1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
3 * fixed some bugs revealed by Paul Dietz' test suite:
4 ** REMOVE-METHOD returns its generic function argument even when
5 the method is not one of the generic functions' methods.
6 ** objects declared as MEMBER types can be admissible arguments to
8 ** printing characters should simply be printed by the FORMAT ~:C
11 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
12 * new platform: experimental support for SBCL x86/Darwin, including
13 MacOS X 10.4.5 on Intel.
14 * new feature: Unicode character names are now known to the system
15 (through CHAR-NAME and NAME-CHAR).
16 * new feature: the filesystem location of SBCL's core file is
17 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
18 * minor incompatible change: the contrib modules SB-POSIX and
19 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
20 this change is to make it easier to distribute
21 location-independent binaries.
22 * bug fix: as implied by AMOP, standardized classes no longer have
23 slots named by external symbols of public packages. (reported by
25 * optimization: calling functions via constant symbols -- as in
26 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
27 via the function object as in (FUNCALL #'FOO).
28 * optimization: on x86 and x86-64, the calling convention for the
29 case of dealing with an unknown number of values has been altered
30 to be friendlier to the prediction heuristics implemented,
31 particularly with reference to CALL and RET pairing. (thanks to
33 * optimization: on x86, the code for access to thread-local symbol
34 values and binding/unbinding of thread-local symbols is smaller.
35 (thanks to Alastair Bridgewater)
36 * enhancement: CONSTANTP is now able to determine constantness of
37 more complex forms, including calls to constant-foldable standardized
38 functions and some special forms beyond QUOTE.
39 * fixed bug: occasional GC crashes on Solaris/x86
40 * optimization: x86-64 supports stack allocation of results of simple
41 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
42 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
43 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
44 (thanks to James Knight)
45 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
46 by the standard (thanks to Stephen Compall)
48 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
49 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
50 be used for bundling the runtime and the core file into one
51 executable binary. This feature is not currently supported on all SBCL
52 platforms. (thanks to James Bielman and NIIMI Satoshi)
53 * new feature: a generational or ephemeral garbage collector is now
54 the default on the PowerPC platform (both Linux and Darwin). The
55 old Cheney (stop and copy) collector is a build-time option.
56 (thanks to Cyrus Harmon, after Raymond Toy)
57 * minor incompatible change: the method by which SBCL finds its
58 contributed modules has changed; it no longer relies on symbolic
59 links from an $SBCL_HOME/systems directory, but searches directly
60 in the subdirectories of $SBCL_HOME.
61 * enhancement: the dynamic heap size on the Linux/PPC platform is
62 markedly larger, even using the older Cheney garbage collector.
63 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
64 certain complicated string types. (reported by Gary King)
65 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
66 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
67 * fixed bug: DOCUMENTATION on structures no longer signals an
68 error if no structure type of the right name exists. (reported by
70 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
71 applicable non-standard primary or :AROUND method on
72 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
73 no longer cause unbound variable errors. (reported by Kevin Reid)
74 * fixed bug: metacircle resolution in cases where methods have slots
75 added before the slots from STANDARD-METHOD. (reported by Jean
77 * fixed bug: the Power PC assembly code for calling into lisp
78 sometimes computed its return address wrongly (depending on the
79 15th bit of the address, fixed at link time). This used to
80 manifest itself in a segmentation violation while building PCL.
81 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
82 many others over the years)
83 * fixed bug: ROOM no longer reports silly numbers for stack usage if
84 the stack pointer (treated as a signed integer) is negative.
85 (thanks to Peter van Eynde)
87 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
88 * new platform: experimental support for the Windows operating
89 system has been added. (thanks to Alastair Bridgewater)
90 * fixed several bugs in and robustified the PPC FFI (including
91 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
92 * bug fix: allow non-simple string symbol names (reported by Paul
94 * bug fix: interrupt handling on NetBSD (thanks to Richard M
96 * bug fix: saving a core corrupted callbacks on x86/x86-64
97 * bug fix: closed a loophole in metacircularity detection and
98 grounding in the PCL implementation of CLOS.
99 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
100 (reported by tomppa on #lisp)
101 * bug fix: clear the direction flag on Lisp -> C transitions, as
102 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
103 (reported by Andrej Grozin and Hendrik Maryns)
104 * optimization: major improvements to GC efficiency on GENCGC platforms
105 * optimization: faster implementation of EQUAL
106 * optimization: emit more efficient opcodes for some common
107 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
108 * optimization: several other minor code-generation improvements on x86-64
109 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
111 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
112 * minor incompatible change: (SETF CLASS-NAME) and (SETF
113 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
114 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
115 by ANSI as a generic function, and for consistency (SETF
116 GENERIC-FUNCTION-NAME) is treated likewise.
117 * fixed bug #233.b: make constraint propagation notice when a variable
118 value is changed after it is referenced but before it is used
119 * fixed bug #296: no more arbitrary behaviour from filenames with
120 odd characters as --load arguments. (reported by Adam Warner)
121 * fixed bug #390: :CHARACTER-SET pathname components now work as
122 expected. (reported by Tim Daly Jr)
123 * fixed bug #391: complicated :TYPE intersections in slot
124 definitions no longer cause an error in PCL internals.
125 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
126 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
127 returning the number of octets which would be written to the
128 file-stream. (thanks to Robert J. Macomber)
129 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
130 arguments to the internal stream functions. (thanks to David
132 * bug fix: stack exhaustion is now handled on the x86/SunOS
133 platform. (thanks to Daisuke Homma)
134 * bug fix: more accurate ROOM results on GENCGC platforms
135 * optimization: improved type inference for arithmetic-for
136 index variables in LOOP
137 * optimization: faster floating-point SQRT on x86-64
138 * fixed some bugs revealed by Paul Dietz' test suite:
139 ** DOCUMENTATION returns NIL instead of "" for method combinations
140 that don't have a docstring
142 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
143 * minor incompatible change: (SETF CLASS-NAME) and (SETF
144 GENERIC-FUNCTION-NAME) are no longer generic functions, and
145 therefore are not customizeable by user code (as seems to be at
146 least permitted and maybe required by AMOP). As a consolation,
147 however, the SBCL implementation of these functions now calls
148 REINITIALIZE-INSTANCE as specified by AMOP.
149 * bug fix: slot-definition documentation is propagated as per ANSI
150 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
151 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
152 argument, as do eslotd initialization methods. (from Pascal
153 Costanza's "Closer" project)
154 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
155 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
157 * bug fix: it is now possible to have more than one subclass of
158 STANDARD-GENERIC-FUNCTION without causing stack overflow.
159 (reported by Bruno Haible, Pascal Costanza and others)
160 * bug fix: the dependent update protocol now works for generic
161 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
163 * bug fix: condition-class instances corresponding to
164 DEFINE-CONDITION forms are now created eagerly. (reported by
165 Kalle Olavi Niemitalo on comp.lang.lisp)
166 * bug fix: floating point printing is more accurate in some
167 circumstances. (thanks to Simon Alexander)
168 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
169 merged with *DEFAULT-PATHNAME-DEFAULTS*.
170 * bug fix: callbacks on OS X now preserve stack-alignment.
171 (thanks to Cyrus Harmon)
172 * enhancement: the x86-64 disassembler is much better at
173 disassembling SSE instructions. (thanks to Lutz Euler)
174 * enhancement: CHECK-TYPE on a variable now allows type-inference
175 (thanks to Vincent Arkesteijn)
176 * optimization: improved performance of EUC-JP external format.
177 (thanks to NIIMI Satoshi)
178 * optimization: performance improvements to IO on file streams of
179 :ELEMENT-TYPE CHARACTER
180 * optimization: much faster memory allocation on x86-64
181 * optimization: faster garbage collections (latency and throughput)
183 * optimization: faster bignum subtraction and division on x86 and x86-64
184 (thanks to Lutz Euler)
185 * optimization: more accurate type-inference for loops that use a
186 floating point index variable or a negative step.
188 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
189 * bug fix: add a workaround to SBCL looping infinitely at startup on
190 Linux kernels with apparently buggy implementations of personality().
191 (thanks to Svein Ove Aas)
192 * bug fix: Unicode symbols are correctly printed in LDB backtraces
193 (thanks to David Lichteblau)
194 * bug fix: local bindings shadow global symbol macros for
195 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
196 * optimization: non-open coded uses of numeric comparison operators
197 (e.g. >) no longer cons when called with more than one parameter
198 on platforms supporting dynamic-extent allocation.
199 * enhancement: saving cores with foreign code loaded is now supported
200 on MIPS/Linux in addition to the previously supported platforms.
201 * bug fix: FIND-RESTART now tests for activity, not applicability when given
202 a restart object as identifier. (reported by Helmut Eller for CMUCL)
203 * bug fix: division by zero in sb-sprof when no samples were collected
204 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
205 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
206 (thanks to Alexander Kjeldaas)
207 * new feature: ignoring the return values of destructive functions
208 like NREVERSE signals a compile-time style-warning.
209 (thanks to Kevin Reid)
210 * new feature: the alignment of alien structure fields can be
211 explicitly specified. (thanks to Cyrus Harmon)
212 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
213 and dump core on SIGQUIT
215 ** incompatible change: threads do not inherit values of specials
216 from their parents (see manual)
217 ** bug fix: threads stacks belonging to dead threads are freed by the
218 next exiting thread, no need to gc to collect thread stacks anymore
219 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
220 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
221 ** bug fix: dynamic variable and thread start related gc lossage
222 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
224 * fixed some bugs revealed by Paul Dietz' test suite:
225 ** SUBTYPEP is slightly more accurate on heinously complicated
226 CONS types where some of the members have uncertain (in the
227 NIL, NIL sense) type relationships to each other.
229 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
230 no longer purified unless :PURIFY T is explicitly specified.
231 ** Non-purified cores are significantly smaller than before
233 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
234 * new feature: timers based on Zach Beane's excellent timer package
235 * added support for the following external formats: koi8-u,
236 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
237 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
238 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
239 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
240 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
241 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
242 * incompatible change: a threaded SBCL will no longer revert to
243 non-threaded mode on non-NPTL systems, but refuse to start entirely.
244 * bug fix: interrupts are disabled until startup is complete; no
245 more sigsegvs when receiving a signal to soon
246 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
248 * bug fix: add a workaround for the memory randomization features in
249 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
250 workaround will only be in effect on systems with the proc filesystem
252 * bug fix: printing objects of type HASH-TABLE signals a
253 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
255 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
256 have been running for over 50 days. (reported by Gilbert Baumann)
257 * bug fix: the logic for getting names of functions gets less
258 confused when confronded with alternate-metaclass
259 funcallable-instances. (reported by Cyrus Harmon)
260 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
261 consistent, even on internal alternate-metaclass objects.
262 * bug fix: flush closure information collected by physical
263 environment analysis prepass before the main pass. (bug reported
265 * bug fix: compiler pack phase does not modify a hashtable, which is
266 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
267 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
268 may be expanded. (reported by Alexander <malishev> on c.l.l)
269 * bug fix: no more style-warnings when compiling code using LAST
270 under high-SPEED low-SPACE settings. (reported by David Wallin
272 * fixed some incompatibilities between SBCL's MOP and the MOP
274 ** the METAOBJECT class is now implemented;
275 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
276 STANDARD-OBJECT, as required;
277 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
278 now compatible (as required by VALIDATE-SUPERCLASS); there
279 remains a consistency requirement over the presence or absence
280 of the FUNCTION class in the superclass list of finalized
281 classes; see the manual for more details;
282 ** the :AROUND method for COMPUTE-SLOTS on
283 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
284 requested slot ordering.
286 ** bug fix: parent thread now can be gc'ed even with a live
288 ** bug fix: binding a special with PROGV to no value is not
289 the same as that symbol not having been bound (thanks to
291 ** bug fix: binding specials is thread safe (thanks to
293 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
294 and SB-SPROF do not die with 'no handler for signal XX in
295 interrupt_handle_now(..)' anymore
296 ** bug fix: WITH-TIMEOUT works with multiple running threads
297 * fixed some bugs revealed by Paul Dietz' test suite:
298 ** ENSURE-GENERIC-FUNCTION should take a method class object for
299 the :method-class keyword argument.
