPEP their class.
1450 * bug fix: the PUSHNEW documentation string has been corrected.
1451 (thanks to Vincent Arkesteijn)
1452 * bug fix: defaulting of the value for the last of an atypically
1453 large number of multiple values being bound was not being
1454 performed correctly on the Alpha or PPC platforms
1455 * optimization: implemented multiplication as a modular
1456 (UNSIGNED-BYTE 32) operation on the x86 backend.
1457 * optimization: SEARCH on simple-base-strings can now be open-coded.
1458 (see also contrib/compiler-extras.lisp for inspiration for
1459 teaching the compiler about the Boyer-Moore algorithm).
1460 * value, returned by MAX (and MIN) called with several EQUALP, but
1461 not EQL, arguments now does not depend on compiler settings.
1462 * fixed some bugs revealed by Paul Dietz' test suite:
1463 ** in stack analysis liveness information is propagated from
1464 non-local entry points.
1465 ** pathwise CAST removing failed when the CAST node did not start
1467 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1468 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1470 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1471 host is already defined.
1472 ** RENAME-FILE works on streams instead of signalling an internal
1474 ** PEEK-CHAR uses the current readtable when determining whether
1475 or not a character is whitespace.
1476 ** MERGE-PATHNAMES handles the case when the pathname does not
1477 specify a name while the default-pathname specifies a version.
1478 ** Pathnames now stand a better chance of respecting print/read
1480 ** Attempting to use standardized file system operators with a
1481 pathname with invalid :DIRECTORY components signals a
1483 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1485 ** DIRECTORY on logical pathnames is more correct.
1486 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1487 signal a TYPE-ERROR if their argument is not a stream.
1488 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1489 designator argument does not designate a stream.
1490 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1491 examining the synonym.
1492 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1494 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1495 element-type larger than ([UN]SIGNED-BYTE 32).
1497 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1498 * When built with the :SB-FUTEX feature, threaded builds now take
1499 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1500 for faster/more reliable mutex and condition variable support.
1501 * Incompatible change (but one you probably shouldn't have been using
1502 anyway): the interface and code for arbitrating between multiple
1503 threads in the same user session has been redesigned.
1504 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1505 error when called without an explicit environment argument.
1506 (thanks to Nikodemus Siivola)
1507 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1508 confused about their position. (thanks to Adam Warner and Gerd
1510 * bug fix: Pretty printing backquoted forms with unquotations in the
1511 argument list position of various code constructs such as LAMBDA
1512 now works correctly. (reported by Paul Dietz)
1513 * bug fix: Pretty printing unquotations no longer loses all
1514 stream position information.
1515 * optimization: performance of string output streams is now less
1516 poor for multiple small sequence writes.
1517 * optimization: performance of CSUBTYPEP in the presence of complex
1518 expressions involving CONS and NOT many times has been improved.
1519 (reported by Paul Dietz)
1520 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1522 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1523 streams for the REPL, for the debugger, and so on. (thanks to
1525 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1526 * fixed some bugs revealed by Paul Dietz' test suite:
1527 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1528 is ignored if it is too small, rather than propagating through
1529 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1530 ** extremely complex negations of CONS types were not being
1531 sufficiently canonicalized, leading to inconsistencies in
1533 ** VALUES tranformer lost derived type.
1535 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1536 * fixed a bootstrapping bug: the build process no longer assumes
1537 that the various BOOLE-related constants have the same value in
1538 host and target lisps. (noted by Paul Dietz' test suite on an
1539 SBCL binary built from CLISP)
1540 * The system can now be dynamically linked on the MIPS platform,
1541 which enables dynamic loading of foreign code from Lisp. (thanks
1542 to Ralf Baechle for discussions on the MIPS ABI)
1543 * The system now records debugging information for its own source
1544 files in a filesystem-position-independent manner, relative to
1545 the "SYS" logical host.
1546 * fixed a compiler bug: MV-LET convertion did not check references
1547 to the "max args" entry point. (reported by Brian Downing)
1548 * tweaked disassembly notes to be less confident about proclaiming
1549 some instruction as an LRA. (thanks to Brian Downing)
1550 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1551 now each have their own history, command character, and other
1552 characteristics. (thanks to David Lichteblau)
1553 * fixed some bugs revealed by Paul Dietz' test suite:
1554 ** compiler failure in compiling LOGAND expressions including a
1556 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1557 shift greater than 32.
1558 ** FLUSH-DEST did not mark blocks for type check regeneration.
1559 ** HANDLER-CASE failed to accept declarations in handler clauses
1560 in some circumstances.
1562 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1563 * New code in contrib/sb-introspect (still probably not entirely
1564 stable yet) provides some support for smart Lisp development
1565 environments like SLIME.
1566 * The conditions signalled for errors occurring when loading .fasl
1567 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1568 in a way which should help ASDF recover gracefully.
1569 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1570 stuff has been cleaned up. If you code contrib/ stuff, this might
1571 affect you, and you can look at contrib/README, contrib/STANDARDS,
1572 and/or the 0.8.4.27 diff to check.
1573 * In full calls the compiler now does not generate checks for declared
1574 argument types for all arguments.
1575 * various threading fixes
1576 ** and some experimental patches which didn't make it into
1577 the main tree for this release, but which are shipped in
1578 contrib/experimental-thread.patch as a possible fix for some
1579 failures (deadlock, spinning...) in GC-intensive multithreaded
1581 * fixed PPC build problem (source code incompatibility of different
1582 library versions): added offsetof() hackery which attempts to divine
1583 where glibc maintainers put uc_mcontext today
1584 * fixed bug 282: compiler does not trust type assertions while passing
1585 arguments to a full call.
1586 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1587 &OPTIONAL VALUES type parameter.
1588 * fix bug 214: algorithm for noting rejected templates is now more
1589 similar to that of template seletion. (also reported by rydis on
1591 * fixed bug 141b: printing backquoted information readably and prettily
1592 inserts a space where necessary.
1593 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1594 (CHARACTER 10) are now reported as errors, rather than propagated
1595 as unknown types. (reported by piso on #lisp)
1596 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1597 with values NIL and :ERROR. (thanks to Milan Zamazal)
1598 * fixed bug 191c: CLOS now does proper keyword argument checking as
1599 described in CLHS 7.6.5 and 7.6.5.1.
1600 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1601 longer raise an error; further, using a list as a for-as-arithmetic
1602 counter now raises a meaningful error.
1603 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1604 now understood by sequence creation functions such as MAKE-SEQUENCE
1606 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1607 a STRING-INPUT-STREAM.
1608 * compiler enhancement: SIGNUM is now better able to derive the type
1610 * type declarations inside WITH-SLOTS are checked. (reported by
1612 * fixed some bugs revealed by Paul Dietz' test suite:
1613 ** incorrect optimization of TRUNCATE for a positive first
1614 argument and negative second.
1615 ** compiler failure in let-convertion during flushing dead code.
1616 ** compiler failure while deriving type of TRUNCATE on an
1617 interval, containing 0.
1618 ** ASH of a negative bignum by a negative bignum count now returns
1620 ** intersection of CONS types now canonicalizes properly, fixing
1621 inconsistencies in SUBTYPEP.
1623 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1624 * incompatible change: The --disable-debugger command line
1625 option now clobbers the debugger at a more fundamental
1626 level, by redefining #'INVOKE-DEBUGGER instead of by
1627 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1628 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1629 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1630 enter the debugger REPL and then suffer recursive errors
1631 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1632 while under the new system, BREAK in --disable-debugger mode
1633 terminates the system just as an unhandled error would.
1634 * fixed compiler performance when processing loops with a step >1;
1635 * bug fix: DOCUMENTATION now retrieves generic function
1636 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1637 support has been systematized, and now supports the methods
1638 specified by ANSI, along with a default method and a method for
1639 slot documentation. (reported by Nathan Froyd)
1640 * bug fix: effective methods associated with a generic function are
1641 no longer cached over a change of that generic function's method
1642 combination. (reported by Andreas Fuchs)
1643 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1644 names a non-existent directory. (thanks to Andreas Fuchs)
1645 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1646 the PPC platform, lead to stack corruption; this has been fixed.
1647 (reported by Rainer Joswig)
1648 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1649 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1650 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1651 arithmetic in general, is now much more reliable on the PPC
1653 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1654 platform now returns the right answer.
1655 * optimization: restored some effective method precomputation in
1656 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1657 precomputation is now tunable.
1658 * optimization: compiler-internal data structure use has been
1659 reviewed, and changes have been made that should improve the
1660 performance of the compiler by about 20%.
1661 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1662 simple-base-strings and simple-bit-vectors is improved.
1663 * optimization: the optimization of 32-bit logical and arithmetic
1664 functions introduced in version 0.8.3 on the x86 has been
1665 implemented on the mips, ppc and sparc platforms; an
1666 implementation of the same facility, but for 64-bit arithmetic,
1667 has been added for the alpha.
1668 * microoptimization: the compiler is better able to make use of the
1669 x86 LEA instruction for multiplication by constants.
1670 * bug fix: in some situations compiler did not report usage of
1671 generic arithmetic in (SPEED 3) policy.
1672 * bug 145b fix: compiler used wrong type specifier while converting
1673 MEMBER-types to numeric.
1674 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1676 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1677 index argument now works properly on the Alpha platform.
1678 * bug fix: floating point exception treatment on the Alpha platform
1680 * bug fix: FILE-POSITION works much better on string input and
1681 output streams. (thanks to Nikodemus Siivola)
1682 * bug fix: many threading/garbage collection symptoms sorted.
1683 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1684 might be pseudo-atomic.
1685 * internal change: Stopping for GC is now done with signals not
1686 ptrace. GC is now done in whichever thread wanted it, instead of
1688 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1690 * bug fix: result form in DO is not contained in the implicit
1692 * incompatible change: ICR structure is changed; the value part of
1693 CONTINUATION is now called LVAR; corresponding functions are
1694 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1695 * added type deriver for ISQRT (thanks to Robert E. Brown).
1696 * bug fix: better support for loading from the command line when an
1697 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1699 * fixed some bugs revealed by Paul Dietz' test suite:
1700 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1701 ** ROUND and FROUND now give the right answer when given very
1702 small float arguments.
1703 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1705 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1706 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1707 ** LCM should return a non-negative integer.
1708 ** PARSE-INTEGER returned the index of a terminator instead of the
1709 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1710 ** PARSE-INTEGER returned an incorrect index being applied to a
1712 ** LCM with two arguments of 0 returns 0 rather than signalling
1714 ** unsigned addition of a 32-bit constant with the high bit set no
1715 longer causes an internal compiler error on the x86.
1716 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1717 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1718 with negative last argument.
1719 ** byte specifiers with zero size and position no longer cause
1720 an error during type derivation.
1721 ** bignum multiplication on the Alpha platform now returns the
1723 * porting: The system now builds on SuSE AMD64, although it still
1724 generates a 32-bit binary.
1725 * .fasl file incompatibility: The fasl file version number has
1726 been incremented (because of the changes to internal compiler
1727 data structures referred to above).
1729 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1730 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1731 more accurately, on the Darwin kernel running on PowerPC hardware.
1732 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1733 * Compiler code deletion notes now signal a condition of type
1734 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1735 an associated MUFFLE-WARNING restart.
1736 * The compiler now performs limited argument count validation of
1737 constant format strings in FORMAT, and where appropriate in ERROR,
1738 CERROR and WARN. (thanks to Gerd Moellmann)
1739 * New ASDF-INSTALL contrib can be used for automatic download and
1740 installation of third-party Lisp code from CCLAN or other sites
1742 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1743 which forces another thread to execute a function supplied by the
1745 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1746 now accept and act upon their :ELEMENT-TYPE keyword argument.