301 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
302 * new port: the Solaris operating system on x86 processors is now
303 mostly supported, though some rough edges in the environment
304 remain. (thanks to Daisuke Homma)
305 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
306 space for its heap. (thanks to Thiemo Seufer)
307 * optimization: SBCL on MIPS platforms now supports dynamic-extent
308 closures. (thanks to Thiemo Seufer)
309 * minor incompatible change: eof selects abort in the debugger.
310 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
311 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
313 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
314 deprecated, and will go away in a future revision of SBCL.
315 * minor incompatible change: GC-ON and GC-OFF are no longer
316 implemented with a counter, it does not matter how many times gc
317 is switched on or off
318 * bug fix: discriminating functions for generic function classes
319 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
320 make invalid assumptions about method precedence order. (reported
322 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
323 platforms. (thanks to Thiemo Seufer)
324 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
325 * bug fix: correct pathname printing with printer escaping is on.
326 (thanks to Kevin Reid)
327 * bug fix: complex VOP definitions in "user-space" no longer trigger
328 package locks. (reported by Zach Beane)
329 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
330 for funcallable-instance functions, and is no different from
332 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
333 characters in the current readtable. (reported by Nicholas Neuss)
334 * bug fix: SUBTYPEP on various CONS types returns more a more
335 accurate acknowledgment of its certainty.
336 * optimizations: REMOVE-DUPLICATES now runs in linear time on
337 lists in some cases. This partially fixes bug 384.
338 * flush all standard streams before prompting in the REPL and the
340 * bug fix: signal handling and triggering gc do not conflict
341 directly anymore, in particular a high frequency sb-sprof does
342 not prevent gc from running
343 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
344 approximation for timezone and DST information between the
345 universal time epoch and the smallest negative 32-bit time_t.
346 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
347 year 1899 so long as the encoded time is non-negative.
348 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
349 the correct answer for non-constant rotate amounts when performing
350 an inline 32-bit rotation.
352 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
353 there is only one thread in the session
354 ** bug fix: memory leak for streams created in one thread and
355 written to in another
356 ** bug fix: lockup when compiled with gcc4
357 ** bug fix: race that allows the gc to be triggered when gc is
359 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
360 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
362 ** bug fix: finished threads are gc'ed properly
363 * fixed some bugs revealed by Paul Dietz' test suite:
364 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
365 has arguments with a different set of applicable methods from
366 the orignal arguments.
367 ** The type error thrown by MAP now has a correct expected-type
369 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
370 name a compiled function.
371 ** Three MISC tests where a large bignum was improperly coerced to
372 a float that couldn't represent that bignum during type
373 derivation were fixed.
374 ** SUBTYPEP can now handle types involving the negation of a
375 list-form FUNCTION type.
376 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
377 as (COMPLEX (AND RATIO (NOT FIXNUM))).
378 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
380 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
381 * New feature: Experimental support for bivalent streams: streams
382 opened with :element-type :default now allow character and binary
383 (unsigned-byte 8) I/O
384 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
385 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
387 * bug fix: no more highly sporadic "couldn't check whether ~S is
388 readable" when reading a stream and an interrupt hits in the middle
389 of a select system call
390 * compiler better recognizes complex arrays (reported by Eduardo
392 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
393 for variables on SBCL built with linkage-tables. (reported by Luis
395 * various error reporting improvements.
396 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
397 (Thanks to Thiemo Seufer)
398 * bug fix: nested reader invokations work correctly
399 * bug fix: it is possible to have simultaneous references to foreign
400 code and foreign data with the same name.
402 ** added x86-64 support
403 ** incompatible change: the threading api now works with thread
404 objects instead of thread ids
405 ** bug fix: threads are protected from signals and interruption when
406 starting up or going down
407 ** bug fix: a race where an exiting thread could lose its stack to gc
408 ** bug fix: don't halt on infinite error in threads if possible
409 ** fixed numerous gc deadlocks introduced in the pthread merge
410 ** bug fix: fixed thread safety issues in read and print
411 ** bug fix: debugger doesn't hang on session lock if interrupted at
412 an inappropriate moment
413 ** bug fix: run-program is now thread safe(r)
414 ** bug fix: inner with-recursive-lock no longer releases the mutex
415 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
416 * fixed some bugs revealed by Paul Dietz' test suite:
417 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
418 EXPECTED-TYPE slots filled.
419 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
421 ** MAKE-CONDITION accepts classes as type-designators.
422 ** COMPILE may never return NIL.
423 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
424 range before calling Unix time functions
426 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
427 * numerous signal handling fixes to increase stability
428 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
429 * minor incompatible change: we now correctly canonize default
430 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
431 as per the MOP, rather than the historical (INITARG INITFUNCTION
432 INITFORM). (reported by Bruno Haible)
433 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
434 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
435 for more information.
436 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
437 pathname is a directory pathname.
438 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
439 * dynamic space size on PPC has been increased to 768Mb. (thanks to
441 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
442 :METACLASS argument in addition to a class name. (reported by
443 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
444 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
445 (reported by Bruno Haible; patch from Gerd Moellmann)
446 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
448 * bug fix: more cleanups to the floating point exception handling on
449 x86-64 (thanks to James Knight)
450 * bug fix: the compiler does not try to constant fold calls of
451 COERCE and BIT-* functions when they return freshly constructed
452 objects. (reported by Nikodemus Siivola and Paul Dietz)
453 * optimization: improved the MIPS versions of generic (in the
454 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
455 * optimization: direct conversion of (unsigned-byte 32) to floats on
456 the PowerPC platform.
457 * optimization: structure instances with raw slots now use less
458 memory, and probably show better memory locality. (thanks to
460 * optimization: DYNAMIC-EXTENT declarations for lists and closures
461 are treated as requests for stack allocation on the x86-64,
462 Alpha-32, PPC and SPARC platforms.
463 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
464 the wrong file descriptor; implementation of SOCKET-OPEN-P.
465 (thanks to Tony Martinez)
467 ** gcing a dead thread can no longer lead to lockups
468 ** threads block signals until they are set up properly
469 ** errno is no longer shared by threads
470 ** interrupt-thread restores the eflags register on x86
471 ** fixed some lockups due to gc/thread interaction
472 * fixed some bugs revealed by Paul Dietz' test suite:
473 ** invalid dotted lists no longer raise a read error when
475 ** signal an error if a symbol that names a declaration is used
476 as the name of a type, or vice versa
477 ** allow using the (declare (typespec var*)) abbreviation for
478 (declare (type typespec var*)) with all type specifiers
479 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
480 ** always return NIL from PPRINT-POP when OBJECT is NIL
481 ** don't signal errors when pretty-printing malformed LABELS,
482 FLET or MACROLET forms
483 ** declarations in a DOLIST body are also in scope for the
485 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
487 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
490 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
491 * fixed cross-compiler leakages that prevented building a 32-bit
492 target with a 64-bit host compiler.
493 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
494 opened with :IF-EXISTS :OVERWRITE.
495 * fixed bug 281: error for an invalid qualifier in a short-form method
496 combination method is not signalled until the faulty method is called.
497 * bug fix: iteration variable type inferrer failed to deal with open
498 intervals. (reported by Alan Shields)
499 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
500 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
501 or more runtime options were provided to the sbcl binary.
502 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
503 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
504 * optimization: there's now a fast-path for fixnum arguments in the
505 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
506 * optimization: the code generated on x86-64 is more compact thanks
507 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
508 * bug fix: floating-point exception handling now partly works on
509 x86-64. (Thanks to James Knight)
510 * improvement to the MIPS backend from Thiemo Seufer: C-style
511 64-bit long long arguments and return values to alien functions
513 * fixed some bugs revealed by Paul Dietz' test suite:
514 ** the type-error signalled from WARN has a filled-in DATUM slot.
515 ** the type-error required when a stream is not associated with
516 a file has the stream as its datum.
517 ** type-errors on single-floats on x86-64 no longer have
518 :INVALID-OBJECT as the datum
519 ** the type-errors signalled for invalid function names now have
520 a correct expected type
521 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
522 for structures defined with a :TYPE. Documentation strings for
523 typed structures are no longer immediately discarded
524 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
525 broadcast streams and synonym streams. FILE-LENGTH now also works
526 on broadcast streams.
528 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
529 * incompatible change: the --noprogrammer option, deprecated since
530 version 0.7.5, has been removed. Please use the equivalent
531 --disable-debugger option instead.
532 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
534 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
535 inoperational for a while now) has been completely removed.
536 * null lexical environments are now printed as #<NULL-LEXENV>,
537 significantly reducing the amount of clutter in typical backtraces.
538 * documentation on weak pointers, finalization, and after GC hooks
539 has been added to the manual.
540 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
541 as other specialized array types, is much faster. SUBSEQ and
542 COPY-SEQ on such arrays have also been sped up.
543 * optimization: EQL is now more efficient when at least other argument
544 is known to be of type (OR FIXNUM (NOT NUMBER)).
545 * fixed bug: the runtime is now less vulnerable to changes in the
546 size of the SBCL object on OS X, and virtual memory is reserved for
547 all spaces that need to be at a fixed address.
548 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
549 * fixed bug: finalizers and after GC hooks that cause consing are now
551 * fixed bug: compiler error messages and summaries are now printed to
552 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
553 * fixed inference of the upper bound of an iteration variable.
554 (reported by Rajat Datta).
555 * fixed bug 211e: calling local functions with duplicated constant
556 keyword argument no longer causes a bogus style warning about an
558 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
559 ftype declaration to be disregarded. (reported by Dave Roberts)
560 * fixed bug 373: caused by erronous compilation of references to alien
561 variables in the runtime on ppc/darwin.
562 * fixed bug 376: CONJUGATE type deriver.
563 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
564 (reported by Baughn on #lisp)
565 * a cross-compiler bug on non-x86 platforms has been identified and
566 fixed. (thanks to Bruno Haible)
567 * improvements to the MIPS runtime code for increased stability.
568 (thanks to Thiemo Seufer)
569 * increased the maximimum compact environment size to allow
570 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
571 * partial workaround for bug 135: don't name the function we're
572 calling for hairy cases of EVAL, so as not to accumulate one environment
573 entry per call to EVAL. (reported by Kevin Reid)
574 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
575 * optimization: type testing for non-vector arrays should be faster.
576 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
578 * bug fix: setting 31st element of a bit vector to zero did not work
580 * bug fix: redefining a class definition which failed due to a
581 previous accessor / function clash now works (but see BUGS entry
582 #380 for more problems in this area). (thanks to Zach Beane)
583 * the long form of DEFINE-METHOD-COMBINATION disables method group
584 checking when given a single method group with pattern *. (thanks
586 * on x86 compiler supports stack allocation of results of simple
587 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
588 * support for building and running on Mac OS X 10.4 "Tiger" has been added
589 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
590 * fixed some bugs related to Unicode integration:
591 ** the restarts for recovering from input and output encoding
592 errors only appear when there is in fact such an error to
594 * fixed some bugs revealed by Paul Dietz' test suite:
595 ** MISC.549 and similar: late transformation of unsafe type
596 assertions into derived types caused unexpected code
598 ** SCALE-FLOAT type deriver is less wrong.
599 ** type derivers for EXP, LOG and similar functions compute result
600 types for complex arguments better.
601 ** (MISC.563) CONJUGATE type deriver works for very restricted
603 ** out-of-line type testers for character strings are available.
604 ** EQUAL compiler transform understands specialness of objects
606 ** accessing double-floats stored on the stack now works on x86-64.
607 ** debugger internals could sometimes create invalid lispobjs,
608 resulting in GC crashes.
609 ** MISC.548: type check weakening can convert required type into
611 ** initialization forms for bindings are not in scope of free special
614 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
615 * incompatible change: thread support for non-NPTL systems has
616 been removed - locking is buggy and unreliable. A threaded
617 SBCL build will now warn at startup and refuse to create
618 new threads, unless futex support is detected in the kernel
619 * incompatible change: the top level REPL now has only an ABORT
620 restart associated with it, not TOPLEVEL and ABORT as it used to.