1747 (reported by Martin Atzmueller, Edi Weitz)
1748 * bug fix: FILE-POSITION now accepts position designators up to
1749 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1750 is the greater. (thanks to Patrik Nordebo)
1751 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1752 Kalle Olavi Niemitalo)
1753 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1754 printable. (reported by Eric Marsden)
1755 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1756 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1757 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1758 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1759 (thanks to Stig Sandoe)
1760 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1761 UNREAD-CHAR now correctly includes the unread character in the
1762 target sequence. (thanks to Gerd Moellmann)
1763 * bug fix in threaded builds: the system can now be suspended and
1764 resumed by shell job control with minimal disruption.
1765 * bug fixes in times and timezones >2038AD
1766 * better handling of "where is GNU make?" problem in build scripts
1767 (thanks to Nikodemus Siivola)
1768 * new optimization: inside a named function any reference to a
1769 function with the same name is considered to be a self-reference;
1770 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1771 optimization quality.
1772 * new optimization on x86: logical functions and + now have
1773 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1774 used when the result is truncated to 32 bits.
1775 * VALUES declaration is partially enabled.
1776 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1777 * fixed some bugs revealed by Paul Dietz' test suite:
1778 ** The system now obeys the constraint imposed by
1779 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1780 types form a lattice under type intersection.
1781 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1782 ** ASSOC now ignores NIL elements in an alist.
1783 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1784 and (1+ MOST-POSITIVE-FIXNUM) answers.
1785 ** The addition of a method with invalid qualifiers to a generic
1786 function does not cause an error to be signalled immediately;
1787 a warning is signalled, and the error is generated only on
1788 calling the generic function.
1789 * changed .fasl file version number, in part to add type codes for
1790 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1791 obscure ANSI requirements
1793 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1794 * fixed bug 148: failure to inline-expand a local function left
1795 garbage, confusing the compiler.
1796 * fixed bugs 3cd: structure slot readers perform type check if the
1797 slot can have an invalid value (i.e. it is either not initialized
1798 or can be written with a less specific slot writer).
1799 * bug fix: the compiler now traps array references to elements off
1800 the end of an array; previously, the bounds checking in some
1801 circumstances could go off-by-one.
1802 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1804 * type declarations for array element types now obey the description
1805 on the CLHS page "Declaration TYPE", as per discussions on
1806 sbcl-help around 2003-05-08. This means that a declaration
1807 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1808 declaration, all references to BAR will be asserted or assumed
1809 (with THE, so dependent on compiler policy) to involve objects of
1810 type FOO. Note that no such declaration is implied in
1811 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1812 * declared types of functions from the "Conditions"
1813 chapter. (reported by Paul Dietz)
1814 * bug fix: CERROR accepts a function as its first argument.
1815 * bug fix: NTH an NTHCDR accept a bignum as index
1816 arguments. (reported by Adam Warner)
1817 * optimization: character compare routines now optimize comparing
1818 against a constant character. (reported by Gilbert Baumann)
1819 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1820 argument now works properly on the MIPS platform.
1821 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1822 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1823 declared to have a functional type, does not cause an error
1825 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1826 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1827 * fixed bug: if last continuation of a deleted block has a
1828 destination, this destination should be deleted too. (reported by
1830 * fixed a bug in the bootstrap process: the host compiler's values
1831 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1832 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1833 test case from Patrik Nordebo)
1834 * improved the ability of the disassembler on the PPC platform to
1835 provide helpful disassembly notes.
1836 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1837 the class in more cases than previously.
1838 * bug fix: FILE-POSITION now understands :START and :END for
1839 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1840 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1841 8) data. (thanks to David Lichteblau)
1842 * bug fix: it is possible to add a method to a generic function
1843 without lambda list.
1844 * bug fix: reader failed to signal END-OF-FILE inside an
1845 object representation. (reported by Nikodemus Siivola)
1846 * fixed some bugs revealed by Paul Dietz' test suite:
1847 ** LAST and [N]BUTLAST should accept a bignum.
1848 ** condition slot accessors are methods.
1849 ** (VECTOR NIL) is a subtype of STRING.
1851 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1852 * minor incompatible change: some nonsensical specialized lambda
1853 lists (used in DEFMETHOD) which were previously ignored now signal
1855 * minor incompatible change: the system is now aware of the types of
1856 variables in the COMMON-LISP package, and will signal errors for
1857 most violations of these type constraints (where previously they
1858 were silently accepted).
1859 * minor incompatible change: COMPILE-FILE now uses the freedom
1860 afforded (ANSI 3.2.2.3) to use derived function types for
1861 functions defined in the same file. This also permits the system
1862 to warn on static type mismatches and function
1863 redefinition. (Currently it does not work with high DEBUG level.)
1864 * minor incompatible change: VALUES declaration is disabled.
1865 * When issuing notes, the compiler now signals a condition of type
1866 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1867 restart for use in user handlers. It is expected that the
1868 COMPILER-NOTE condition will eventually become a condition
1869 supertype to a hierarchy of note types, which will then be
1870 handleable in a similar fashion. However, at the moment, no such
1871 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1872 but it's an implementation detail, not a classification for the
1874 * Changes in type checking closed the following bugs:
1875 ** type checking of unused values (192b, 194d, 203);
1876 ** template selection based on unsafe type assertions (192c, 236);
1877 ** type checking in branches (194bc).
1878 * A short form of VALUES type specifier has ANSI meaning (it has
1879 increased the number of situations when SBCL cannot perform type
1881 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1882 keywords or constants is permissible.
1883 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1884 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1885 classes. (thanks to Antonio Martinez)
1886 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1887 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1888 argument to a pretty-print stream. Instead, it leaves any such
1889 operations to DESCRIBE-OBJECT methods.
1890 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1892 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1893 lambda lists. (thanks to David Lichteblau)
1894 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1895 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1896 (thanks to Gerd Moellmann)
1897 * fixed bug 246: increased compilation speed of long
1898 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1900 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1902 * DEFCONSTANT now throws a condition of type
1903 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1904 constant to a non-EQL value; CONTINUE and ABORT restarts
1905 respectively change and preserve the value.
1906 * fixed bug 63: The code walker, part of the implementation of CLOS,
1907 is now better at handling symbol macros.
1908 * bug fix: There is no longer an internal implementation type named
1909 CL:LENGTH. (reported by Raymond Toy)
1910 * bug fix: In macro-like defining macros/special operators the
1911 implicit block does not enclose the lambda list.
1912 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1913 enclosed in parentheses) are not suitable as type specifiers, and
1914 their use properly signals an error now.
1915 * bug fix: An argument count mismatch for a type specifier in code
1916 being compiled no longer causes an unhandled error at compile
1917 time, but signals a compile-time warning.
1918 * fixed simple vector readable printing
1919 * bug fix: DESCRIBE takes more care over whether the class
1920 precedence list slot of a class is bound before accessing it.
1921 (reported by Markus Krummenacker)
1922 * bug fix: FORMATTER can successfully compile pretty-printer format
1923 strings which use variants of the ~* directive inside.
1924 * bug fix: SEARCH now applies its TEST predicate to the elements of
1925 the arguments in the correct order. (thanks to Wolfhard Buss)
1926 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1927 in type inference. (thanks to Robert E. Brown)
1928 * bug fix: Reading in symbols with an explicit package name of ""
1929 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1930 * fixed some bugs revealed by Paul Dietz' test suite:
1931 ** NIL is now allowed as a structure slot name.
1932 ** Arbitrary numbers, not just REALs, are allowed in certain
1933 circumstances in LOOP for-as-arithmetic clauses.
1934 ** Multiple class redefinitions before slot access no longer
1935 causes a type error.
1936 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1937 association between the name and a class.
1938 ** Generic functions with non-standard method-combination and over
1939 five methods all of which return constants no longer return NIL
1940 after the first few invocations. (thanks to Gerd Moellmann)
1941 ** CALL-NEXT-METHOD with no arguments now passes the original
1942 values of the arguments, even in the presence of assignment.
1943 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1945 ** DISASSEMBLE works with closures and funcallable instances.
1946 ** ADD-METHOD now returns the generic function, not the new method.
1947 ** FIND-METHOD signals an error if the lengths of the specializers
1948 is incompatible with the generic function, even if the ERRORP
1950 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1951 which its argument is a member.
1952 ** DEFCLASS only redefines the class named by its class-name
1953 argument if that name is the proper name of the class;
1954 otherwise, it creates a new class.
1955 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1956 of the UNBOUND-SLOT condition to the name of the slot.
1957 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1958 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1959 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1960 SLOT-MAKUNBOUND in the specified fashion.
1962 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1963 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1964 cross-compilation host. As a consequence, we can now bootstrap our
1965 way up to SBCL starting with a bare gcc toolchain and human-readable
1966 source code (first the source to CLISP, then the source to SBCL).
1967 * A contributed module containing a partial implementation of the
1968 simple-streams interface has been included. (thanks to Rudi
1970 * A contributed module implementing the RFC1321 Message Digest
1971 Algorithm, known as MD5, has been included.
1972 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1973 no longer has any effect, as the code controlled by this feature
1974 has been deleted. (As far as we know, no-one has ever built using
1975 this feature, and its semantics were confused in any case).
1976 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1977 work on conditions (as required by the ANSI specification),
1978 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1979 expected behaviour on conditions. Users should note, however,
1980 that such behaviour is not required by the ANSI specification,
1981 and so use of this behaviour may render their code unportable.
1982 * fixed some bugs revealed by Paul Dietz' test suite:
1983 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1985 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1986 ** SXHASH on condition objects no longer returns NIL.
1987 ** :ALLOCATION :CLASS slots are better treated; their values are
1988 updated on class redefinition, and initforms inherited from
1989 superclasses are applied.
1990 ** REMOVE-METHOD returns its generic function argument even when
1991 no method was removed.
1992 ** SHARED-INITIALIZE now initializes the values of the requested
1993 slots, including those with :ALLOCATION :CLASS.
1994 ** ALLOCATE-INSTANCE now works on structure classes defined via
1995 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1997 ** SLOT-EXISTS-P now works on conditions, as well as structures
1999 ** MAKE-LOAD-FORM now has the required methods on
2000 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2001 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2002 keyword, and now implements the SLOT-NAMES argument.
2003 ** methods with &OPTIONAL arguments no longer allow too many
2004 arguments to be passed in the call without error.
2005 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2006 option is consistent with the required arguments of the generic
2007 function lambda list.
2008 * bug fix: REQUIRE accepts a string designator. (Thanks to
2010 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2011 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2012 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2013 * bug fix: The compiler now checks for duplicated variables in macro
2015 * bug fix: SETQ on globals returns the correct value.
2016 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2017 not a COMPILER-ERROR (followed by some other strange error on
2018 choosing the CONTINUE restart).
2019 * bug fix: make.sh and friends are now more consistent in the way that
2020 they look for GNU "make".
2022 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2023 * experimental native threads support (on x86 Linux >=2.4 only).
2024 This is not compiled in by default: you need to add :SB-THREAD to
2025 the target features. See the "Beyond ANSI" chapter of the manual
2027 * fix for longstanding nonANSIism: The old distinction between
2028 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2029 The return value from CL:FIND-CLASS is now a CLOS class, and
2030 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2031 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2032 * An interface to the MetaObject Protocol, as described in Kiczales,
2033 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2034 MIT Press, 1991, is available from the SB-MOP package.