621 TOP and TOPLEVEL are now available as debugger commands for
622 returning to the top level.
623 * incompatible change: forms evaluated in the REPL now use the
624 global optimization policy.
625 * incompatible change: user- and system-initialization files are
626 no longer processed with LOAD, but by READ and EVAL; hence the
627 global optimization policy, startup package, readtable, etc,
629 * The .fasl file format number has been incremented because of
630 various incompatible changes.
631 * internal entry point details and argument counts no longer appear
632 in backtraces unless explicitly requested by setting
633 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
634 * built-in and standard functions no longer have names like "top
635 level local call to FOO".
636 * fixed bug 32: functions defined in non-null lexical environments
637 now have more legible printed representation
638 * fixed bug 33: functions defined in non-null lexical environemnts
639 are now more amenable to inspection by INSPECT.
640 * workaround for bug 354: XEPs no longer appear in backtraces unless
641 explicitly requested.
642 * fixed bug: receiving the signal which results from attempting to
643 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
644 SIGBUS on other *BSDs) on architectures where the C stack is also
645 the Lisp stack (x86 and x86-64) from foreign code no longer leads
646 to debugger confusion or wild instability. (reported by Cheuksan
648 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
649 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
650 (reported by Lutz Euler)
651 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
652 compiled code. (reported by Paul Dietz)
653 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
654 specialized parameter to a method to something that is not TYPEP
655 the specializer is now possible.
656 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
657 face of package deletion.
658 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
659 pathname. (reported by David Tolpin on c.l.l)
660 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
661 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
662 STRING-STREAM and FILE-STREAM, respectively.
663 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
664 than previously; in addition, its test suite is now run on build.
665 (thanks to Luke Gorrie)
666 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
667 * optimization: added a immediate representation for single-floats
669 * fixed some bugs revealed by Paul Dietz' test suite:
670 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
671 ** attempting to create a package with a colliding nickname causes
672 correctable errors to be signalled.
673 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
674 ** several x86-64 backend bugs related to sign-extension of immediate
677 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
678 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
679 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
681 * fixed loading of multiply forward-referenced layouts.
682 (thanks to Cheuksan Wang)
683 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
684 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
685 (see COMPILE-FILE documentation for details of :PRINT option)
686 * fixed bugs 19 and 317: fixed-format floating point printing is
687 more accurate. This also fixes a bug reported by Adam Warner
688 related to the ~@F format directive.
689 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
691 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
692 dispatch table if the from-char is a dispatch macro character.
693 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
694 by the wordsize no longer ignore the last word. (reported by Lutz
696 * fixed bug in type checking of dynamic-extent variables. (reported
698 * optimization: sequence traversal functions use their freedom to
699 coerce function designators to functions.
700 * optimization: code with many calls to CLOS methods specialized on
701 CLOS classes has had redundant type checks removed.
702 * optimization: added declarations to speed up operations that access
703 the internal character database (for example STRING-UPCASE)
704 * optimization: comparison operations between floats and sufficiently small
705 fixnums no longer create extra rationals
706 * fixed some bugs related to Unicode integration:
707 ** portions of multibyte characters at the end of buffers for
708 character-based file input are correctly transferred to the
709 start of the buffer at the next read.
710 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
711 passing it through to OPEN.
712 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
713 argument given to internal calls to OPEN.
714 ** the built-in comment readers (introduced by character sequences
715 ";" and "#|") are more forgiving to encoding errors; they will
716 STYLE-WARN and then attempt to resync the stream at a character
717 boundary. (thanks to Teemu Kalvas)
718 * fixed some bugs in the x86-64 port:
719 ** Negative short int return values from c-calls are sign-extended
721 ** The stack is aligned to 16-bytes for c-calls, as required by
722 the ABI. (thanks to Cheuksan Wang)
723 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
724 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
725 for enums. (thanks to Vincent Arkesteijn)
726 ** Multiple small optimizations and bugfixes for floating point
728 * fixed some bugs revealed by Paul Dietz' test suite:
729 ** Space, Tab, Linefeed, Return and Page have the invalid
730 secondary constituent character trait.
731 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
733 ** WITH-INPUT-FROM-STRING should only update the index place on
735 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
737 ** Bit-array operations (BIT-AND and similar) worked incorrectly
738 with one-dimensional arrays with fill pointers.
739 ** TYPE-OF failed on a complex with an integer realpart and a
741 ** compiler failure during type inference for the code of form
742 (IF (EQL X (THE ...)) ...) (MISC.535).
744 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
745 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
746 platform. The port supports 61-bit fixnums, large memory spaces
747 and reloading shared object files.
748 * enhancement: saving cores with foreign code loaded is now
749 supported on ppc/Darwin in addition to the previously supported
751 * enhancement: the statistical profiler now walks deeper into the
752 call stack for more meaningful call-graphs and accrued time
753 reports (x86/x86-64 only). It also now reports time spent in
755 * enhancement: it is now possible to trace most individual methods
756 of a generic function in addition to tracing the generic function
758 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
760 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
761 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
762 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
763 produces an error. (thanks to Vincent Arkesteijn)
764 * bug fix: NAMESTRING on pathnames with :WILD components in their
765 directories works correctly. (thanks to Artem V. Andreev)
766 * fixed bug 125: compiler preserves identity of closures. (reported
768 * bug fixed: functions with &REST arguments sometimes failed with
769 "Undefined function" when compiled with (DEBUG 3). (reported by
771 * bug fix: overflow during compiling of setting element of a bit
772 vector with constant index and value. (reported by Timmy Douglas)
773 * build fix: fixed the dependence on *LOAD-PATHNAME* and
774 *COMPILE-FILE-PATHNAME* being absolute pathnames.
775 * on x86 compiler partially supports stack allocation of dynamic-extent
777 * GO and RETURN-FROM do not check the extent of their exit points
778 when compiled with SAFETY 0.
779 * fixed some bugs related to Unicode integration:
780 ** encoding and decoding errors are now much more robustly
781 handled; it should now be possible to recover even from invalid
782 input or output to the terminal. (thanks to Teemu Kalvas)
783 ** provided a first cut at implementing STRING-TO-OCTETS and
784 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
785 ** altered the SB-MD5 contributed module slightly, changing the
786 interface just enough for it to be supportable for builds where
787 lisp characters are not eight bits.
788 * fixed some bugs revealed by Paul Dietz' test suite:
789 ** the FORMATTER-generated functions for ~V[ conditionals require
790 the correct number of arguments.
791 ** READ-FROM-STRING returns the mandated second value when applied
792 to displaced strings.
793 ** the #\Rubout and #\Backspace characters are treated as invalid
794 constituent characters by the tokenizer.
796 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
797 * new feature: reloading changed shared object files with
798 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
799 * new feature: references to foreign variables and functions
800 can now be compiled and loaded before the corresponding shared
801 object file is loaded, as long as the foreign definitions are
802 available at runtime.
803 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
804 supported. (thanks to Dan Debertin)
805 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
806 just keywords. (thanks to Vincent Arkesteijn)
807 * fixed bug #331: structure-class instances corresponding to
808 DEFSTRUCT forms are now created eagerly.
809 * fixed bug #345: backtraces from calls to undefined functions work
810 on x86 as well. Related bug #61 is now also partially fixed on x86
811 (backtraces from throws to unknown catch tags.)
812 * bug fix: lambda-list parsing is now stricter vrt. order and number
813 of lambda-list keywords.
814 * bug fix: as specified by AMOP, an error is signalled if a
815 class-option appears multiple times in a DEFCLASS form. (reported
817 * bug fix: RANDOM can be compiled when the compiler derives the type
818 of its numeric argument as a disjoint set of small integers.
819 (reported by Paul Dietz)
820 * bug fix: starting a core saved with shared objects loaded when
821 those objects are not available no longer causes threaded SBCL to
822 hang. (reported by Sean Ross)
823 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
825 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
826 argument, which must always be NIL. (reported by Kalle Niemitalo)
827 * bug fix: printing 1.0d+23 no longer results in an error.
828 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
829 * bug fix: structure slot setters preserve evaluation order. (thanks
831 * fixed some bugs related to Unicode integration:
832 ** RUN-PROGRAM can allow its child to take input from a Lisp
833 stream. (reported by Stefan Scholl)
834 ** ASDF-INSTALL successfully downloads .tar.gz files in all
835 locales. (reported by Ken Causey)
836 * fixed some bugs revealed by Paul Dietz' test suite:
837 ** INCF, DECF and REMF evaluate their place form as specified in
839 ** FORMATTER expands ~{ ~} iteration directives with V or #
840 parameters correctly.
841 ** FORMATTER deals with the ~@[ ~] conditional directive where the
842 consequent uses no arguments correctly.
843 ** the system has a partial understanding of the (COMPLEX RATIO)
845 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
848 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
849 * new feature: a build-time option (controlled by the :SB-UNICODE
850 keyword feature, enabled by default) for building the system with
851 support for the entire 21-bit character space defined by the
853 * new feature: the system now has rudimentary external-format
854 support; the primary user-visible change at this time is that
855 characters with the high bit set (such as lower-case-e-acute) will
856 print correctly to a terminal in a UTF-8 environment.
857 * minor incompatible change: BASE-CHAR no longer names a class;
858 however, CHARACTER continues to do so, as required by ANSI.
859 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
860 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
861 has been moved to the SB-EXT package (temporarily re-exported from
863 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
864 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
866 * fixed bug #351: better error-handlind and reporting for malformed
868 * fixed bug #350: bignum-printing is now more memory-efficient,
869 allowing printing of very large bignums, eg. (expt 2 10000000).
870 (reported by Bruno Haible)
871 * fixed bug #302: better primitive-type selection for intersection
873 * fixed bug #308: non-graphic characters now all have names, as
874 required. (reported by Bruno Haible)
875 * bug fix: redefining a class with different superclasses now correctly
876 removes it from the direct-subclasses of its previous superclasses.
877 (reported by David Morse)
878 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
879 new value now works. (reported by Bruno Haible)
880 * bug fix: correct canonicalization of multiple non-standard slot
881 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
882 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
883 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
884 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
885 now exists, an signals an error.
886 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
887 during macroexpansion for non-iteration variables as well. (reported
888 by Bruno Haible for CMUCL)
889 * bug fix: Cyclic structures and unprintable objects in compiler
890 messages no longer cause errors. (reported by Bruno Haible)
891 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
892 and READ-FROM-STRING all now return a primary value of NIL if
893 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
894 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
895 (reported by Bruno Haible for CMUCL)
896 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
897 do not designate a condition. (reported by Bruno Haible for
899 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
900 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
901 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
902 permitting resolution in favour of any of the conflicting symbols.
903 (reported by Bruno Haible for CMUCL)
904 * bug fix: EQUAL compiler optimizations is less aggressive on
905 strings which can potentially compare true despite having distinct
906 specialized array element types.
907 * bug fix: unit enumerations can be defined without dividing by
908 zero. (thanks to Vincent Arkesteijn)
909 * FORMAT compile-time argument count checking has been enhanced.
910 (report from Bruno Haible for CMUCL)
911 * a partial workaround for the bug 262: the compiler does not try to
912 inline-expand a local function doing RETURN-FROM from a deleted
913 BLOCK. (thanks to Peter Denno for the bug report and to David
914 Wragg for the simple test case)
915 * fixed some bugs revealed by Paul Dietz' test suite:
916 ** RENAME-PACKAGE allows all package designators as new package
918 ** constraint propagation and lambda variable substitution are
919 more cautious in dealing with partially deleted code.
920 ** compiler sometimes lost reoptimization passes.
921 ** CERROR, given a condition as condition designator, passes its
922 remaining arguments to the continue format control without
924 ** Case-altering FORMAT directives work correctly on non-ASCII
926 ** The REAL type specifier handles bounds outside the floating
927 point ranges without signalling FLOATING-POINT-OVERFLOW.
928 ** Functions with IR1-transformations can create intercomponent
929 references to global functions.
930 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
932 ** FORMAT ~R treats a nil value for its first parameter correctly.
934 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
935 * enhancement: saving cores with foreign code loaded is now
936 supported on x86/NetBSD and sparc/Linux in addition to the previously
938 * bug fix: on some platforms repeated installations caused multiple
939 copies of HTML documentation to be installed -- should not happen
940 any more. (reported by Stefan Scholl)
941 * bug fix: parsing self-recursive alien record types multiple times
942 no longer causes infinite recursion. (reported by Thomas F. Burdick,
943 original patch by Helmut Eller for CMUCL)
944 * bug fix: stack-exhaustion detection works now on NetBSD as well.
945 (thanks to Richard Kreuter)
946 * bug fix: defining classes whose accessors are methods on existing
947 generic functions in other (locked) packages no longer signals
948 bogus package lock violations. (reported by François-René Rideau)
949 * bug fix: special variables as DEFMETHOD parameters no longer have
950 associated bogus type declarations. (reported by David Wragg and
952 * bug fix: read-write consistency on streams of element-type
953 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
954 * bug fix: redefiniton of the only method of a generic function with
955 no DEFGENERIC no longer emits a full WARNING. In addition,
956 redefinition of generic functions with no DEFGENERIC to an
957 incompatible lambda list now signals an error. (thanks to Zach
959 * bug fix: DEFGENERIC now works even when there's a function of the
960 same name in an enclosing lexical environment. (thanks to Zach
962 * fixed compiler failure, caused by instrumenting code during
963 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
964 * optimization: added loop analysis and improved register allocation
965 * fixed some bugs revealed by Paul Dietz' test suite:
966 ** POSITION on displaced vectors with non-zero displacement
967 returns the right answer.
968 ** (SIMPLE-STRING) is a valid type specifier for sequence
970 ** *PRINT-LEVEL* handling for slotless structures is pedantically
972 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
973 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
975 ** The justification version of the FORMAT ~< directive treats
976 non-zero minpad parameter correctly.
978 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
979 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
980 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
981 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
982 the supported interface.
983 * new feature: Single-stepping of code compiled with DEBUG 2 or
984 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
985 * new feature: saving cores with foreign code loaded is now
986 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
987 Timothy Moore's work for CMUCL)
988 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
989 parameters to * instead of NIL if no initform is supplied.
990 (reported by Johan Bockgaard)
991 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
992 correctly when FUNCALL appears as the car of the form. Note:
993 despite this FUNCALL forms are not currently subject to
994 compiler-macro expansion. (port of Raymond Toy's fix for the
995 same from CMUCL, reported by Johan Bockgaard)
996 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
997 (thanks for Teemu Kalvas)
998 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
999 case instead of terminating the iteration (thanks for Julian
1000 Squires, Sean Champ and Raymond Toy)
1001 * bug fix: incorrect expansion of defgeneric that caused a style
1002 warning. (thanks for Zach Beane)
1003 * bug fix: slot accessor effective method computation works properly
1004 for classes with multiple non-standard applicable methods on
1005 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1006 * on x86 compiler supports stack allocation of results of LIST and
1007 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1008 CMUCL implementation by Gerd Moellmann)
1009 * fixed some bugs revealed by Paul Dietz' test suite:
1010 ** FORMAT strings with both the ~<~:;~> form of the justification
1011 directive and pretty-printing directives cause an error.
1013 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1014 * incompatible change: the internal functions
1015 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1016 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1017 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1018 instead of the old functions.
1019 * new feature: on platforms where "dladdr" is available foreign
1020 function names now appear in backtraces. (based on Helmut Eller's
1022 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1023 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1025 * documentation: ASDF manual is now distributed with SBCL.
1026 * bug fix: SBCL can now load its contributed modules with REQUIRE
1027 even if the system-provided entries have been removed from
1028 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1030 * bug fix: backtraces involving undefined functions or assembly
1031 routines are more informative. (thanks to Brian Downing)
1032 * bug fix: mutually referent alien structures now work correctly.
1033 (reported by Rick Taube)
1034 * bug fix: structures defined by WITH-ALIEN can be referred to
1035 within other definitions in the same WITH-ALIEN.
1036 * bug fix: division operators (MOD, TRUNCATE and the like) with
1037 constant zero divisors and integer dividends no longer generate
1039 * bug fix: provide default methods for INPUT-STREAM-P and
1040 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1041 * bug fix: improve the detection and resolution of MOP metacycles
1042 (where effective method computation depends on the generic function
1043 having its effective method computed). (reported by Bruno Haible)
1044 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1045 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1046 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1047 * bug fix: pathnames with relative directory components can be
1048 represented relative to default pathnames.
1049 * optimization: in taking the GCD of bignums, reduce the two bignums
1050 to approximately the same size (using Euclid's algorithm) before
1051 applying the more sophisticated binary GCD. (thanks to Juho
1053 * optimization: COUNT on bitvectors now operates word-at-a-time.
1054 * optimization: ASH with a positive, but not necessarily constant,
1055 (leftwards) shift, when in a modular context, is compiled to a
1057 * fixed some bugs revealed by Paul Dietz' test suite:
1058 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1059 if the corresponding argument is NIL.
1060 ** FORMAT directives accepting a minpad parameter treat negative
1062 ** literal commas as character parameters to FORMAT directives are
1064 ** literal spaces directly after ~<Newline> directives within a
1065 format-logical-block (~:< ~@:>) do not induce :FILL-style
1066 conditional newlines.
1067 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1068 as specified: it no longer includes conditional newlines).
1069 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1071 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1072 :PER-LINE-PREFIX argument does not evaluate to a string.
1074 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1075 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1076 the manual for details; for now, package locks can be disabled by
1077 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1078 if no major problems are found then it is likely that they will be
1079 compiled in unconditionally.
1080 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1081 unsupported operators on all platforms. To load a shared library
1082 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1083 object file, link it into a shared library outside of SBCL and
1084 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1086 * bug workaround: ROOM T and the GC were not working together
1087 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1088 reporting facility aren't preserved. That reporting has been
1089 disabled, so now until and unless someone figures out how to make it
1090 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1091 * minor incompatible change: as threatened around sbcl-0.8.0, the
1092 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1093 The symbols are also exported from SB-PCL for backwards
1094 compatibility, but more so than before SB-PCL should be treated as
1095 an implementation-internal package.
1096 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1098 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1099 It is known as of this release not to work on the Alpha, however.
1100 * fixed bug #167: errors signalled due to illegal syntax in method
1101 bodies are now more legible.
1102 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1103 designators and can hence be used with TYPEP.
1104 * fixed bug #333: CHECK-TYPE now ensures that the type error
1105 signalled, if any, has the right object to be accessed by
1106 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1107 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1108 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1110 * fixed bug #334: programmatic addition of slots using specialized
1111 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1112 and :ALLOCATION :CLASS slots.
1113 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1114 float exponents. (rereported by Peter Seibel)
1115 * fixed a bug: #\Space (and other whitespace characters) are no
1116 longer considered to be macro characters in standard syntax by
1117 GET-MACRO-CHARACTER.
1118 * fixed bug: initialization of condition class metaobjects no longer
1119 causes an instance of the condition to be created. (reported by
1121 * fixed bug: it is now possible to have slots such that
1122 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1123 description is neither :INSTANCE nor :CLASS.
1124 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1125 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1126 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1127 non-standard SLOT-DEFINITION-ALLOCATION values.
1128 * fixed bugs: various race conditions handling exiting threads.
1129 CL-PPCRE's thread torture test now passes.
1130 * fixed arguably-a-bug: GC time proportional to number of threads in
1131 system even when most of them are idle
1132 * optimization: improved performance of BIT and SBIT on bit-vectors.
1133 * .fasl file incompatibility: The fasl file version number has
1134 been incremented because of changes associated with package locks.
1136 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1137 * minor incompatible change: the system no longer provides
1138 optimization hints (conditions which are TYPEP
1139 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1140 mode; these hints are still emitted when higher SPEED optimization
1142 * new contrib module: a sampling profiler (profiling by statistical
1143 sampling, rather than by instrumenting functions) is available as
1144 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1145 * the behaviour of the standard function ED is now customizeable by
1146 third parties through a hook variable: see ED's documentation
1147 string for information on the protocol.
1148 * the compiler no longer emits efficiency notes for (FUNCALL X)
1149 when the type of X is uncertain under default optimization
1151 * fixed bug 276: mutating a binding of a specialized parameter to a
1152 method to something that is not TYPEP the specializer is now
1154 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1155 exhibit the required behaviour on the x86 platform. (thanks to
1156 Peter van Eynde, Eric Marsden and Bruno Haible)
1157 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1158 even for difficult arguments. (reported by Peter Graves)
1159 * fixed bug 141a: the backquote printer now descends quoted
1161 * fixed another bug in backquote printing: no more destructive
1162 modification of the form's list structure. (reported by Brian
1164 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1165 move between its address being taken and the call to
1166 interrupt_thread, fixing a crashing race condition.
1167 * the SB-POSIX contrib implementation has been adjusted so that it
1168 no longer exhibits ridiculously poor performance when constructing
1169 instances corresponding to C structs.
1171 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1172 * minor incompatible change: the sb-grovel contrib now treats C
1173 structures as alien (in the sense of SB-ALIEN) objects rather than
1174 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1175 has implications for memory management of client code
1176 (sb-grovel-returned objects must now be manually managed) and for
1177 type safety (alien objects now have full types).
1178 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1179 used to control emission of compiler diagnostics, rather than the
1180 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1181 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1182 quality should be considered deprecated.
1183 * install.sh now installs the user manual as well
1184 * (not quite a new documentable feature, but worth considering in
1185 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1186 beginnings of a semantically meaningful condition hierarchy is
1187 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1189 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1190 designator as the defaults argument.
1191 * fixed bug: Displaced arrays whose displaced-to array has become
1192 too small now cause ARRAY-DIMENSION to signal an error, providing
1193 for safer bounds-checking. (reported by Bruno Haible)
1194 * fixed bug: DEFCLASS slot definitions with identical :READER and
1195 :WRITER names now signal a reasonable error. (reported by Thomas
1197 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1199 * fixed bug: Invalid slot specification errors now print correctly.
1200 (thanks to Zach Beane)
1201 * fixed bug 320: Shared to local slot value transfers in class
1202 redefinitions now happen corrently from superclasses as
1203 well. (reported by Bruno Haible)
1204 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1206 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1207 improper lists correctly. (reported by Bruno Haible)
1208 * fixed bug 313: source-transform for <fun-name> was erroneously
1209 applied to a call of a value of a variable with name <fun-name>.
1210 (reported by Antonio Menezes Leitao)
1211 * fixed bug 307: The obsolete instance protocol ensures that
1212 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1213 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1214 remove dead unknown-values globs from the middle of the stack.
1215 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1216 easier to resume long computations after using *BREAK-ON-SIGNALS*
1217 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1218 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1219 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1220 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1221 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1222 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1224 * fixed a (fairly theoretical) bug in string printing: if
1225 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1226 does not have array-element-type equal to the most general string
1228 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1229 function name. (thanks to Nikodemus Siivola)
1230 * fixed bug: compile-time format string checker failed on
1231 non-closed ~{. (reported by Thomas F Burdick)
1232 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1233 don't warn on legal loop code involving more than one aggregate
1234 boolean. (thanks to Nikodemus Siivola)
1235 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1236 in keyword position to MAKE-PACKAGE was not being reported as
1238 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1239 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1241 * fixed bugs 280 and 312: the checking for multiple definitions in a
1242 file is less likely to become confused by uses of inline
1244 * fixed bug: the #S reader macro performs the keyword coercion
1245 specified for slot names. (reported by Kalle Niemitalo)
1246 * fixed bug: lambda lists may contain symbols whose names start with
1247 & but are not lambda-list-keywords; their occurrence triggers a
1249 * fixed bug 321: define-method-combination argument lambda lists do
1250 not cause invalid code to be generated when &optional or &aux
1251 variables have default values. (reported by Bruno Haible)
1252 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1253 have CLOS classes; this appears also to have fixed reported
1254 instability in stack exhaustion detection.