2035 * incompatible change: the SB-PCL package should now be considered
2036 a private implementation detail, and no longer a semi-private MOP
2038 * minor incompatible change: due to rearrangement for threads, the
2039 control stack and binding stack are now allocated at arbitrary
2040 addresses instead of being hardcoded per-port. Users affected by
2041 this probably have to be doing advanced things with shared
2042 libraries, and will know who they are.
2043 * minor incompatible change: Previously, all --eval forms used were
2044 processed with READ before any of them were processed with EVAL.
2045 Now each --eval form is processed with both READ and EVAL before
2046 the next --eval form is processed. (Thus package operations like
2047 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2048 work as the user might reasonably expect.)
2049 * minor incompatible change: *STANDARD-INPUT* is now only an
2050 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2052 * minor incompatible change: Y-OR-N-P is now character-oriented,
2053 not line oriented. Also, YES-OR-NO-P now works without errors.
2054 (thanks to Antonio Martinez)
2055 * sb-aclrepl module improvements: an integrated inspector, added
2056 repl features, and a bug fix to :trace command.
2057 * Known functions, which cannot be open coded by the backend, are
2058 considered to be able to check types of their arguments. (fixing
2059 a bug report by Nathan J. Froyd)
2060 * fixed a bug in computing method discriminating functions: It is
2061 now possible to define methods specialized on classes which have
2062 forward-referenced superclasses. (thanks to Gerd Moellmann)
2063 * fixed evaluation order in optional entries (reported by Gilbert
2065 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2066 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2067 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2068 required-by-AMOP NAME argument, as well as CLASS and
2069 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2070 * fixed bug 20: DEFMETHOD can define methods using names that are
2071 not the proper names of classes to designate class specializers.
2072 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2073 Unix file descriptors, instead of blowing up. (thanks to Antonio
2075 * Garbage collection refactoring: user-visible change is that a
2076 call to the GC function during WITHOUT-GCING will not do garbage
2077 collection until the end of the WITHOUT-GCING. If you were doing
2078 this you were probably losing anyway.
2079 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2080 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2081 TYPEP the latter but not the former.
2082 * The compiler issues a full WARNING for calls to undefined functions
2083 with names from the CL package.
2084 * MAP-INTO for a vector destination is open coded. (reported by
2085 Brian Downing on c.l.l)
2086 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2087 documentation string.
2088 * fixed some bugs revealed by Paul Dietz' test suite:
2089 ** COPY-ALIST now signals an error if its argument is a dotted
2091 ** Condition slots are now accessed more correctly in the presence
2092 of multiple initargs for a given slot.
2093 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2094 exclude restarts of the same name associated with a different
2096 ** DEFCLASS of forward-referenced classes with another
2097 forward-referenced class in the superclasses list no longer
2099 ** Condition slots are now initialized once each, not multiple
2100 times. (thanks to Gerd Moellmann)
2101 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2103 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2104 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2105 arguments contain duplicated elements.
2106 ** RESTART-CASE understands local macros.
2107 ** RESTART-CASE associates exactly its own restarts with a condition.
2108 ** ENDP in safe mode checks its argument to be of type LIST.
2109 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2110 in question is unbound.
2111 ** Optimized MAKE-INSTANCE functions no longer cause internal
2112 assertion failures in the presence of duplicate initargs.
2113 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2114 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2115 propagated even in the ctor optimized implementation of
2117 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2119 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2120 * a better implementation of SXHASH on (simple) bit vectors,
2121 measured both in execution speed and in distribution of results
2122 over the positive fixnums, has been installed. Likewise, a better
2123 implementation of EQUAL for simple bit vectors is now available.
2124 * fixed CEILING optimization for a divisor of form 2^k.
2125 * fixed bug 240 (emitting extra style warnings "using the lexical
2126 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2127 by Antonio Martinez)
2128 * fixed SXHASH, giving different results for NIL depending on type
2129 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2130 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2131 variable. (found by Rolf Wester)
2132 * SBCL does not ignore type declarations for special
2133 variables. (reported by rif on c.l.l 2003-03-05)
2134 * some bug fixes in contrib/sb-aclrepl/
2135 * fixed some bugs revealed by Paul Dietz' test suite:
2136 ** a bug in the CONS type specifier, whereby the CAR and CDR
2137 types got intertwined, has been fixed;
2138 ** the type system is now able to reason about the interaction
2139 between INTEGER and RATIO types more completely;
2140 ** APPEND, [N]REVERSE and NRECONC check that those their
2141 arguments, which must be proper lists, are really so;
2142 ** An array specialized to be unable to hold elements has been
2143 implemented, as required -- yes, really -- by ANSI;
2144 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2145 on malformed property lists;
2147 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2148 * incompatible packaging change: in line with Unix convention,
2149 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2150 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2151 * REQUIRE and PROVIDE are now optionally capable of doing something
2152 useful. See the documentation string for REQUIRE.
2153 * infrastructure for a managed SBCL contrib system: contributed
2154 modules in this release include:
2155 ** the ASDF system definition facility;
2156 ** an interface to the BSD Sockets API;
2157 ** an ACL-like convenience interface to the repl;
2158 (thanks to Kevin Rosenberg)
2159 ** an implementation of ROTATE-BYTE, with efficient implementation
2161 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2162 gives a closer approximation to the right answer than previously.
2163 (thanks to Raymond Toy)
2164 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2165 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2166 optional environment arguments, as required by ANSI.
2167 * fixed bugs in other functions taking environment objects, allowing
2168 calls with an explicit NIL environment argument to be compiled
2170 * fixed bug 228: primary return values from
2171 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2172 COMPILE or FUNCTION.
2173 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2174 :TYPE will no longer signal errors on innocuous objects.
2175 * fixed bug 231b: SETQ is better at respecting type declarations in
2176 the lexical environment.
2177 * fixed a bug in DEFCLASS: classes named by symbols with no or
2178 unprintable packages can now be defined.
2179 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2180 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2181 * fixed compiler failure related to checking types of functions.
2182 (reported by Robert E. Brown)
2183 * the compiler is now much more consistent in its error-checking
2184 treatment of bounding index arguments to sequence functions: in
2185 (SAFETY 3) code, errors will be signalled in almost all cases if
2186 invalid sequence bounding indices are passed to functions defined
2187 by ANSI to operate on sequences.
2188 * fixed a bug in the build procedure: documentation of SBCL-specific
2189 packages is now preserved and available in the final Lisp image.
2190 * lifted FDEFINITION lookup out of loops in the implementation of
2191 many list operations. (thanks to Robert E. Brown)
2192 * fixed a bug in the reader: the #n# reader macro now works for
2193 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2194 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2195 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2196 E. Brown and Tony Martinez respectively)
2197 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2198 SLOT-BOUNDP: in particular, it is now able to optimize them much
2199 better, and is now not vulnerable to having packages renamed.
2200 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2201 always called when a slot is not present in an instance. (thanks
2203 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2204 longer signal an error on unreading a character following EOF on
2205 the previous constituent stream. (thanks to Tony Martinez)
2206 * fixed some bugs revealed by Paul Dietz' test suite:
2207 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2208 not just nonnegative fixnums;
2209 ** the logical bit-array operators such as BIT-AND now accept an
2210 explicit NIL for their "opt-arg" argument (to indicate a
2211 freshly-consed result bit-array);
2212 ** ELT now signals an error on an invalid sequence index in safe
2214 ** the type system is now cleverer about negations of numeric
2215 types, and consequently understands the BIGNUM and RATIO types
2217 ** the type system is now cleverer about the interaction between
2218 INTEGER and RATIO types: while bugs still remain, many more
2219 cases are accurately computed;
2220 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2221 if it is in the last clause;
2222 ** CONSTANTLY now correctly returns a side-effect-free function in
2224 ** DECLARE is no longer treated as a special-operator; in
2225 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2226 * incremented fasl file version number due to the change in the
2227 DEFSTRUCT-SLOT-DESCRIPTION structure.
2229 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2230 * minor incompatible change: code processed by the "interpreter" or
2231 EVAL now has a compilation optimization policy of (DEBUG 2)
2232 (changed from (DEBUG 1)) to improve debuggability of interactive
2233 development, and to allow the use of the debug RETURN command in
2235 * an experimental implementation of the RETURN command for the
2236 debugger has been included. (thanks to Frederik Kuivinen)
2237 * fixed bug 62: constraints were not propagated into a loop.
2238 * fixed bug in embedded calls of SORT (reported and investigated by
2240 * fixed some bugs revealed by Paul Dietz' test suite:
2241 ** printing and reading of arrays with some dimensions having
2242 length 0 (thanks to Gerd Moellmann);
2243 ** BOA constructor with &AUX argument without a default value does
2244 not cause a type error;
2245 ** CONSTANTP now returns true for all self-evaluating objects.
2247 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2248 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2249 accessors that are related by inheritance, as specified in the
2250 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2251 Valtteri Vuorikoski)
2252 * The compiler is now able to inline functions that were defined in
2253 a complex lexical environment (e.g. inside a MACROLET).
2254 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2256 * Support for the upcoming FreeBSD-5.0 release has been included.
2257 (thanks to Dag-Erling Smorgrav)
2258 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2259 effect when it is not in a toplevel context.
2260 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2261 stand a better chance of being correct. (thanks to Gerd
2263 * fixed bug in COERCE, which now signals an error on coercing a
2264 rational to a bounded real type which excludes the expected
2266 * The compiler is now able to derive types more accurately from the
2267 COERCE and COMPILE functions.
2268 * fixed bug 223: functional binding is considered to be constant
2269 only for symbols in the CL package.
2270 * fixed bug 231: SETQ did not check the type of a variable being set
2271 (reported by Robert E. Brown)
2272 * A new optimization for MAKE-INSTANCE has been included, fixing
2273 various bugs (including relating to :ALLOCATION :CLASS slots and
2274 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2276 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2277 ** As required by ANSI, LOOP now disallows anonymous collection
2278 clauses such as COLLECT I in conjunction with aggregate boolean
2279 clauses such as THEREIS (= I 1);
2280 ** LOOP now signals an error when any variable is reused in the
2281 same loop (including the potentially useful construct analogous
2282 to WITH A = 1 WITH A = (1+ A);
2283 ** IT is only a special loop symbol within the first clause of a
2284 conditional loop clause;
2285 ** LOOP with a typed iteration variable over a hashtable now
2286 signals a type error iff it should.
2287 * fixed some other bugs revealed by Paul Dietz' test suite:
2288 ** FILE-STREAM now names the class previously known as FD-STREAM;
2289 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2290 argument) no longer signals an error;
2291 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2292 of the slot symbol, rather than using the current package
2293 ((:CONC-NAME "") continues to intern the slot's name in the
2295 * incremented fasl file version number, because of the incompatible
2296 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2297 of the new implementation of DEFINE-COMPILER-MACRO.
2299 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2300 * Support for building SBCL for MIPS platforms running in
2301 little-endian mode has now been checked in, and basic
2302 functionality on said platforms verified.
2303 * minor incompatible change: PCL now records the pathname of a file
2304 in which methods and the like are defined, rather than its
2306 * minor incompatible change: TRUENAME now considers the truename of
2307 a file naming a directory to be the pathname with :DIRECTORY
2308 component indicating that directory.
2309 * minor incompatible change: a NAMED clause in the extended form of
2310 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2311 reason for the previous behaviour is unclear.