1255 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1256 MUFFLE-WARNING when no associated restart is present is now
1258 * optimization: rearranged the expansion of various defining macros
1259 so that each expands into only one top-level form in a
1260 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1262 * optimization: used a previously little-used slot in symbols to
1263 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1265 * fixed some bugs revealed by Paul Dietz' test suite:
1266 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1268 ** Fixed an optimization bug related to inheritance of initforms
1269 from local to shared slots.
1270 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1271 ** CAST optimizer forgot to flush argument derived type.
1272 ** print/read consistency on floats is now orders of magnitude
1273 more likely. (thanks also to Bruno Haible for a similar report
1275 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1276 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1277 for objects of type REAL. Make it so.
1278 ** SXHASH is specified (infelicitously) to respect similarity,
1279 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1280 it so. (thanks to Markus Ziegler)
1281 ** on the Alpha, the compiler succeeds in compiling functions
1282 returning a known number of arguments greater than 63.
1283 ** fixed handling of invalid NIL arguments in keyword position
1285 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1287 ** provided a readably-printable representation for RANDOM-STATE
1289 ** ensured that pathnames, where they have a namestring, always
1290 print using #P"..." syntax.
1292 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1293 * Support for the forthcoming 2.0 version of the NetBSD kernel
1294 running on x86 hardware has been added. (thanks to Perry
1295 E. Metzger most immediately, and others for their past work)
1296 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1297 current release version; SBCL's OpenBSD support had been broken
1298 since about the time of OpenBSD's switch to ELF binary format.)
1299 * [placeholder for DX summary]
1300 ** user code with &REST lists declared dynamic-extent, under high
1301 speed or space and low safety and debug optimization policy.
1302 * The manual has been converted to Texinfo format and the debugger
1303 chapter from the cmucl manual has been added.
1304 * A facility has been added to extract documentation strings from
1305 sbcl and store them as Texinfo-formatted snippets for inclusion in
1306 the manual (via Texinfo's @include directive)
1307 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1308 the test case to Dave Roberts)
1309 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1310 pointers. (reported by Sean Ross)
1311 * bug fix: PROFILE output is printed nicely even for large numerical
1312 values. (thanks to Zach Beane)
1313 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1314 greater than 32 handle EOF correctly.
1315 * bug fix: on X86 an immediate argument of the IMUL instruction is
1316 correctly printed in disassembly. (thanks to Lutz Euler)
1317 * bug fix: class slots in redefined classes preserve their old
1318 values. (thanks to Bruno Haible and Nikodemus Siivola)
1319 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1320 succeeds. (reported by Marco Baringer)
1321 * fixed some bugs revealed by Paul Dietz' test suite:
1322 ** READ-SEQUENCE now works on ECHO-STREAMs.
1323 ** RATIONALIZE works more according to its specification. (thanks
1326 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1327 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1328 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1329 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1330 should matter to you only if you rebind the printer control
1331 variables and then find you want different bindings in the
1332 debugger than in the ordinary execution of your program.)
1333 * The runtime build system has been tweaked to support building
1334 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1335 assembler and linker. (thanks to Nikodemus Siivola)
1336 * Unbound, undefined, undeclared variables now trigger full
1337 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1338 more often programmer error than deliberate exploitation of undefined
1340 * optimization: the hash algorithm for strings has changed to one
1341 that is less vulnerable to spurious collisions. (thanks to Juho
1343 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1344 needless bounds checking. (thanks to Juho Snellman)
1345 * optimization: implemented multiplication as a modular
1346 (UNSIGNED-BYTE 32) operation on the PPC backend.
1347 * fixed some bugs revealed by Paul Dietz' test suite:
1348 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1350 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1351 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1352 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1354 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1356 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1358 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1359 their output stream on EOF from read.
1360 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1361 have been read to end-of-file.
1362 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1364 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1365 description of determination of which consecutive characters
1367 ** Printing the "Space" character with escaping on now yields "#\\ ",
1368 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1369 ** Reading floating-point numbers with *READ-BASE* set to a number
1370 less than 10 works correctly.
1371 ** Reading floating-point numbers with *READ-BASE* set to a number
1372 more than 10 works correctly.
1373 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1374 the readtable currently in effect.
1376 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1377 * minor incompatible change: parsing of namestrings on a physical
1378 (Unix) host has changed; numbers after the final #\. in a
1379 namestring are no longer interpreted as a version field. This is
1380 intented to be largely invisible to the user, except that the
1381 meaning of the namestring "*.*.*" has changed: it now refers to a
1382 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1383 should usually be replaced by
1384 (make-pathname :name :wild :type :wild :version :wild)
1385 with the added benefit that this is more likely to be portable.
1386 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1387 option to OPEN now signals an error if the file being opened
1388 exists; this may have an impact on existing code.
1389 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1390 signals received, on the PowerPC platforms (both Linux and
1391 Darwin). (thanks to Pierre Mai for pointing out the location of
1393 * several fixes on OS X: The system now builds and runs cleanly on
1394 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1395 (thanks to Brian Mastenbrook)
1396 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1397 second-resolution: integer multiples of 1/3600 between -24 and 24.
1398 (thanks to Vincent Arkesteijn)
1399 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1400 type when called with 1 argument; PEEK-CHAR checked type of
1401 PEEK-TYPE only after having read first character from a
1402 stream. (reported by Peter Graves)
1403 * bug fix: the garbage collector now has much better locality
1404 behaviour, and in particular no longer treats all memory as being
1405 exhausted when many small objects point to each other in a deeply
1407 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1408 recognized as being TYPEP their class.
1409 * bug fix: the PUSHNEW documentation string has been corrected.
1410 (thanks to Vincent Arkesteijn)
1411 * bug fix: defaulting of the value for the last of an atypically
1412 large number of multiple values being bound was not being
1413 performed correctly on the Alpha or PPC platforms
1414 * optimization: implemented multiplication as a modular
1415 (UNSIGNED-BYTE 32) operation on the x86 backend.
1416 * optimization: SEARCH on simple-base-strings can now be open-coded.
1417 (see also contrib/compiler-extras.lisp for inspiration for
1418 teaching the compiler about the Boyer-Moore algorithm).
1419 * value, returned by MAX (and MIN) called with several EQUALP, but
1420 not EQL, arguments now does not depend on compiler settings.
1421 * fixed some bugs revealed by Paul Dietz' test suite:
1422 ** in stack analysis liveness information is propagated from
1423 non-local entry points.
1424 ** pathwise CAST removing failed when the CAST node did not start
1426 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1427 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1429 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1430 host is already defined.
1431 ** RENAME-FILE works on streams instead of signalling an internal
1433 ** PEEK-CHAR uses the current readtable when determining whether
1434 or not a character is whitespace.
1435 ** MERGE-PATHNAMES handles the case when the pathname does not
1436 specify a name while the default-pathname specifies a version.
1437 ** Pathnames now stand a better chance of respecting print/read
1439 ** Attempting to use standardized file system operators with a
1440 pathname with invalid :DIRECTORY components signals a
1442 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1444 ** DIRECTORY on logical pathnames is more correct.
1445 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1446 signal a TYPE-ERROR if their argument is not a stream.
1447 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1448 designator argument does not designate a stream.
1449 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1450 examining the synonym.
1451 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1453 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1454 element-type larger than ([UN]SIGNED-BYTE 32).
1456 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1457 * When built with the :SB-FUTEX feature, threaded builds now take
1458 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1459 for faster/more reliable mutex and condition variable support.
1460 * Incompatible change (but one you probably shouldn't have been using
1461 anyway): the interface and code for arbitrating between multiple
1462 threads in the same user session has been redesigned.
1463 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1464 error when called without an explicit environment argument.
1465 (thanks to Nikodemus Siivola)
1466 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1467 confused about their position. (thanks to Adam Warner and Gerd
1469 * bug fix: Pretty printing backquoted forms with unquotations in the
1470 argument list position of various code constructs such as LAMBDA
1471 now works correctly. (reported by Paul Dietz)
1472 * bug fix: Pretty printing unquotations no longer loses all
1473 stream position information.
1474 * optimization: performance of string output streams is now less
1475 poor for multiple small sequence writes.
1476 * optimization: performance of CSUBTYPEP in the presence of complex
1477 expressions involving CONS and NOT many times has been improved.
1478 (reported by Paul Dietz)
1479 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1481 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1482 streams for the REPL, for the debugger, and so on. (thanks to
1484 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1485 * fixed some bugs revealed by Paul Dietz' test suite:
1486 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1487 is ignored if it is too small, rather than propagating through
1488 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1489 ** extremely complex negations of CONS types were not being
1490 sufficiently canonicalized, leading to inconsistencies in
1492 ** VALUES tranformer lost derived type.
1494 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1495 * fixed a bootstrapping bug: the build process no longer assumes
1496 that the various BOOLE-related constants have the same value in
1497 host and target lisps. (noted by Paul Dietz' test suite on an
1498 SBCL binary built from CLISP)
1499 * The system can now be dynamically linked on the MIPS platform,
1500 which enables dynamic loading of foreign code from Lisp. (thanks
1501 to Ralf Baechle for discussions on the MIPS ABI)
1502 * The system now records debugging information for its own source
1503 files in a filesystem-position-independent manner, relative to
1504 the "SYS" logical host.
1505 * fixed a compiler bug: MV-LET convertion did not check references
1506 to the "max args" entry point. (reported by Brian Downing)
1507 * tweaked disassembly notes to be less confident about proclaiming
1508 some instruction as an LRA. (thanks to Brian Downing)
1509 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1510 now each have their own history, command character, and other
1511 characteristics. (thanks to David Lichteblau)
1512 * fixed some bugs revealed by Paul Dietz' test suite:
1513 ** compiler failure in compiling LOGAND expressions including a
1515 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1516 shift greater than 32.
1517 ** FLUSH-DEST did not mark blocks for type check regeneration.
1518 ** HANDLER-CASE failed to accept declarations in handler clauses
1519 in some circumstances.
1521 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1522 * New code in contrib/sb-introspect (still probably not entirely
1523 stable yet) provides some support for smart Lisp development
1524 environments like SLIME.
1525 * The conditions signalled for errors occurring when loading .fasl
1526 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1527 in a way which should help ASDF recover gracefully.
1528 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1529 stuff has been cleaned up. If you code contrib/ stuff, this might
1530 affect you, and you can look at contrib/README, contrib/STANDARDS,
1531 and/or the 0.8.4.27 diff to check.
1532 * In full calls the compiler now does not generate checks for declared
1533 argument types for all arguments.
1534 * various threading fixes
1535 ** and some experimental patches which didn't make it into
1536 the main tree for this release, but which are shipped in
1537 contrib/experimental-thread.patch as a possible fix for some
1538 failures (deadlock, spinning...) in GC-intensive multithreaded
1540 * fixed PPC build problem (source code incompatibility of different
1541 library versions): added offsetof() hackery which attempts to divine
1542 where glibc maintainers put uc_mcontext today
1543 * fixed bug 282: compiler does not trust type assertions while passing
1544 arguments to a full call.
1545 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1546 &OPTIONAL VALUES type parameter.
1547 * fix bug 214: algorithm for noting rejected templates is now more
1548 similar to that of template seletion. (also reported by rydis on
1550 * fixed bug 141b: printing backquoted information readably and prettily
1551 inserts a space where necessary.
1552 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1553 (CHARACTER 10) are now reported as errors, rather than propagated
1554 as unknown types. (reported by piso on #lisp)
1555 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1556 with values NIL and :ERROR. (thanks to Milan Zamazal)
1557 * fixed bug 191c: CLOS now does proper keyword argument checking as
1558 described in CLHS 7.6.5 and 7.6.5.1.
1559 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1560 longer raise an error; further, using a list as a for-as-arithmetic
1561 counter now raises a meaningful error.
1562 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1563 now understood by sequence creation functions such as MAKE-SEQUENCE
1565 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1566 a STRING-INPUT-STREAM.
1567 * compiler enhancement: SIGNUM is now better able to derive the type
1569 * type declarations inside WITH-SLOTS are checked. (reported by
1571 * fixed some bugs revealed by Paul Dietz' test suite:
1572 ** incorrect optimization of TRUNCATE for a positive first
1573 argument and negative second.
1574 ** compiler failure in let-convertion during flushing dead code.
1575 ** compiler failure while deriving type of TRUNCATE on an
1576 interval, containing 0.
1577 ** ASH of a negative bignum by a negative bignum count now returns
1579 ** intersection of CONS types now canonicalizes properly, fixing
1580 inconsistencies in SUBTYPEP.
1582 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1583 * incompatible change: The --disable-debugger command line
1584 option now clobbers the debugger at a more fundamental
1585 level, by redefining #'INVOKE-DEBUGGER instead of by
1586 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1587 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1588 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1589 enter the debugger REPL and then suffer recursive errors
1590 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1591 while under the new system, BREAK in --disable-debugger mode
1592 terminates the system just as an unhandled error would.
1593 * fixed compiler performance when processing loops with a step >1;
1594 * bug fix: DOCUMENTATION now retrieves generic function
1595 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1596 support has been systematized, and now supports the methods
1597 specified by ANSI, along with a default method and a method for
1598 slot documentation. (reported by Nathan Froyd)
1599 * bug fix: effective methods associated with a generic function are
1600 no longer cached over a change of that generic function's method
1601 combination. (reported by Andreas Fuchs)
1602 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1603 names a non-existent directory. (thanks to Andreas Fuchs)
1604 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1605 the PPC platform, lead to stack corruption; this has been fixed.
1606 (reported by Rainer Joswig)
1607 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1608 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1609 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1610 arithmetic in general, is now much more reliable on the PPC
1612 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1613 platform now returns the right answer.
1614 * optimization: restored some effective method precomputation in
1615 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1616 precomputation is now tunable.
1617 * optimization: compiler-internal data structure use has been
1618 reviewed, and changes have been made that should improve the
1619 performance of the compiler by about 20%.
1620 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1621 simple-base-strings and simple-bit-vectors is improved.
1622 * optimization: the optimization of 32-bit logical and arithmetic
1623 functions introduced in version 0.8.3 on the x86 has been
1624 implemented on the mips, ppc and sparc platforms; an
1625 implementation of the same facility, but for 64-bit arithmetic,
1626 has been added for the alpha.
1627 * microoptimization: the compiler is better able to make use of the
1628 x86 LEA instruction for multiplication by constants.
1629 * bug fix: in some situations compiler did not report usage of
1630 generic arithmetic in (SPEED 3) policy.
1631 * bug 145b fix: compiler used wrong type specifier while converting
1632 MEMBER-types to numeric.
1633 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1635 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1636 index argument now works properly on the Alpha platform.
1637 * bug fix: floating point exception treatment on the Alpha platform
1639 * bug fix: FILE-POSITION works much better on string input and
1640 output streams. (thanks to Nikodemus Siivola)
1641 * bug fix: many threading/garbage collection symptoms sorted.
1642 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1643 might be pseudo-atomic.
1644 * internal change: Stopping for GC is now done with signals not
1645 ptrace. GC is now done in whichever thread wanted it, instead of
1647 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1649 * bug fix: result form in DO is not contained in the implicit
1651 * incompatible change: ICR structure is changed; the value part of
1652 CONTINUATION is now called LVAR; corresponding functions are
1653 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1654 * added type deriver for ISQRT (thanks to Robert E. Brown).
1655 * bug fix: better support for loading from the command line when an
1656 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1658 * fixed some bugs revealed by Paul Dietz' test suite:
1659 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1660 ** ROUND and FROUND now give the right answer when given very
1661 small float arguments.
1662 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1664 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1665 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1666 ** LCM should return a non-negative integer.
1667 ** PARSE-INTEGER returned the index of a terminator instead of the
1668 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1669 ** PARSE-INTEGER returned an incorrect index being applied to a
1671 ** LCM with two arguments of 0 returns 0 rather than signalling
1673 ** unsigned addition of a 32-bit constant with the high bit set no
1674 longer causes an internal compiler error on the x86.
1675 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1676 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1677 with negative last argument.
1678 ** byte specifiers with zero size and position no longer cause
1679 an error during type derivation.
1680 ** bignum multiplication on the Alpha platform now returns the
1682 * porting: The system now builds on SuSE AMD64, although it still
1683 generates a 32-bit binary.
1684 * .fasl file incompatibility: The fasl file version number has
1685 been incremented (because of the changes to internal compiler
1686 data structures referred to above).
1688 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1689 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1690 more accurately, on the Darwin kernel running on PowerPC hardware.
1691 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1692 * Compiler code deletion notes now signal a condition of type
1693 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1694 an associated MUFFLE-WARNING restart.
1695 * The compiler now performs limited argument count validation of
1696 constant format strings in FORMAT, and where appropriate in ERROR,
1697 CERROR and WARN. (thanks to Gerd Moellmann)
1698 * New ASDF-INSTALL contrib can be used for automatic download and
1699 installation of third-party Lisp code from CCLAN or other sites
1701 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1702 which forces another thread to execute a function supplied by the
1704 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1705 now accept and act upon their :ELEMENT-TYPE keyword argument.
1706 (reported by Martin Atzmueller, Edi Weitz)
1707 * bug fix: FILE-POSITION now accepts position designators up to
1708 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1709 is the greater. (thanks to Patrik Nordebo)
1710 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1711 Kalle Olavi Niemitalo)
1712 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1713 printable. (reported by Eric Marsden)
1714 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1715 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1716 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1717 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1718 (thanks to Stig Sandoe)
1719 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1720 UNREAD-CHAR now correctly includes the unread character in the
1721 target sequence. (thanks to Gerd Moellmann)
1722 * bug fix in threaded builds: the system can now be suspended and
1723 resumed by shell job control with minimal disruption.
1724 * bug fixes in times and timezones >2038AD
1725 * better handling of "where is GNU make?" problem in build scripts
1726 (thanks to Nikodemus Siivola)
1727 * new optimization: inside a named function any reference to a
1728 function with the same name is considered to be a self-reference;
1729 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1730 optimization quality.
1731 * new optimization on x86: logical functions and + now have
1732 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1733 used when the result is truncated to 32 bits.
1734 * VALUES declaration is partially enabled.
1735 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1736 * fixed some bugs revealed by Paul Dietz' test suite:
1737 ** The system now obeys the constraint imposed by
1738 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1739 types form a lattice under type intersection.
1740 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1741 ** ASSOC now ignores NIL elements in an alist.
1742 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1743 and (1+ MOST-POSITIVE-FIXNUM) answers.
1744 ** The addition of a method with invalid qualifiers to a generic
1745 function does not cause an error to be signalled immediately;
1746 a warning is signalled, and the error is generated only on
1747 calling the generic function.
1748 * changed .fasl file version number, in part to add type codes for
1749 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1750 obscure ANSI requirements
1752 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1753 * fixed bug 148: failure to inline-expand a local function left
1754 garbage, confusing the compiler.
1755 * fixed bugs 3cd: structure slot readers perform type check if the
1756 slot can have an invalid value (i.e. it is either not initialized
1757 or can be written with a less specific slot writer).
1758 * bug fix: the compiler now traps array references to elements off
1759 the end of an array; previously, the bounds checking in some
1760 circumstances could go off-by-one.
1761 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1763 * type declarations for array element types now obey the description
1764 on the CLHS page "Declaration TYPE", as per discussions on
1765 sbcl-help around 2003-05-08. This means that a declaration
1766 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1767 declaration, all references to BAR will be asserted or assumed
1768 (with THE, so dependent on compiler policy) to involve objects of
1769 type FOO. Note that no such declaration is implied in
1770 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1771 * declared types of functions from the "Conditions"
1772 chapter. (reported by Paul Dietz)
1773 * bug fix: CERROR accepts a function as its first argument.
1774 * bug fix: NTH an NTHCDR accept a bignum as index
1775 arguments. (reported by Adam Warner)
1776 * optimization: character compare routines now optimize comparing
1777 against a constant character. (reported by Gilbert Baumann)
1778 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1779 argument now works properly on the MIPS platform.
1780 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1781 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1782 declared to have a functional type, does not cause an error
1784 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1785 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1786 * fixed bug: if last continuation of a deleted block has a
1787 destination, this destination should be deleted too. (reported by
1789 * fixed a bug in the bootstrap process: the host compiler's values
1790 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1791 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1792 test case from Patrik Nordebo)
1793 * improved the ability of the disassembler on the PPC platform to
1794 provide helpful disassembly notes.
1795 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1796 the class in more cases than previously.
1797 * bug fix: FILE-POSITION now understands :START and :END for
1798 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1799 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1800 8) data. (thanks to David Lichteblau)
1801 * bug fix: it is possible to add a method to a generic function
1802 without lambda list.
1803 * bug fix: reader failed to signal END-OF-FILE inside an
1804 object representation. (reported by Nikodemus Siivola)
1805 * fixed some bugs revealed by Paul Dietz' test suite:
1806 ** LAST and [N]BUTLAST should accept a bignum.
1807 ** condition slot accessors are methods.
1808 ** (VECTOR NIL) is a subtype of STRING.
1810 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1811 * minor incompatible change: some nonsensical specialized lambda
1812 lists (used in DEFMETHOD) which were previously ignored now signal
1814 * minor incompatible change: the system is now aware of the types of
1815 variables in the COMMON-LISP package, and will signal errors for
1816 most violations of these type constraints (where previously they
1817 were silently accepted).
1818 * minor incompatible change: COMPILE-FILE now uses the freedom
1819 afforded (ANSI 3.2.2.3) to use derived function types for
1820 functions defined in the same file. This also permits the system
1821 to warn on static type mismatches and function
1822 redefinition. (Currently it does not work with high DEBUG level.)
1823 * minor incompatible change: VALUES declaration is disabled.
1824 * When issuing notes, the compiler now signals a condition of type
1825 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1826 restart for use in user handlers. It is expected that the
1827 COMPILER-NOTE condition will eventually become a condition
1828 supertype to a hierarchy of note types, which will then be
1829 handleable in a similar fashion. However, at the moment, no such
1830 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1831 but it's an implementation detail, not a classification for the
1833 * Changes in type checking closed the following bugs:
1834 ** type checking of unused values (192b, 194d, 203);
1835 ** template selection based on unsafe type assertions (192c, 236);
1836 ** type checking in branches (194bc).
1837 * A short form of VALUES type specifier has ANSI meaning (it has
1838 increased the number of situations when SBCL cannot perform type
1840 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1841 keywords or constants is permissible.
1842 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1843 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1844 classes. (thanks to Antonio Martinez)
1845 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1846 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1847 argument to a pretty-print stream. Instead, it leaves any such
1848 operations to DESCRIBE-OBJECT methods.
1849 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1851 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1852 lambda lists. (thanks to David Lichteblau)
1853 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1854 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1855 (thanks to Gerd Moellmann)
1856 * fixed bug 246: increased compilation speed of long
1857 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1859 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1861 * DEFCONSTANT now throws a condition of type
1862 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1863 constant to a non-EQL value; CONTINUE and ABORT restarts
1864 respectively change and preserve the value.
1865 * fixed bug 63: The code walker, part of the implementation of CLOS,
1866 is now better at handling symbol macros.
1867 * bug fix: There is no longer an internal implementation type named
1868 CL:LENGTH. (reported by Raymond Toy)
1869 * bug fix: In macro-like defining macros/special operators the
1870 implicit block does not enclose the lambda list.
1871 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1872 enclosed in parentheses) are not suitable as type specifiers, and
1873 their use properly signals an error now.
1874 * bug fix: An argument count mismatch for a type specifier in code
1875 being compiled no longer causes an unhandled error at compile
1876 time, but signals a compile-time warning.
1877 * fixed simple vector readable printing
1878 * bug fix: DESCRIBE takes more care over whether the class
1879 precedence list slot of a class is bound before accessing it.
1880 (reported by Markus Krummenacker)
1881 * bug fix: FORMATTER can successfully compile pretty-printer format
1882 strings which use variants of the ~* directive inside.