2312 * more systematization and improvement of CLOS and MOP conformance
2313 in PCL (thanks to Gerd Moellman and Pierre Mai):
2314 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2316 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2317 primary methods with no specializers;
2318 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2320 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2321 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2322 FORWARD-REFERENCED-CLASSes; error reporting on
2323 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2325 ** SXHASH on CLOS instances now uses a slot internal to the
2326 instance to return different numbers on distinct instances,
2327 while preserving the same return value through invocations of
2329 ** DEFMETHOD signals errors when methods with longer incongruent
2330 lambda lists are added to generic functions;
2331 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2332 CLASS, as specified in AMOP;
2333 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2334 on the order returned by the primary method for classes of
2335 class STANDARD-CLASS;
2336 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2337 * fixed some bugs shown by Paul Dietz' test suite:
2338 ** DOLIST puts its body in TAGBODY;
2339 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2341 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2342 value producing form;
2343 ** if more variables are given to PROGV than values, extra
2344 variables are bound and made to have no value;
2345 ** NSUBSTITUTE on list arguments gets the right answer with
2347 ** ELT signals an error of type TYPE-ERROR when the index argument
2348 is not a valid sequence index;
2349 ** LOOP signals (at macroexpansion time) an error of type
2350 PROGRAM-ERROR when duplicate variable names are found;
2351 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2352 ** FUNCALL of special-operators now cause an error of type
2354 ** PSETQ now works as required in the presence of side-effecting
2355 symbol-macro places;
2356 ** NCONC accepts any object as its last argument;
2357 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2359 ** loop-for-as-package does not require a package to be explicitely
2361 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2363 * fixed bug 166: compiler preserves "there is a way to go"
2364 invariant when deleting code.
2365 * fixed bug 172: macro lambda lists with required arguments after
2366 &REST arguments now cause an error to be signalled. (thanks to
2368 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2369 bug. (thanks to Matthew Danish)
2370 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2372 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2373 arguments are lexically rebound. (thanks to Gerd Moellmann and
2375 * fixed bug 194: error messages are now more informative when there
2376 is no primary method applicable in a call to a generic
2377 function. (thanks to Gerd Moellmann)
2378 * fixed bug in command line argument checking (thanks to Julian
2380 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2381 (thanks to Matthew Danish)
2382 * incremented fasl file version number, because of the
2383 SXHASH-related changes in the layout of CLOS data structures
2385 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2386 * minor incompatible change: The runtime (the Unix executable named
2387 "sbcl") is now much pickier about the .core files it will load.
2388 Essentially it now requires .core files to descend from the same
2389 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2390 as the runtime does. (The intent is to prevent the crashes which
2391 can occur, and which can even be reported as mysterious failures,
2392 when people patch the sources or change the build parameters
2393 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2394 sbcl and .core files.)
2395 * fixed bug: VALUES-LIST is no longer optimized away.
2396 * fixed bug 142: The FFI conversion of C string values to Lisp
2397 string values no longer conses excessively. (thanks to Nathan
2398 Froyd porting Raymond Toy's fix to CMU CL)
2399 * began to systematize and improve MOP conformance in PCL (thanks to
2400 Nathan Froyd, Gerd Moellman and Pierre Mai):
2401 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2403 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2404 ** FINALIZE-INHERITANCE is now called on class finalization;
2405 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2406 argument precedence order.
2407 * fixed bug 202: The compiler no longer fails on functions whose
2408 derived types contradict their declared type.
2409 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2410 so it can be non-toplevel.
2411 * The fasl file version number has changed (because of the new
2412 implementation of DEFMACRO).
2413 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2414 safe code, that any length requirement by their type-specifier
2415 argument is valid. The exceptions to this are described in bug
2417 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2418 functions now check the types of their inputs as required by ANSI.
2419 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2420 introduced symbol is DECLAREd to be SPECIAL.
2421 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2422 * fixed bug: :COUNT argument to sequence functions may be negative
2423 * fixed bug: body of DO-SYMBOLS may contain declarations
2424 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2425 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2426 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2427 symbol macro only once
2428 * fixed printing of call frame when argument list is unavailable
2429 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2430 * compiler no longer signals WARNING on unknown keyword
2433 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2434 * A beta-quality port to the mips architecture running Linux,
2435 based on the old CMUCL backend, has been made. It has been tested
2436 on a big-endian kernel, and works sufficiently well to be able to
2437 rebuild itself; it has not been tested in little-endian mode.
2438 * fixed an inconsistency between gencgc.c and purify.c which made
2439 dumping/loading .core files unreliable
2440 * fixed bug 120a: The compiler now deals correctly with IFs where
2441 the consequent is the same as the alternative, instead of
2442 misderiving the return type. (thanks to Alexey Dejneka)
2443 * fixed bug 113: Logical pathnames are now dumpable (the logical
2444 host is resolved at load-time, throwing an error if it is not
2446 * fixed bug 174: FORMAT's error message is slightly clearer when a
2447 non-printing character is used in a format directive.
2448 * fixed several bugs in compiler checking of type declarations, i.e.
2449 violations of the Python "declarations are assertions" principle
2450 (thanks to Alexey Dejneka)
2451 * fixed several bugs in PCL's error checking (thanks to Gerd
2453 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2455 * fixed bug in compilation of functions as first class values
2456 (thanks to Antonio Martinez-Shotton)
2457 * The compiler's handling TYPE-ERRORs which it can prove will
2458 inevitably happen at runtime has been cleaned up and corrected
2459 in several ways. (thanks to Alexey Dejneka)
2460 * improved argument type checking for various basic arithmetic
2461 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2462 lavished on them in the past that they can be compiled in many
2463 ways in different special cases
2464 * fixed bug 181: compiler checks validity of user supplied type
2466 * cleaned up code flushing in optimization: Function calls which
2467 should signal errors for safety purposes (e.g. which ANSI says
2468 should signal errors when their arguments are of incorrect type)
2469 are no longer optimized away.
2470 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2471 * incremented fasl file version number, because changes in the
2472 implementation of sequence functions like COERCE caused
2473 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2474 in old inline expansions) to become undefined. (Actually these
2475 changes were later undone, so we might very well be binary
2476 compatible with 0.7.7 after all, but leaving the version number
2477 incremented seemed like the simplest and most conservative
2480 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2481 * An alpha-quality port to the parisc architecture running Linux,
2482 based on the old CMUCL backend, has been made. This, even more so
2483 than the other backends, should be considered still a work in
2484 progress; known problems include that the Linux kernel in 64-bit
2485 mode does not propagate the correct sigcontext structure to
2486 userspace, and consequently SBCL on a parisc64 kernel will not
2488 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2489 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2490 suppressed.) Also now that the compiler is looking at declarations
2491 in the environment, it checks optimization declarations as well,
2492 and suppresses inlining when (> DEBUG SPEED).
2493 * More fixes have been made to treatment of floating point exception
2494 treatment and other Unix signals. In particular, floating point
2495 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2496 * The detection and handling of control stack exhaustion (infinite
2497 or very deeply nested recursion) has changed. Stack exhaustion
2498 detection is now done by write-protecting pages at the OS level
2499 and applies at all optimization settings; when found, a
2500 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2501 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2502 be caught using IGNORE-ERRORS.
2503 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2504 correctly when transforming recursive LABELS functions to LETs.
2505 (thanks to Alexey Dejneka)
2506 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2507 that are names of constants or global variables.
2508 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2509 alien routines with docstrings.
2510 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2511 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2513 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2514 (thanks to Eric Marsden)
2515 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2516 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2517 * Bug fix: The SPARC backend can now compile functions involving
2518 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2519 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2520 to a functions expecting a PATHNAME-DESIGNATOR.
2521 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2522 lambda lists. (thanks to Alexey Dejneka)
2523 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2524 functions so that EXPORTing the name of the class doesn't cause
2525 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2526 to Antonio Martinez for reporting this)
2527 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2528 constant-folded without failing an assertion. (thanks to Einar
2529 Floystad Dorum for reporting this)
2530 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2531 types are now dealt with more correctly by the compiler.
2532 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2533 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2534 behaviour analogous to RENAME-FILE. This puts its behaviour more
2535 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2537 * The fasl file version number has changed again. (because of the
2538 bug fix involving the names of PCL MAKE-INSTANCE functions)
2540 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2541 * bug fix: Floating point exceptions are treated much more
2542 consistently on the x86/Linux and PPC/Linux platforms.
2543 * Array initialization with :INITIAL-ELEMENT is now much faster for
2544 cases when the compiler cannot open code the array creation, but
2545 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2546 array accesses have also seen a speed increase.
2547 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2548 specified. (This was at the root of some bad interactions between
2549 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2551 * bug fix: Internal error arguments for undefined functions are now
2552 computed correctly on the PPC/Linux platform.
2553 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2554 Raymond Toy's patch for CMU CL)
2555 * Support for the Solaris 9 operating environment has been included
2556 (thanks to Daniel Merritt)
2557 * A very ugly but hopefully complete draft of the missing FFI chapter
2558 of the manual has been created by reformatting the corresponding
2559 CMU CL manual chapter into (currently very ugly and incoherent)
2560 DocBook and bringing it up to date for SBCL behavior. Thus, the
2561 manual is now essentially complete, at least by my extreme
2562 once-and-only-once standards, whereby it's acceptable to refer to
2563 the doc strings of SB-EXT functions as the primary documentation.
2564 * The fasl file version number has changed again, due to cleanup of
2565 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2566 is no longer a static symbol.)
2568 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2569 * SBCL now builds with OpenMCL (version 0.12) as the
2570 cross-compilation host; also, more progress has been made toward
2571 bootstrapping under CLISP.
2572 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2574 * bug 158 fixed: The compiler can now deal with integer loop
2575 increments different from 1; fixing this turned out also to fix
2577 * bug 169 fixed: no more bogus warnings about using lexical bindings
2578 despite the presence of perfectly good SPECIAL declarations (thanks
2579 to David Lichteblau)
2580 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2581 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2582 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2584 * bug 180 fixed: Method combination specifications no longer ignore
2585 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2586 * bug fix: Structure type predicate functions now check their argument
2587 count as they should.
2588 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2589 correctly. (thanks to Pierre Mai)
2590 * minor incompatible change: The --noprogrammer option is deprecated
2591 in favor of the new --disable-debugger option, which is very similar.
2592 (The major difference is that it takes effect at a slightly different
2593 time at startup, causing handling of errors in --sysinit and
2594 --userinit files will be affected differently.) The
2595 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2596 been added to allow this functionality to be controlled from ordinary
2597 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2598 maintainers, who might want to run non-interactive scripts to
2599 build SBCL cores which will later be used interactively.)
2600 * minor incompatible change: The LOAD function no longer, when given
2601 a wild pathname to load, loads all files matching that pathname.
2602 Instead, an error of type FILE-ERROR is signalled.
2604 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2605 * bug 147 fixed: The compiler preserves its block link/count
2606 invariants more correctly now so that it doesn't crash. (thanks
2608 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2610 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2611 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2612 no-value-for-this-character as NIL (as specified by ANSI).
2613 * HOST-NAMESTRING on physical pathnames now returns a string that is
2614 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2615 (thanks to Christophe Rhodes)
2616 * The Alpha port handles icache flushing more correctly. (thanks to
2618 * More progress has been made toward bootstrapping under CLISP. (thanks
2619 to Christophe Rhodes)
2620 * The fasl file format has changed again, because dynamic loading
2621 on OpenBSD (which has non-ELF object files) motivated some cleanups
2622 in the way that foreign symbols are transformed and passed around.
2623 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2624 is no longer treated as whitespace by the reader, but instead as
2625 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2626 |A
\7fB|, instead of A as it used to.
2628 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2629 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2630 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2631 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2632 this as of 0.7.1.45, but was left out of the previous news section
2633 (thanks to Dan Barlow)
2634 * SBCL now runs on the Solaris operating system on SPARC architectures
2635 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2636 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2637 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2638 * SPARC backend cleanups, allowing builds of cores optimized for V8
2639 and V9 SPARCS, and also emission of code targeted to a particular
2640 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2642 * SBCL is closer to bootstrapping under CLISP, thanks to various
2643 fixes by Christophe Rhodes.
2644 * The fasl file format has changed again, to allow the compiler's
2645 INFO database to support symbol macros.
2646 * The user manual (in doc/) is formatted into HTML more nicely.
2647 (thanks to coreythomas)
2648 * The system is smarter about SUBTYPEP relationships, especially
2649 those involving NOT types (including types such as ATOM which are
2650 represented internally using NOT types). Thus SUBTYPEP is less
2651 likely to return (VALUES NIL NIL) in general, and in particular
2652 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2654 * The fasl file format has changed again, because the internal
2655 representation of types now includes a new slot to support the new
2656 SUBTYPEP-of-NOT-types logic.
2657 * (not a change in the main branch of SBCL, but a related prototype
2658 which can hopefully be merged into the main branch of SBCL in the
2659 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2660 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2661 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2662 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2663 dispatch macro character combinations. (thanks to Alexey Dejneka)
2664 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2665 that superficially look like logical namestrings correctly.
2666 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2668 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2669 default supplied-p) for &optional and &key arguments. (thanks to
2672 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2673 * incompatible change: The compiler is now less aggressive about
2674 tail call optimization, doing it only when (> SPACE DEBUG) or
2675 (> SPEED DEBUG). (This is an incompatible change because there are
2676 programs which relied on the old CMU-CL-style behavior to optimize
2677 away their unbounded recursion which will now die of stack overflow.)
2678 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2679 for non-GENCGC systems has been increased to 20M (since that
2680 seems much closer to the likely performance optimum for modern
2681 systems than the old 4M value was)
2682 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2683 and *DEBUG-PRINT-LEVEL*
2684 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2685 of CMU CL's support for SPARC, and various endianness and other
2686 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2687 * new syntactic sugar for the Unix command line: --load foo.bar is now
2688 an alternate notation for --eval '(load "foo.bar")'.
2690 ** The system now detects stack overflow and handles it gracefully,
2691 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2692 optimization settings. (This is a good thing in general, and
2693 its introduction in this version should be particularly timely
2694 for anyone whose code fails because of suppression of tail
2696 ** The system now hunts for the C variable "environ" in a more
2697 devious way, to avoid segfaults when the C library version differs
2698 between compile time and run time. (thanks to Christophe Rhodes)
2699 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2700 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2701 ** The compiler no longer issues bogus style warnings for undefined
2702 classes in the same source file as the DEFCLASSes which defined
2703 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2705 ** fixes in CONDITION class precedence list for undefined function
2706 errors (thanks to Alexei Dejneka)
2707 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2708 correctly. (thanks to Dan Barlow)
2709 ** portability fixes aiming at bootstrapping under CLISP (thanks
2710 to Dave McDonald and Christophe Rhodes)
2711 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2712 ** fixes in type translation and and type inference (thanks to
2714 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2715 ** various fixes in the new ports (thanks to Dan Barlow)
2716 * several changes related to debugging:
2717 ** suppression of tail recursion, as noted above
2718 ** stack overflow detection, as noted above
2719 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2720 is now the default. (For some time encapsulation has been more
2721 reliable than the breakpoint-based :ENCAPSULATE NIL
2722 implementation, at least on X86 systems; and I just noticed that
2723 encapsulation also seems closer to the spirit of the ANSI
2726 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2728 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2729 up properly again. (There was a packaging bug in 0.7.0 which
2730 left their definitions in SB-SYS::LOAD-FOREIGN and
2731 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2732 vital for most things which interface to C-level interfaces,
2733 like extensions working with sockets or databases or
2734 Perl-compatible regexes or whatever, and the need to fix
2735 this bug is the main reason that 0.7.1 was released so
2737 ** DEFGENERIC is now choosier about the methods it redefines, so that
2738 reLOADing a previously-LOADed file containing DEFGENERICs does
2739 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2740 cycle now works as it should. (thanks to Alexey Dejneka)
2741 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2742 fixed by Christophe Rhodes. (He actually submitted this patch
2743 months ago, and I delayed until after 0.7.0.)
2744 ** Bug 111 (internal compiler confusion about runtime checks on
2745 FUNCTION types) was fixed by Alexey Dejneka.
2746 * Some internal cleanups (getting rid of variables which aren't
2747 needed now that the byte interpreter is gone) caused the fasl
2748 file format number to change again.
2750 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2751 * major incompatible change: The default fasl file extension, i.e. the
2752 default extension for files produced by COMPILE-FILE, has changed
2753 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2755 ** There are many changes in the implementation of the compiler.
2756 SBCL is now essentially a compiler-only implementation of ANSI
2757 Common Lisp. EVAL still "interprets" a few special cases, but
2758 almost all the interesting cases are handled by creating
2759 a LAMBDA expression, calling COMPILE on it, then calling
2760 FUNCALL on the result.
2761 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2762 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2763 Since the code is newer, there might still be some new bugs
2764 (though not as many as before Martin Atzmueller's fixes:-). But
2765 the new code is substantially simpler and clearer, and hopefully
2766 any remaining bugs will be simpler, less fundamental, and more
2767 fixable then the bugs in the old code.
2768 ** The revised compiler is still a little unsteady on its feet.
2770 *** The debugging information it produces (particularly the names
2771 of FUNCTION objects) is sometimes much less useful than what
2772 the old compiler produced.
2773 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2774 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2775 within a MACROLET) has been temporarily weakened.
2776 ** There are new compiler optimizations for various functions:
2777 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2778 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2779 *** the math functions TRUNCATE, FLOOR, and CEILING
2780 *** the function-of-all-trades COERCE
2781 Mostly these should be transparent, but there's one
2782 potentially-annoying problem (bug 117): when the compiler
2783 inline-expands a function and does type analysis on the result,
2784 it can create control paths which have type mismatches, and
2785 when it can't prove that those control paths aren't taken,
2786 it will issue WARNINGs about the type mismatches. This is
2787 a particular problem in practice for the new sequence functions.
2788 It's not clear how this should be fixed, and for now, a
2789 workaround is given in the entry for 117 in the BUGS file.
2790 ** (Because of the interaction between the two previous items --
2791 occasional inlining problems and new inline expansions -- some
2792 of the new sequence function optimizations won't really kick in
2793 completely until debugging information, and then inlining, are
2794 straightened out in some future version.)
2795 * minor incompatible changes:
2796 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2797 DIRECTORY no longer implicitly promotes NIL slots of its
2798 pathname argument to :WILD. In particular, when you ask for the
2799 contents of a directory (which you used to be able to do without
2800 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2801 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2802 ** changes in behavior that ANSI explicitly defines to be
2803 implementation dependent:
2804 *** The new compiler-only implementation still conforms with ANSI,
2805 but acts a little different than before. Besides the obvious
2806 changes in performance tradeoffs (that the cost per form passed
2807 to EVAL has gone up, and the cost per form executed by EVAL
2808 has gone down), the behavior of the system changes a little
2809 because there are no longer any interpreted function objects.
2810 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2811 e.g. doing COMPILE on the output of interactive DEFUN is
2813 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2815 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2816 has changed from (:CL) to NIL.
2817 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2818 e.g. #\Newline and #\Tab, don't have names specified in the
2819 ANSI Common Lisp standard, is now based on their ASCII symbolic
2820 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2821 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2822 are no longer used for output.
2823 ** changes in internal implementation constants:
2824 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2825 4 million. (If your application spends a lot of time GCing and
2826 you have a lot of RAM, you might want to experiment with
2827 increasing it even more.)
2828 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2829 However, almost all old code should still continue to work without
2830 immediate update, as SB-C-CALL is now a (deprecated) nickname
2832 ** Old operator names in the style DEF-FOO are now deprecated in
2833 favor of new corresponding names DEFINE-FOO, for consistency with
2834 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2835 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2836 internal symbols, but a few supported extensions like
2837 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2838 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2839 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2840 etc. as you get deeper into recursive calls to the debugger
2841 command loop, instead of the old "5]", "5]]", "5]]]"
2842 sequence. (I was motivated to do this when squabbles between
2843 ILISP and SBCL left me very deeply nested in the debugger. In the
2844 short term, this change will probably provoke more ILISP/SBCL
2845 squabbles, but hopefully it will be an improvement in the long run.)
2846 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2847 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2848 function, since declaiming return types involving aliens is
2849 (1) annoyingly messy to do by hand and (2) vital to efficient
2850 compilation of code which calls such functions.
2851 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2852 longer reexported by the SB-EXT package. They're solely useful
2853 for alien code, so it seems more logical that you should get
2854 them from the SB-ALIEN package, not in SB-EXT.
2855 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2856 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2857 features. Instead, the code that they used to control is always
2858 built into the system.
2859 * many other bug fixes
2860 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2861 advantage of the new EVAL-WHEN stuff and to clean them up in
2862 general, and they are now more ANSI-compliant in a number of
2863 ways. Martin Atzmueller is responsible for a lot of this.
2864 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2866 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2867 *** correct ERROR type for various file operations
2868 *** some fixes for Lisp streams
2869 *** DEFMETHOD syntax checking
2870 *** changing old weird representation of debug information as
2871 strings (which, among their other deficiencies, don't transform
2872 correctly when you rename packages, and don't change their
2873 print representation when you change things like *PACKAGE*
2874 and *PRINT-LENGTH*) to symbols and lists of symbols
2875 He also made several improvements and fixed several bugs in DESCRIBE.
2876 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2878 *** misbehavior of WRITE-STRING/WRITE-LINE
2879 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2880 and several other LOOP problems as well
2881 *** DIRECTORY when similar filenames are present
2882 *** DEFGENERIC with :METHOD options
2883 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2884 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2885 *** argument ordering in FIND with :TEST option
2886 *** mishandled package designator argument in APROPOS-LIST
2887 *** various problems in the backquote readmacro
2888 *** a bug in APROPOS
2889 *** probably some others that I'm not describing very well here,
2890 since the CVS log documents them by reference to sbcl-devel
2891 messages, and the SourceForge archives aren't working well.:-(
2892 ** Dan Barlow improved the Alpha port (and is making progress on the
2893 PPC port, for those of you who think different).
2894 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2895 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2896 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2897 continues to work on the SPARC port (for those of you in a position
2898 to look down upon our little PC-compatible boxes from a great height).
2899 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2900 and uses that copy, rather than the current dynamic value, when
2901 it's trying to decide whether to truncate output. Thus e.g.
2902 (let ((*print-lines* 50))
2903 (pprint-logical-block (stream nil)
2905 (let ((*print-lines* 8))
2906 (print (aref possiblybigthings i) stream)))))
2907 should now truncate the logical block only at 50 lines, instead of
2908 often truncating it at 8 lines, as it did before.
2909 * The doc/cmucl/ directory, containing old CMU CL documentation
2910 from the time of the fork, is no longer part of the base system.