1883 * bug fix: SEARCH now applies its TEST predicate to the elements of
1884 the arguments in the correct order. (thanks to Wolfhard Buss)
1885 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1886 in type inference. (thanks to Robert E. Brown)
1887 * bug fix: Reading in symbols with an explicit package name of ""
1888 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1889 * fixed some bugs revealed by Paul Dietz' test suite:
1890 ** NIL is now allowed as a structure slot name.
1891 ** Arbitrary numbers, not just REALs, are allowed in certain
1892 circumstances in LOOP for-as-arithmetic clauses.
1893 ** Multiple class redefinitions before slot access no longer
1894 causes a type error.
1895 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1896 association between the name and a class.
1897 ** Generic functions with non-standard method-combination and over
1898 five methods all of which return constants no longer return NIL
1899 after the first few invocations. (thanks to Gerd Moellmann)
1900 ** CALL-NEXT-METHOD with no arguments now passes the original
1901 values of the arguments, even in the presence of assignment.
1902 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1904 ** DISASSEMBLE works with closures and funcallable instances.
1905 ** ADD-METHOD now returns the generic function, not the new method.
1906 ** FIND-METHOD signals an error if the lengths of the specializers
1907 is incompatible with the generic function, even if the ERRORP
1909 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1910 which its argument is a member.
1911 ** DEFCLASS only redefines the class named by its class-name
1912 argument if that name is the proper name of the class;
1913 otherwise, it creates a new class.
1914 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1915 of the UNBOUND-SLOT condition to the name of the slot.
1916 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1917 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1918 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1919 SLOT-MAKUNBOUND in the specified fashion.
1921 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1922 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1923 cross-compilation host. As a consequence, we can now bootstrap our
1924 way up to SBCL starting with a bare gcc toolchain and human-readable
1925 source code (first the source to CLISP, then the source to SBCL).
1926 * A contributed module containing a partial implementation of the
1927 simple-streams interface has been included. (thanks to Rudi
1929 * A contributed module implementing the RFC1321 Message Digest
1930 Algorithm, known as MD5, has been included.
1931 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1932 no longer has any effect, as the code controlled by this feature
1933 has been deleted. (As far as we know, no-one has ever built using
1934 this feature, and its semantics were confused in any case).
1935 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1936 work on conditions (as required by the ANSI specification),
1937 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1938 expected behaviour on conditions. Users should note, however,
1939 that such behaviour is not required by the ANSI specification,
1940 and so use of this behaviour may render their code unportable.
1941 * fixed some bugs revealed by Paul Dietz' test suite:
1942 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1944 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1945 ** SXHASH on condition objects no longer returns NIL.
1946 ** :ALLOCATION :CLASS slots are better treated; their values are
1947 updated on class redefinition, and initforms inherited from
1948 superclasses are applied.
1949 ** REMOVE-METHOD returns its generic function argument even when
1950 no method was removed.
1951 ** SHARED-INITIALIZE now initializes the values of the requested
1952 slots, including those with :ALLOCATION :CLASS.
1953 ** ALLOCATE-INSTANCE now works on structure classes defined via
1954 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1956 ** SLOT-EXISTS-P now works on conditions, as well as structures
1958 ** MAKE-LOAD-FORM now has the required methods on
1959 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1960 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1961 keyword, and now implements the SLOT-NAMES argument.
1962 ** methods with &OPTIONAL arguments no longer allow too many
1963 arguments to be passed in the call without error.
1964 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1965 option is consistent with the required arguments of the generic
1966 function lambda list.
1967 * bug fix: REQUIRE accepts a string designator. (Thanks to
1969 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1970 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1971 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1972 * bug fix: The compiler now checks for duplicated variables in macro
1974 * bug fix: SETQ on globals returns the correct value.
1975 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1976 not a COMPILER-ERROR (followed by some other strange error on
1977 choosing the CONTINUE restart).
1978 * bug fix: make.sh and friends are now more consistent in the way that
1979 they look for GNU "make".
1981 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1982 * experimental native threads support (on x86 Linux >=2.4 only).
1983 This is not compiled in by default: you need to add :SB-THREAD to
1984 the target features. See the "Beyond ANSI" chapter of the manual
1986 * fix for longstanding nonANSIism: The old distinction between
1987 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1988 The return value from CL:FIND-CLASS is now a CLOS class, and
1989 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1990 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1991 * An interface to the MetaObject Protocol, as described in Kiczales,
1992 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1993 MIT Press, 1991, is available from the SB-MOP package.
1994 * incompatible change: the SB-PCL package should now be considered
1995 a private implementation detail, and no longer a semi-private MOP
1997 * minor incompatible change: due to rearrangement for threads, the
1998 control stack and binding stack are now allocated at arbitrary
1999 addresses instead of being hardcoded per-port. Users affected by
2000 this probably have to be doing advanced things with shared
2001 libraries, and will know who they are.
2002 * minor incompatible change: Previously, all --eval forms used were
2003 processed with READ before any of them were processed with EVAL.
2004 Now each --eval form is processed with both READ and EVAL before
2005 the next --eval form is processed. (Thus package operations like
2006 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2007 work as the user might reasonably expect.)
2008 * minor incompatible change: *STANDARD-INPUT* is now only an
2009 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2011 * minor incompatible change: Y-OR-N-P is now character-oriented,
2012 not line oriented. Also, YES-OR-NO-P now works without errors.
2013 (thanks to Antonio Martinez)
2014 * sb-aclrepl module improvements: an integrated inspector, added
2015 repl features, and a bug fix to :trace command.
2016 * Known functions, which cannot be open coded by the backend, are
2017 considered to be able to check types of their arguments. (fixing
2018 a bug report by Nathan J. Froyd)
2019 * fixed a bug in computing method discriminating functions: It is
2020 now possible to define methods specialized on classes which have
2021 forward-referenced superclasses. (thanks to Gerd Moellmann)
2022 * fixed evaluation order in optional entries (reported by Gilbert
2024 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2025 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2026 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2027 required-by-AMOP NAME argument, as well as CLASS and
2028 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2029 * fixed bug 20: DEFMETHOD can define methods using names that are
2030 not the proper names of classes to designate class specializers.
2031 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2032 Unix file descriptors, instead of blowing up. (thanks to Antonio
2034 * Garbage collection refactoring: user-visible change is that a
2035 call to the GC function during WITHOUT-GCING will not do garbage
2036 collection until the end of the WITHOUT-GCING. If you were doing
2037 this you were probably losing anyway.
2038 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2039 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2040 TYPEP the latter but not the former.
2041 * The compiler issues a full WARNING for calls to undefined functions
2042 with names from the CL package.
2043 * MAP-INTO for a vector destination is open coded. (reported by
2044 Brian Downing on c.l.l)
2045 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2046 documentation string.
2047 * fixed some bugs revealed by Paul Dietz' test suite:
2048 ** COPY-ALIST now signals an error if its argument is a dotted
2050 ** Condition slots are now accessed more correctly in the presence
2051 of multiple initargs for a given slot.
2052 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2053 exclude restarts of the same name associated with a different
2055 ** DEFCLASS of forward-referenced classes with another
2056 forward-referenced class in the superclasses list no longer
2058 ** Condition slots are now initialized once each, not multiple
2059 times. (thanks to Gerd Moellmann)
2060 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2062 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2063 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2064 arguments contain duplicated elements.
2065 ** RESTART-CASE understands local macros.
2066 ** RESTART-CASE associates exactly its own restarts with a condition.
2067 ** ENDP in safe mode checks its argument to be of type LIST.
2068 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2069 in question is unbound.
2070 ** Optimized MAKE-INSTANCE functions no longer cause internal
2071 assertion failures in the presence of duplicate initargs.
2072 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2073 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2074 propagated even in the ctor optimized implementation of
2076 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2078 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2079 * a better implementation of SXHASH on (simple) bit vectors,
2080 measured both in execution speed and in distribution of results
2081 over the positive fixnums, has been installed. Likewise, a better
2082 implementation of EQUAL for simple bit vectors is now available.
2083 * fixed CEILING optimization for a divisor of form 2^k.
2084 * fixed bug 240 (emitting extra style warnings "using the lexical
2085 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2086 by Antonio Martinez)
2087 * fixed SXHASH, giving different results for NIL depending on type
2088 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2089 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2090 variable. (found by Rolf Wester)
2091 * SBCL does not ignore type declarations for special
2092 variables. (reported by rif on c.l.l 2003-03-05)
2093 * some bug fixes in contrib/sb-aclrepl/
2094 * fixed some bugs revealed by Paul Dietz' test suite:
2095 ** a bug in the CONS type specifier, whereby the CAR and CDR
2096 types got intertwined, has been fixed;
2097 ** the type system is now able to reason about the interaction
2098 between INTEGER and RATIO types more completely;
2099 ** APPEND, [N]REVERSE and NRECONC check that those their
2100 arguments, which must be proper lists, are really so;
2101 ** An array specialized to be unable to hold elements has been
2102 implemented, as required -- yes, really -- by ANSI;
2103 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2104 on malformed property lists;
2106 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2107 * incompatible packaging change: in line with Unix convention,
2108 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2109 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2110 * REQUIRE and PROVIDE are now optionally capable of doing something
2111 useful. See the documentation string for REQUIRE.
2112 * infrastructure for a managed SBCL contrib system: contributed
2113 modules in this release include:
2114 ** the ASDF system definition facility;
2115 ** an interface to the BSD Sockets API;
2116 ** an ACL-like convenience interface to the repl;
2117 (thanks to Kevin Rosenberg)
2118 ** an implementation of ROTATE-BYTE, with efficient implementation
2120 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2121 gives a closer approximation to the right answer than previously.
2122 (thanks to Raymond Toy)
2123 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2124 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2125 optional environment arguments, as required by ANSI.
2126 * fixed bugs in other functions taking environment objects, allowing
2127 calls with an explicit NIL environment argument to be compiled
2129 * fixed bug 228: primary return values from
2130 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2131 COMPILE or FUNCTION.
2132 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2133 :TYPE will no longer signal errors on innocuous objects.
2134 * fixed bug 231b: SETQ is better at respecting type declarations in
2135 the lexical environment.
2136 * fixed a bug in DEFCLASS: classes named by symbols with no or
2137 unprintable packages can now be defined.
2138 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2139 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2140 * fixed compiler failure related to checking types of functions.
2141 (reported by Robert E. Brown)
2142 * the compiler is now much more consistent in its error-checking
2143 treatment of bounding index arguments to sequence functions: in
2144 (SAFETY 3) code, errors will be signalled in almost all cases if
2145 invalid sequence bounding indices are passed to functions defined
2146 by ANSI to operate on sequences.
2147 * fixed a bug in the build procedure: documentation of SBCL-specific
2148 packages is now preserved and available in the final Lisp image.
2149 * lifted FDEFINITION lookup out of loops in the implementation of
2150 many list operations. (thanks to Robert E. Brown)
2151 * fixed a bug in the reader: the #n# reader macro now works for
2152 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2153 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2154 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2155 E. Brown and Tony Martinez respectively)
2156 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2157 SLOT-BOUNDP: in particular, it is now able to optimize them much
2158 better, and is now not vulnerable to having packages renamed.
2159 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2160 always called when a slot is not present in an instance. (thanks
2162 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2163 longer signal an error on unreading a character following EOF on
2164 the previous constituent stream. (thanks to Tony Martinez)
2165 * fixed some bugs revealed by Paul Dietz' test suite:
2166 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2167 not just nonnegative fixnums;
2168 ** the logical bit-array operators such as BIT-AND now accept an
2169 explicit NIL for their "opt-arg" argument (to indicate a
2170 freshly-consed result bit-array);
2171 ** ELT now signals an error on an invalid sequence index in safe
2173 ** the type system is now cleverer about negations of numeric
2174 types, and consequently understands the BIGNUM and RATIO types
2176 ** the type system is now cleverer about the interaction between
2177 INTEGER and RATIO types: while bugs still remain, many more
2178 cases are accurately computed;
2179 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2180 if it is in the last clause;
2181 ** CONSTANTLY now correctly returns a side-effect-free function in
2183 ** DECLARE is no longer treated as a special-operator; in
2184 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2185 * incremented fasl file version number due to the change in the
2186 DEFSTRUCT-SLOT-DESCRIPTION structure.
2188 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2189 * minor incompatible change: code processed by the "interpreter" or
2190 EVAL now has a compilation optimization policy of (DEBUG 2)
2191 (changed from (DEBUG 1)) to improve debuggability of interactive
2192 development, and to allow the use of the debug RETURN command in
2194 * an experimental implementation of the RETURN command for the
2195 debugger has been included. (thanks to Frederik Kuivinen)
2196 * fixed bug 62: constraints were not propagated into a loop.