2911 SourceForge has shut down its anonymous FTP service, and with it
2912 my original plan for distributing the old CMU CL documentation
2913 there. For now, if you need these files you can download an old
2914 SBCL source release and extract them from it.
2915 * The fasl file version number changed again, for dozens of reasons,
2916 some of which are apparent above.
2918 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2919 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2920 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2921 patch, so that DISASSEMBLE output is much nicer.
2922 * The code in the SB-PROFILE package now seems reasonably stable.
2923 I still haven't decided what the final interface should look like
2924 (I'd like PROFILE to interact cleanly with TRACE, since both
2925 facilities use function encapsulation) but if you have a need
2926 for profiling now, you can probably use it successfully with
2927 the current CMU-CL-style interface.
2928 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2929 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2930 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2931 link now returns the dangling link itself, and for similar
2932 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2933 link itself. (In these cases the old code signalled an error and
2934 looped endlessly, respectively.) Thus, DIRECTORY now works even
2935 in the presence of dangling and cyclic symbolic links.
2936 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2937 is now a supported extension again, since the consensus on
2938 sbcl-devel was that it can be useful for ordinary development
2939 work, not just for debugging SBCL itself.
2940 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2941 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2942 that currently-defined functions might be redefined later with
2943 different return types.
2944 * Hash tables can be printed readably, as inspired by CMU CL code
2945 of Eric Marsden and SBCL code of Martin Atzmueller.
2946 * better error handling in CLOS method combination, thanks to
2947 Martin Atzmueller porting Pierre Mai's CMU CL patches
2948 * more overflow fixes for >16Mbyte I/O buffers
2949 * A bug in READ has been fixed, so that now a single Ctrl-D
2950 character suffices to cause end-of-file on character streams.
2951 In particular, now you only need one Ctrl-D at the command
2952 line (not two) to exit SBCL.
2953 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2955 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2956 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2957 does the right thing.
2958 * The implementation of some type tests, especially for CONDITION
2959 types, is now tidier and maybe faster, due to CMU CL code
2960 originally by Douglas Crosher, ported by Martin Atzmueller.
2961 * Some math functions have been fixed, and there are new
2962 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2963 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2964 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2965 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2966 likely to be built into sbcl-0.7.0.)
2967 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2968 CL, ported to SBCL by Martin Atzmueller)
2969 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2970 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2971 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2972 * There's a new slam.sh hack to shorten the edit/compile/debug
2973 cycle for low-level changes to SBCL itself, and a new
2974 :SB-AFTER-XC-CORE target feature to control the generation of
2975 the after-xc.core file needed by slam.sh.
2976 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2977 COMPILE-FILE is no longer supported, so that now every function
2978 gets an entry point, so that block compilation looks a little
2979 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2980 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2981 returns the number of bytes consed since the system started,
2982 rather than the number consed since the first time the function
2983 was called. (The new definition parallels ANSI functions like
2984 CL:GET-INTERNAL-RUN-TIME.)
2985 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2986 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2987 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2988 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2989 * Fasl file version numbers are now independent of the target CPU,
2990 since historically most system changes which required version
2991 number changes have affected all CPUs equally. Similarly,
2992 the byte fasl file version is now equal to the ordinary
2995 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2996 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2997 is no longer recognized. I apologize for this, because it was
2998 listed in SB-EXT as a supported extension, but I found that
2999 its existing behavior was poorly specified, as well as incorrectly
3000 specified, and it looked like too much of a mess to straighten it
3001 out. I have enough on my hands trying to get ANSI stuff to work..
3002 * many patches ported from CMU CL by Martin Atzmueller, with
3003 half a dozen bug fixes in pretty-printing and the debugger, and
3004 half a dozen others elsewhere
3005 * fixed bug 13: Floating point infinities are now supported again.
3006 They might still be a little bit flaky, but thanks to bug reports
3007 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3008 as flaky as they were.
3009 * The --noprogrammer command line option is now supported. (Its
3010 behavior is slightly different in detail from what the old man
3011 page claimed it would do, but it's still appropriate under the
3012 same circumstances that the man page talks about.)
3013 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3014 are now supported, and enabled by default. Thus, the compiler can
3015 handle many floating point and complex operations much less
3016 inefficiently. (Thus e.g. you can implement a complex FFT
3018 * The compiler now detects type mismatches between DECLAIM FTYPE
3019 and DEFUN better, and implements CHECK-TYPE more correctly, and
3020 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3021 thanks to patches from Martin Atzmueller.
3022 * various fixes to make the cross-compiler more portable to
3023 ANSI-conforming-but-different cross-compilation hosts (notably
3024 Lispworks for Windows, following bug reports from Arthur Lemmens)
3025 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3026 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3027 fixed thanks to Pierre Mai's CMU CL patches.
3028 * fixes to keep the system from overflowing internal counters when
3029 it tries to use i/o buffers larger than 16M bytes
3030 * fixed bug 45a: Various internal functions required to support
3031 complex special functions have been merged from CMU CL sources.
3032 (When I was first setting up SBCL, I misunderstood a compile-time
3033 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3034 * improved support for type intersection and union, fixing bug 12
3035 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3036 more obscure bugs as well
3037 * some steps toward byte-compiling non-performance-critical
3038 parts of the system, courtesy of patches from Martin Atzmueller
3039 * Christophe Rhodes has made some debian packages of sbcl at
3040 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3041 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3042 stable, but are nonetheless usable. When he's ready, I'd be happy
3043 to add them to the SourceForge "File Releases" section. (And if
3044 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3045 * new fasl file format version number (because of changes in
3046 internal representation of (OR ..) types to accommodate the new
3047 support for (AND ..) types, among other things)
3049 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3050 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3052 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3054 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3055 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3056 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3057 * DESCRIBE now works on structure objects again.
3058 * Most function call argument type mismatches are now handled as
3059 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3060 know whether the function will be redefined before the call is
3061 executed. (The compiler could flag local calls with full WARNINGs,
3062 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3063 it doesn't keep track of enough information to know whether calls
3064 are local in this sense.)
3065 * Compiler output is now more verbose, with messages truncated
3066 later than before. (There should be some supported way for users
3067 to override the default verbosity, but I haven't decided how to
3068 provide it yet, so this behavior is still controlled by the internal
3069 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3070 src/compiler/ir1util.lisp.)
3071 * Fasl file format version numbers have increased again, because
3072 support for the Gray streams extension changes the layout of the
3073 system's STREAM objects.
3074 * The Gray subclassable streams extension now works, thanks to a
3075 patch from Martin Atzmueller.
3076 * The full LOAD-FOREIGN extension (not just the primitive
3077 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3078 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3079 but like most other programs, it defaults to copying the Unix
3080 environment from the original process instead of starting the
3081 new process in an empty environment.
3082 * Extensions which manipulate the Unix environment now support
3083 an :ENVIRONMENT keyword option which doesn't smash case or
3084 do other bad things. The CMU-CL-style :ENV option is retained
3085 for porting convenience.
3086 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3087 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3089 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3091 * A patch from Martin Atzmueller seems to have solved the SIGINT
3092 problem, and as far as we know, signal-handling now works cleanly.
3093 (If you find any new bugs, please report them!)
3094 * The system no longer defaults Lisp source file names to types
3095 ".l", ".cl", or ".lsp", but only to ".lisp".
3096 * The compiler no longer uses special default file extensions for
3097 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3098 seems to expect a single default extension for all compiled code,
3099 and there's no compelling reason to try to stretch the standard
3100 to allow two different extensions.) Instead, byte-compiled files
3101 default to the same extension as native-compiled files.
3102 * Fasl file format version numbers have increased again, because
3103 a rearrangement of internal implementation packages made some
3104 dumped symbols in old fasl files unreadable in new cores.
3105 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3106 many fewer weird special cases.
3107 * Bug #17 (differing COMPILE-FILE behavior between logical and
3108 physical pathnames) has been fixed, and some related misbehavior too,
3109 thanks to a patch from Martin Atzmueller.
3110 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3111 by Tim Moore, ported to SBCL by Martin Atzmueller.
3112 * Martin Atzmueller fixed several filesystem-related problems,
3113 including bug #36, in part by porting CMU CL patches, which were
3114 written in part by Paul Werkowski.
3115 * More compiler warnings in src/runtime/ are gone, thanks to
3116 more patches from Martin Atzmueller.
3117 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3120 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3122 * DESCRIBE now works on CONDITION objects.
3123 * The debugger now handles errors which arise when trying to print
3124 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3126 * The build system now uses an additional file, customize-target-features.lisp,
3127 to allow local modifications to the target *FEATURES* list. (The point of
3128 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3129 debugging features enabled, without having to worry about propagating it
3130 into everyone's system when I do a "cvs update".) When no
3131 customize-target-features.lisp file exists, the target *FEATURES* list
3132 should be constructed the same way as before.
3133 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3134 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3135 warn about behavior which is undefined under the ANSI spec. Note: This
3136 is specified by ANSI, but it's not very popular with programmers.
3137 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3138 macro in the SBCL sources for an example of a workaround which you
3139 might use to make portable ANSI-standard code which does what you want.
3140 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3141 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3142 when it's used as a non-toplevel form. (This is required in order
3143 to implement the DEFCONSTANT-EQX macro.)
3144 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3145 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3146 funny things, due to bugs in the implementation of EVAL-WHEN
3147 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3148 non-toplevel DEFCONSTANTs are uncommon.)
3149 * The core file version number and fasl file version number have been
3150 incremented, because the old noncompliant DEFCONSTANT behavior involved
3151 calling functions which no longer exist, and because I also took the
3152 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3153 * fixed bug 1 (error handling before read-eval-print loop starts), and
3154 redid debugger restarts and related debugger commands somewhat while
3156 ** The QUIT debugger command is gone, since it did something
3157 rather different than the SB-EXT:QUIT command, and since it never
3158 worked properly outside the main toplevel read/eval/print loop.
3159 Invoking the new TOPLEVEL restart provides the same functionality.
3160 ** The GO debugger command is also gone, since you can just invoke
3161 the CONTINUE restart directly instead.
3162 ** The TOP debugger command is also gone, since it's redundant with the
3163 FRAME 0 command, and since it interfered with abbreviations for the
3165 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3167 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3168 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3169 some porting work by Martin Atzmueller)
3170 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3171 with a boatload of patches which he ported to SBCL. Now that those
3173 ** The system tries to make sure that its low-priority messages
3174 are prefixed by semicolons, to help people who like to use
3175 syntax highlighting in their ILISP buffer. (This patch
3176 was originally due to Raymond Toy.)
3177 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3178 patches originally written by Raymond Toy.
3179 ** The compiler understands coercion between single-value and
3180 multiple-VALUES type expressions better, getting rid of some very
3181 weird behavior, thanks to patches originally by Robert MacLachlan
3182 and Douglas Crosher.
3183 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3184 lambda lists, thanks to a patch originally by Pierre Mai.
3185 ** The system no longer bogusly warns about "abbreviated type
3187 ** The compiler gets less confused by inlining and RETURN-FROM,
3188 thanks to some patches originally by Tim Moore.
3189 ** The system no longer hangs when dumping circular lists to fasl
3190 files, thanks to a patch originally from Douglas Crosher.
3191 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3192 undefined function error.
3193 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3194 in structure slot initforms) for now, documented workaround instead:-|
3195 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3196 * fixed bug 5: added stubs for various Gray stream functions called
3197 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3198 installed, at least appropriate type errors are generated
3199 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3200 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3201 consistently in DEFMETHOD forms.
3202 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3203 it doesn't seem to affect SBCL after all
3204 * The C runtime system now builds with better optimization and many
3205 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3207 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3209 * The system is now under CVS at SourceForge (instead of the
3210 CVS repository on my home machine).
3211 * The new signal handling code has been tweaked to treat register
3212 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3213 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3214 downstream, e.g. in debug-int.lisp, has implicit dependencies
3215 on the unsignedness of integer representation of machine words,
3216 and that caused the system to bomb out with infinite regress
3217 when trying to recover from type errors involving signed values,
3218 e.g. (BUTLAST '(1 2 3) -1).)
3219 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3220 (This was one of the bugs Peter Van Eynde reported back in July.)
3221 * The system now uses code inspired by Colin Walters' O(N)
3222 implementation of MAP (from the cmucl-imp@cons.org mailing
3223 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3224 inline the MAP operation, and there is more than one
3225 sequence argument to the MAP call (so that it can't just
3226 do ETYPECASE once and for all based on the type of the
3227 single sequence argument). (The old non-inline implementation
3228 of the general M-argument sequence-of-length-N case required
3229 O(M*N*N) time when any of the sequence arguments were LISTs.)
3230 * The QUIT :UNIX-CODE keyword argument has been renamed to
3231 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3232 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3234 * Raymond Wiker's patches to port dynamic loading from Linux to
3235 FreeBSD have been added.
3236 * The BUGS file is now more nearly up to date, thanks in large part
3237 to Martin Atzmueller's review of it.
3238 * The debugger now flushes standard output streams before it begins
3239 its output ("debugger invoked" and so forth).
3240 * The core version number and fasl file version number have both
3241 been incremented, because of incompatible changes in the layout
3243 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3244 used to not be called for a saved Lisp image.)
3245 * Martin Atzmueller's version of a patch to fix a compiler crash,
3246 as posted on sbcl-devel 13 September 2000, has been installed.
3247 * Instead of installing Martin Atzmueller's patch for the
3248 compiler transform for SUBSEQ, I deleted the compiler transform,
3249 and transforms for some similar consing operations.
3250 * A bug in signal handling which kept TRACE from working on OpenBSD
3252 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3253 away by constant folding
3254 * The system now defines its address space constants in one place
3255 (in the Lisp sources), and propagates them automatically elsewhere
3256 (through GENESIS and the sbcl.h file). Therefore, patching the
3257 address map is less unnecessarily tedious and error-prone. The
3258 Lisp names of address space constants have also been systematized.
3259 * CVS tags like dollar-Header-dollar have been removed from
3260 the sources, because they have never saved me trouble and
3261 they've been source of trouble working with patches and other
3262 diff-related operations.
3263 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3264 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3266 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3268 * The system has been ported to OpenBSD.
3269 * The system now compiles with a simple "sh make.sh" on the systems
3270 that it's supported on. I.e., now you no longer need to tweak
3271 text in the target-features.lisp-expr and symlinks in src/runtime/
3272 by hand, the make.sh takes care of it for you.
3273 * The system is no longer so grossly inefficient when compiling code
3274 involving vectors implemented as general (not simple) vectors (VECTOR T),
3275 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3276 which dares to use the various sequence functions on non-simple
3277 vectors, takes less of a performance hit.
3278 * There is now a primitive type predicate VECTOR-T-P
3279 to test for the (VECTOR T) type, so that e.g.
3280 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3281 can now be compiled with some semblance of efficiency. (The old code
3282 turned the type declaration into a full call to %TYPEP at runtime!)
3283 * AREF on (VECTOR T) is still not fast, since it's still compiled
3284 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3285 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3286 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3287 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3288 without making full calls to SUBTYPEP at runtime.
3289 (Some analogous efficiency issues for non-simple vectors specialized to
3290 element types other than T, or for non-simple multidimensional arrays,
3291 have not been addressed. They could almost certainly be handled the
3292 same way if anyone is motivated to do so.)
3293 * The changes in array handling break binary compatibility, so
3294 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3295 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3296 instead of (VALUES T T).
3297 * By following the instructions that Dan Barlow posted to sbcl-devel
3298 on 2 July 2000, I was able to enable primitive dynamic object
3299 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3300 functionality is not implemented (since it calls ld to resolve
3301 library references automatically, requiring RUN-PROGRAM for its
3302 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3303 try to resolve library references) is now supported.
3304 * The system now flushes the standard output streams when it terminates,
3305 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3306 them at several other probably-convenient times, e.g. in each pass of
3307 the toplevel read-eval-print loop, and after evaluating a form given
3308 as an "--eval" command-line option. (These changes were motivated by a
3309 discussion of stream flushing issues on cmucl-imp in August 2000.)
3310 * The source transform for TYPEP of array types no longer assumes
3311 that an array whose element type is a not-yet-defined type
3312 is implemented as an array of T, but instead punts, so that the
3313 type will be interpreted at runtime.
3314 * There is now some support for cross-compiling in make.sh: each of
3315 the phases of make.sh has its own script. (This should be transparent
3316 to people doing ordinary, non-cross-compile builds.)
3317 * Since my laptop doesn't have hundreds of megabytes of memory like
3318 my desktop machine, I became more motivated to do some items on
3319 my to-do list in order to reduce the size of the system a little:
3320 ** Arrange for various needed-only-at-cold-init things to be
3321 uninterned after cold init. To support this, those things have
3322 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3323 symbols with such names are now uninterned after cold init).
3324 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3325 fasl files for cold load.
3326 ** Remove the old compiler structure pooling code (which used to
3327 be conditional on the target feature :SB-ALLOC) completely.
3328 ** Redo the representation of some data in cold init to be more compact.
3329 (I also looked into supporting byte compiled code at bootstrap time,
3330 which would probably reduce the size of the system a lot, but that
3331 looked too complicated, so I punted for now.)
3332 * The maximum signal nesting depth in the src/runtime/ support code has
3333 been reduced from 4096 to 256. (I don't know any reason for the very
3334 large old value. If the new smaller value turns out to break something,
3335 I'll probably just bump it back up.)
3336 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3338 * Many, many bugs reported by Peter Van Eynde have been added to
3339 the BUGS list; some have even been fixed.
3340 * While enabling dynamic object file loading, I tried to make the
3341 code easier to understand, renaming various functions and variables
3342 with less ambiguous names, and changing some function calling
3343 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3344 * While trying to figure out how to do the OpenBSD port, I tried to
3345 clean up some of the code in src/runtime/. In particular, I dropped
3346 support for non-POSIX signal handling, added various comments,
3347 tweaked the code to reduce the number of compilation warnings, and
3348 renamed some files to increase consistency.
3349 * To support the new automatic configuration functionality in make.sh,
3350 the source file target-features.lisp-expr has been replaced with the
3351 source file base-target-features.lisp-expr and the machine-generated
3352 file local-target-features.lisp-expr.
3353 * fixed a stupid quoting error in make.sh so that using CMU CL
3354 "lisp -batch" as cross-compilation host works again
3356 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3358 * DESCRIBE no longer tries to call itself recursively to describe
3359 bound/fbound values, so that it no longer fails on symbols which are
3360 bound to themselves (like keywords, T, and NIL).
3361 * DESCRIBE now works on generic functions.
3362 * The printer now prints less-screwed-up representations of closures
3363 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3364 * A private symbol is used instead of the :EMPTY keyword previously
3365 used to mark empty slots in hash tables. Thus
3366 (DEFVAR *HT* (MAKE-HASH-TABLE))
3367 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3368 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3369 now does what ANSI says that it should. (You can still get
3370 similar noncompliant behavior if bang on the hash table
3371 implementation with all the symbols you get back from
3372 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3373 This breaks binary compatibility, since tests for equality to
3374 :EMPTY are wired into things like the macroexpansion of
3375 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3377 * There's now a minimal placeholder implementation for CL:STEP,
3378 as required by ANSI.
3379 * An obscure bug in the interaction of the normal compiler, the byte
3380 compiler, inlining, and structure predicates has been patched
3381 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3382 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3384 * Missing ordinary arguments in a macro call are now detected even
3385 when the macro lambda list contains &KEY or &REST.
3386 * The debugger no longer complains about encountering the top of the
3387 stack when you type "FRAME 0" to explicitly instruct it to go to
3388 the top of the stack. And it now prints the frame you request even
3389 if it's the current frame (instead of saying "You are here.").
3390 * As specified by ANSI, the system now always prints keywords
3391 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3393 * The default initial SIZE of HASH-TABLEs is now smaller.
3394 * Type information from CLOS class dispatch is now propagated
3395 into DEFMETHOD bodies, so that e.g.
3396 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3398 is now basically equivalent to
3399 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3400 (DECLARE (TYPE SINGLE-FLOAT X))
3402 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3403 operation, without having to do run-time type dispatch.
3404 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3405 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3406 * Since I don't seem to be making big file reorganizations very often
3407 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3408 is overflowing my ability to conveniently back them up), I've finally
3409 checked the system into CVS. (The CVS repository is on my home system,
3410 not at SourceForge -- putting it on SourceForge might come later.)
3411 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3412 high-level GC-NOTIFY-FOO functions send their output. (There's
3413 still very little control of where low-level verbose GC functions
3414 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3415 less than it used to -- the GC-NOTIFY-FOO functions are now under
3416 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3417 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3418 in only one place in the source code, and propagates it automatically
3419 everywhere that it's needed. Thus e.g. when I bump the version from
3420 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3421 * The C source files now include boilerplate legalese and documentation
3422 at the head of each file (just as the Lisp source files already did).
3423 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3424 to his page will be replaced with a link to his new CLiki service.
3426 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3428 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3429 * The build process now looks for GNU make under the default name "gmake",
3430 instead of "make" as it used to. If GNU make is not available as "gmake"
3431 on your system, you can change this default behavior by setting the
3432 GNUMAKE environment variable.
3433 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3434 can build without error under CMU CL.
3436 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3438 * There is now a partial SBCL user manual (with some new text and some
3439 text cribbed from the CMU CL manual).
3440 * The beginnings of a profiler have been added (starting with the
3441 CMU CL profiler and simplifying and cleaning up). Eventually the
3442 main interface should be through the TRACE macro, but for now,
3443 it's still accessed through vaguely CMU-CL-style functions and macros
3444 exported from the package SB-PROFILE.
3445 * Some problems left over from porting CMU CL to the new
3446 cross-compilation bootstrap process have been cleaned up:
3447 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3448 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3449 part when getting rid of the compiler *BACKEND* stuff.)
3450 ** The value of *NULL-TYPE* was screwed up, because it was
3451 being initialized before the type system knew the final
3452 definition of the 'NULL type. This screwed up several key
3453 optimizations in the compiler, causing inefficiency in all sorts
3454 of places. (I found it because I wanted to understand why
3455 GET-INTERNAL-RUN-TIME was consing.)
3456 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3457 PROCLAIM FTYPE information. Unfortunately this broke binary
3458 compatibility again, since now the forms output by DEFGENERIC
3459 to refer to functions which didn't exist in 0.6.3.
3460 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3461 can use the new (as of 0.6.3) transform for SOME into MAP into
3463 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3464 values to (INTEGER 0 1000000), so that the time code will no longer
3465 occasionally get blown up by Linux returning 1000000 microseconds
3466 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3467 its output conform to the ANSI spec. (Alas, this makes its output
3468 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3469 * A full call to MAP NIL with a single sequence argument no longer conses.
3470 * fixes to problems pointed out by Martin Atzmueller:
3471 * The manual page no longer talks about multiprocessing as though
3472 it were currently supported.
3473 * The ILISP support patches have been removed from the distribution,
3474 because as of version 5.10.1, ILISP now supports SBCL without us
3475 having to maintain patches.
3476 * added a modified version of Raymond Toy's recent CMU CL patch for
3477 EQUALP comparison of HASH-TABLE
3479 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3481 * The system still can't cross-compile itself with
3482 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3483 entails), but at least it can compile more of itself that way
3484 than it used to be able to, and various buglets which were uncovered
3485 by trying to cross-compile itself that way have now been fixed.
3486 * This release breaks binary compatibility again. This time
3487 at least I've incremented the FASL file format version to 2, so that the
3488 problem can be detected reliably instead of just causing weird errors.
3489 * various new style warnings:
3490 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3491 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3492 ** using the lexical binding of a variable named in the *FOO* style
3493 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3494 as specified by ANSI.
3495 * *RANDOM-STATE* is no longer automatically initialized from
3496 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3497 default behavior of the system is to repeat its behavior every time
3498 it's run. If you'd like to change this behavior, you can always
3499 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3500 old convention there was no comparably easy way to get the system to
3501 repeat its behavior every time it was run.
3502 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3503 FTYPE declarations has been removed, in favor of their ANSI
3504 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3505 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3506 the types of their sequence arguments can be determined at compile time.
3507 This is done through a new open code expansion for MAP which eliminates
3508 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3509 argument types can be determined at compile time.
3510 * The optimizer now transforms COERCE into an identity operation when it
3511 can prove that the coerced object is already of the correct type. (This
3512 can be a win for machine generated code, including the output of other
3513 optimization transforms, such as the MAP transform above.)
3514 * Credit information has been moved from source file headers into CREDITS.
3515 * Source file headers have been made more standard.
3516 * The CASE macro now compiles without complaining even when it has
3519 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3521 * (Note that the way that the PCL macroexpansions were rewritten
3522 to accommodate the change in DEFGENERIC below breaks binary
3523 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3524 not run under sbcl-0.6.2. Once we get out of alpha releases,
3525 i.e. hit release 1.0.0, we'll probably try to maintain binary
3526 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3527 and sbcl-1.4.4. Until then, however, it might be fairly common
3528 for maintenance releases to break binary compatibility.)
3529 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3531 * The reporting of unhandled signals has been changed to print some
3532 explanatory text as well as the report form. (Previously only
3533 the report form was printed.)
3534 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3535 it defines, so that the compiler no longer issues undefined function
3536 warnings for compiled-but-not-yet-loaded generic functions.
3537 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3538 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3539 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3540 section of the ANSI spec are supported.
3541 * The "" nickname for the "KEYWORD" package has been removed.
3542 The reader still handles symbol tokens which begin with a package marker
3543 as keywords, but it doesn't expose its mechanism for doing so in the
3544 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3545 * The system now issues STYLE-WARNINGs for contradictory TYPE
3546 proclamations. (Warnings for contradictory FTYPE proclamations would
3547 be nice too, but those can't be done usefully unless the type system
3548 is made smarter about FUNCTION types.)
3549 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3550 systematized, so that "*target-*.lisp is supposed to exist only on the
3551 target and imply that there's a related file which exists on the
3552 host, and *host-*.lisp is supposed to exist only on the host and imply
3553 that there's a related file which exists on the target. This involves a
3554 lot of renaming. Hopefully the acute confusion caused by the renaming
3555 will be justified by the reduction in chronic confusion..
3556 ** runtime-type.lisp -> early-target-type.lisp
3557 ** target-type.lisp -> late-target-type.lisp
3558 ** early-host-format.lisp -> early-format.lisp
3559 ** late-host-format.lisp -> late-format.lisp
3560 ** host-error.lisp -> misc-error.lisp
3561 ** early-error.lisp -> early-target-error.lisp
3562 ** late-error.lisp -> late-target-error.lisp
3563 ** host-defboot.lisp -> early-defboot.lisp
3564 ** code/misc.lisp -> code/target-misc.lisp
3565 ** code/host-misc.lisp -> code/misc.lisp
3566 ** code/numbers.lisp -> code/target-numbers.lisp
3567 ** code/early-numbers.lisp -> numbers.lisp
3568 ** early-host-type.lisp -> early-type.lisp
3569 ** late-host-type.lisp -> late-type.lisp
3570 ** host-typep.lisp -> typep.lisp
3571 ** load.lisp -> target-load.lisp
3572 ** host-load.lisp -> load.lisp
3573 ** host-disassem.lisp -> disassem.lisp
3574 ** host-insts.lisp -> insts.lisp
3575 ** byte-comp.lisp -> target-byte-comp.lisp
3576 ** host-byte-comp.lisp -> byte-comp.lisp
3577 ** host-signal.lisp -> signal.lisp
3578 ** host-defstruct.lisp -> defstruct.lisp
3579 ** late-target-type.lisp -> deftypes-for-target.lisp
3580 Furthermore, several other previously target-only files foo.lisp (e.g.
3581 hash-table.lisp and random.lisp) have been split into a target-and-host
3582 foo.lisp file and a target-only target-foo.lisp file, with their key type
3583 definitions in the target-and-host part, so that the cross-compiler will
3584 know more about target types.
3585 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3586 gone away. In their place are various *BACKEND-FOO* variables
3587 corresponding to the slots of the old structure.
3588 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3589 into the target SBCL has been fixed.
3590 * The chill.lisp system for loading cold code into a running SBCL
3592 * Support for the CMU CL "scavenger hook" extension has been removed.
3593 (It was undocumented and unused in the CMU CL sources that SBCL was
3594 derived from, and stale in sbcl-0.6.1.)
3595 * Various errors in the cross-compiler type system were detected
3596 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3597 (enabling various consistency checks). Many of them were fixed,
3598 but some hard problems remain, so the compiler is back to
3599 running without *TYPE-SYSTEM-INITIALIZED* for now.
3600 * As part of the cross-compiler type system cleanup, I implemented
3601 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3602 * I have started adding UNCROSS calls throughout the type system
3603 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3604 on cross-compiler input files will be able to go away, and various
3606 * CONSTANTP now returns true for quoted forms (as explicitly required
3609 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3611 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3612 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3613 (MAKE-LIST -1). In the longer term, it ought to become true
3614 that declarations are assertions even at SAFETY 1. For now, it's not
3615 quite true even at SAFETY 3, but it's at least more nearly true..
3616 (Note that this change seems to increases the size of the system by
3617 O(5%) and to decrease the speed of the compiler by 20% or more.)
3618 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3619 for the problem of printing dozens of lines of distracting information
3620 about low-level system machinery as part of the top stack frame
3621 on entry to the debugger when an undefined function was called.
3622 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3623 is not reset to COMMON-LISP-USER.
3624 * Compilation of stuff related to dyncount.lisp has been made conditional
3625 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3626 smaller. The various dyncount-related symbols have been moved into
3627 a new "SB-DYNCOUNT" package.
3628 * tty-inspect.lisp has been renamed to inspect.lisp.
3629 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3630 feature has gone away. (When we eventually port to other flavors of
3631 libc and/or Unix, we'll try to make the differences between flavors
3632 invisible at the user level.)
3633 * Various other *FEATURES* tags, and/or their associated conditionals,
3634 have been removed if obsolescent, or given better documentation, or
3635 sometimes given more-mnemonic names.
3637 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3639 * tidied up "make.sh" script
3640 * tidied up system directory structure
3641 * better "clean.sh" behavior
3642 * added doc/FOR-CMUCL-DEVELOPERS
3643 * many many small tweaks to output format, e.g. removing possibly-confusing
3644 trailing #\. character in DESCRIBE-INSTANCE
3645 * (EQUALP #\A 'A) no longer signals an error.
3646 * new hashing code, including EQUALP hashing
3647 * tidied up Lisp initialization and toplevel
3648 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3649 * command line argument processing
3650 * added POSIX-GETENV function to deal with Unix-ish environment variables
3651 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3652 terminating SBCL on EOF
3653 * non-verbose GC by default
3654 * There is no more "sbcl" shell script; the sbcl file is now the C
3655 runtime executable (just like CMU CL).
3656 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3658 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3659 symbol and package information as Lisp native symbol and package objects
3660 instead of strings naming symbols and strings naming packages. This way,
3661 whenever packages are renamed (as in warm init), debug information is
3662 transformed along with everything else.
3663 * tweaked the optimization policy declarations which control the building
3664 of SBCL itself. Now, among other things, the system no longer saves
3665 source location debugging information. (This helps two problems at once
3666 by reducing SBCL size and by keeping SBCL from trying to look for its
3667 sources -- which may not exist -- when reporting errors.)
3668 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3669 debugging and testing purposes
3670 * cleaned up printing, making the printer call PRINT-OBJECT for
3671 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3672 methods, giving nearly-ANSI behavior
3673 * converted almost all special variables to use *FOO* naming convention
3674 * deleted PARSE-TIME functionality, since it can be done portably
3675 * moved some files out of cold init into warm init
3676 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3678 * regularized formatting of source files
3679 * added an install.sh script
3680 * fixed ridiculous memory usage of cross-compiler by making
3681 compiler/alloc.lisp not try to do pooling unless it can hook
3682 itself into the GC of the cross-compilation host. Now the system
3683 builds nicely on my old laptop.
3684 * added :SB-ALLOC in target-features.lisp-expr
3685 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3686 was not implemented)
3687 * re-did condition handling and note reporting in the compiler. Notes
3688 are no longer handled by signalling conditions. Style warnings
3689 and warnings are handled more correctly and reported in such a way
3690 that it's easy to find one or the other in your output (so that you
3691 can e.g. figure out which of many problems caused COMPILE-FILE to
3693 * changed the severity of several compiler warnings from full WARNING
3694 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3695 compiler note reporting so that it doesn't use the condition system
3696 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3697 COMPILE-FILE command)
3698 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3699 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3700 no longer supported, since their implementation was deeply intertwingled
3701 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3702 * removed bogus "support" for compiler macros named (SETF FOO), and
3703 removed the compiler macro for SETF INFO (but only after making a fool
3704 of myself on the cmucl-imp mailing list by posting a bogus patch for
3705 DEFINE-COMPILER-MACRO..)
3706 * Compiled files containing forms which have side effects on the Lisp
3707 reader (such as DEFPACKAGE forms) are now handled more correctly.
3708 (Compiler queuing of top level lambdas has been suppressed by setting
3709 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3710 * deleted various currently-unused source files, e.g. gengc.lisp. They
3711 may be added back at some point e.g. when porting to other architectures,
3712 but until they are it's distracting to distribute them and to try to
3714 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3715 there were so many of them they're just distractions, and UNCROSS is
3716 known to be able to handle the current sources
3717 * moved PROFILE functionality into TRACE, so that it will be clear
3718 how the wrapping and unwrapping of functions when you profile them
3719 interacts with the wrapping and unwrapping of functions when you
3720 trace them. (Actually, the functionality isn't there yet, but at least
3721 the interface specification is there. Hopefully, the functionality will
3722 arrive with some future maintenance release.)
3723 * removed host-oops.lisp
3724 * changed signature of QUIT function to allow UNIX-CODE argument
3725 * fixed READ-SEQUENCE bug
3726 * tweaked verbose GC output so that it looks more like the progress
3727 output that ANSI specifies for functions like LOAD
3728 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3729 * added <http://sbcl.sourceforge.com> to the banner information printed by