2197 * fixed bug in embedded calls of SORT (reported and investigated by
2199 * fixed some bugs revealed by Paul Dietz' test suite:
2200 ** printing and reading of arrays with some dimensions having
2201 length 0 (thanks to Gerd Moellmann);
2202 ** BOA constructor with &AUX argument without a default value does
2203 not cause a type error;
2204 ** CONSTANTP now returns true for all self-evaluating objects.
2206 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2207 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2208 accessors that are related by inheritance, as specified in the
2209 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2210 Valtteri Vuorikoski)
2211 * The compiler is now able to inline functions that were defined in
2212 a complex lexical environment (e.g. inside a MACROLET).
2213 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2215 * Support for the upcoming FreeBSD-5.0 release has been included.
2216 (thanks to Dag-Erling Smorgrav)
2217 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2218 effect when it is not in a toplevel context.
2219 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2220 stand a better chance of being correct. (thanks to Gerd
2222 * fixed bug in COERCE, which now signals an error on coercing a
2223 rational to a bounded real type which excludes the expected
2225 * The compiler is now able to derive types more accurately from the
2226 COERCE and COMPILE functions.
2227 * fixed bug 223: functional binding is considered to be constant
2228 only for symbols in the CL package.
2229 * fixed bug 231: SETQ did not check the type of a variable being set
2230 (reported by Robert E. Brown)
2231 * A new optimization for MAKE-INSTANCE has been included, fixing
2232 various bugs (including relating to :ALLOCATION :CLASS slots and
2233 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2235 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2236 ** As required by ANSI, LOOP now disallows anonymous collection
2237 clauses such as COLLECT I in conjunction with aggregate boolean
2238 clauses such as THEREIS (= I 1);
2239 ** LOOP now signals an error when any variable is reused in the
2240 same loop (including the potentially useful construct analogous
2241 to WITH A = 1 WITH A = (1+ A);
2242 ** IT is only a special loop symbol within the first clause of a
2243 conditional loop clause;
2244 ** LOOP with a typed iteration variable over a hashtable now
2245 signals a type error iff it should.
2246 * fixed some other bugs revealed by Paul Dietz' test suite:
2247 ** FILE-STREAM now names the class previously known as FD-STREAM;
2248 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2249 argument) no longer signals an error;
2250 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2251 of the slot symbol, rather than using the current package
2252 ((:CONC-NAME "") continues to intern the slot's name in the
2254 * incremented fasl file version number, because of the incompatible
2255 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2256 of the new implementation of DEFINE-COMPILER-MACRO.
2258 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2259 * Support for building SBCL for MIPS platforms running in
2260 little-endian mode has now been checked in, and basic
2261 functionality on said platforms verified.
2262 * minor incompatible change: PCL now records the pathname of a file
2263 in which methods and the like are defined, rather than its
2265 * minor incompatible change: TRUENAME now considers the truename of
2266 a file naming a directory to be the pathname with :DIRECTORY
2267 component indicating that directory.
2268 * minor incompatible change: a NAMED clause in the extended form of
2269 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2270 reason for the previous behaviour is unclear.
2271 * more systematization and improvement of CLOS and MOP conformance
2272 in PCL (thanks to Gerd Moellman and Pierre Mai):
2273 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2275 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2276 primary methods with no specializers;
2277 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2279 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2280 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2281 FORWARD-REFERENCED-CLASSes; error reporting on
2282 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2284 ** SXHASH on CLOS instances now uses a slot internal to the
2285 instance to return different numbers on distinct instances,
2286 while preserving the same return value through invocations of
2288 ** DEFMETHOD signals errors when methods with longer incongruent
2289 lambda lists are added to generic functions;
2290 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2291 CLASS, as specified in AMOP;
2292 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2293 on the order returned by the primary method for classes of
2294 class STANDARD-CLASS;
2295 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2296 * fixed some bugs shown by Paul Dietz' test suite:
2297 ** DOLIST puts its body in TAGBODY;
2298 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2300 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2301 value producing form;
2302 ** if more variables are given to PROGV than values, extra
2303 variables are bound and made to have no value;
2304 ** NSUBSTITUTE on list arguments gets the right answer with
2306 ** ELT signals an error of type TYPE-ERROR when the index argument
2307 is not a valid sequence index;
2308 ** LOOP signals (at macroexpansion time) an error of type
2309 PROGRAM-ERROR when duplicate variable names are found;
2310 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2311 ** FUNCALL of special-operators now cause an error of type
2313 ** PSETQ now works as required in the presence of side-effecting
2314 symbol-macro places;
2315 ** NCONC accepts any object as its last argument;
2316 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2318 ** loop-for-as-package does not require a package to be explicitely
2320 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2322 * fixed bug 166: compiler preserves "there is a way to go"
2323 invariant when deleting code.
2324 * fixed bug 172: macro lambda lists with required arguments after
2325 &REST arguments now cause an error to be signalled. (thanks to
2327 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2328 bug. (thanks to Matthew Danish)
2329 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2331 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2332 arguments are lexically rebound. (thanks to Gerd Moellmann and
2334 * fixed bug 194: error messages are now more informative when there
2335 is no primary method applicable in a call to a generic
2336 function. (thanks to Gerd Moellmann)
2337 * fixed bug in command line argument checking (thanks to Julian
2339 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2340 (thanks to Matthew Danish)
2341 * incremented fasl file version number, because of the
2342 SXHASH-related changes in the layout of CLOS data structures
2344 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2345 * minor incompatible change: The runtime (the Unix executable named
2346 "sbcl") is now much pickier about the .core files it will load.
2347 Essentially it now requires .core files to descend from the same
2348 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2349 as the runtime does. (The intent is to prevent the crashes which
2350 can occur, and which can even be reported as mysterious failures,
2351 when people patch the sources or change the build parameters
2352 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2353 sbcl and .core files.)
2354 * fixed bug: VALUES-LIST is no longer optimized away.
2355 * fixed bug 142: The FFI conversion of C string values to Lisp
2356 string values no longer conses excessively. (thanks to Nathan
2357 Froyd porting Raymond Toy's fix to CMU CL)
2358 * began to systematize and improve MOP conformance in PCL (thanks to
2359 Nathan Froyd, Gerd Moellman and Pierre Mai):
2360 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2362 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2363 ** FINALIZE-INHERITANCE is now called on class finalization;
2364 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2365 argument precedence order.
2366 * fixed bug 202: The compiler no longer fails on functions whose
2367 derived types contradict their declared type.
2368 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2369 so it can be non-toplevel.
2370 * The fasl file version number has changed (because of the new
2371 implementation of DEFMACRO).
2372 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2373 safe code, that any length requirement by their type-specifier
2374 argument is valid. The exceptions to this are described in bug
2376 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2377 functions now check the types of their inputs as required by ANSI.
2378 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2379 introduced symbol is DECLAREd to be SPECIAL.
2380 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2381 * fixed bug: :COUNT argument to sequence functions may be negative
2382 * fixed bug: body of DO-SYMBOLS may contain declarations
2383 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2384 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2385 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2386 symbol macro only once
2387 * fixed printing of call frame when argument list is unavailable
2388 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2389 * compiler no longer signals WARNING on unknown keyword
2392 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2393 * A beta-quality port to the mips architecture running Linux,
2394 based on the old CMUCL backend, has been made. It has been tested
2395 on a big-endian kernel, and works sufficiently well to be able to
2396 rebuild itself; it has not been tested in little-endian mode.
2397 * fixed an inconsistency between gencgc.c and purify.c which made
2398 dumping/loading .core files unreliable
2399 * fixed bug 120a: The compiler now deals correctly with IFs where
2400 the consequent is the same as the alternative, instead of
2401 misderiving the return type. (thanks to Alexey Dejneka)
2402 * fixed bug 113: Logical pathnames are now dumpable (the logical
2403 host is resolved at load-time, throwing an error if it is not
2405 * fixed bug 174: FORMAT's error message is slightly clearer when a
2406 non-printing character is used in a format directive.
2407 * fixed several bugs in compiler checking of type declarations, i.e.
2408 violations of the Python "declarations are assertions" principle
2409 (thanks to Alexey Dejneka)
2410 * fixed several bugs in PCL's error checking (thanks to Gerd
2412 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2414 * fixed bug in compilation of functions as first class values
2415 (thanks to Antonio Martinez-Shotton)
2416 * The compiler's handling TYPE-ERRORs which it can prove will
2417 inevitably happen at runtime has been cleaned up and corrected
2418 in several ways. (thanks to Alexey Dejneka)
2419 * improved argument type checking for various basic arithmetic
2420 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2421 lavished on them in the past that they can be compiled in many
2422 ways in different special cases
2423 * fixed bug 181: compiler checks validity of user supplied type
2425 * cleaned up code flushing in optimization: Function calls which
2426 should signal errors for safety purposes (e.g. which ANSI says
2427 should signal errors when their arguments are of incorrect type)
2428 are no longer optimized away.
2429 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2430 * incremented fasl file version number, because changes in the
2431 implementation of sequence functions like COERCE caused
2432 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2433 in old inline expansions) to become undefined. (Actually these
2434 changes were later undone, so we might very well be binary
2435 compatible with 0.7.7 after all, but leaving the version number
2436 incremented seemed like the simplest and most conservative
2439 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2440 * An alpha-quality port to the parisc architecture running Linux,
2441 based on the old CMUCL backend, has been made. This, even more so
2442 than the other backends, should be considered still a work in
2443 progress; known problems include that the Linux kernel in 64-bit
2444 mode does not propagate the correct sigcontext structure to
2445 userspace, and consequently SBCL on a parisc64 kernel will not
2447 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2448 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2449 suppressed.) Also now that the compiler is looking at declarations
2450 in the environment, it checks optimization declarations as well,
2451 and suppresses inlining when (> DEBUG SPEED).
2452 * More fixes have been made to treatment of floating point exception
2453 treatment and other Unix signals. In particular, floating point
2454 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2455 * The detection and handling of control stack exhaustion (infinite
2456 or very deeply nested recursion) has changed. Stack exhaustion
2457 detection is now done by write-protecting pages at the OS level
2458 and applies at all optimization settings; when found, a
2459 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2460 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2461 be caught using IGNORE-ERRORS.
2462 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2463 correctly when transforming recursive LABELS functions to LETs.
2464 (thanks to Alexey Dejneka)
2465 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2466 that are names of constants or global variables.
2467 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2468 alien routines with docstrings.
2469 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2470 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2472 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2473 (thanks to Eric Marsden)
2474 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2475 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2476 * Bug fix: The SPARC backend can now compile functions involving
2477 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2478 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2479 to a functions expecting a PATHNAME-DESIGNATOR.
2480 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2481 lambda lists. (thanks to Alexey Dejneka)
2482 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2483 functions so that EXPORTing the name of the class doesn't cause
2484 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2485 to Antonio Martinez for reporting this)
2486 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2487 constant-folded without failing an assertion. (thanks to Einar
2488 Floystad Dorum for reporting this)
2489 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2490 types are now dealt with more correctly by the compiler.
2491 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2492 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2493 behaviour analogous to RENAME-FILE. This puts its behaviour more
2494 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2496 * The fasl file version number has changed again. (because of the
2497 bug fix involving the names of PCL MAKE-INSTANCE functions)
2499 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2500 * bug fix: Floating point exceptions are treated much more
2501 consistently on the x86/Linux and PPC/Linux platforms.
2502 * Array initialization with :INITIAL-ELEMENT is now much faster for
2503 cases when the compiler cannot open code the array creation, but
2504 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2505 array accesses have also seen a speed increase.
2506 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2507 specified. (This was at the root of some bad interactions between
2508 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2510 * bug fix: Internal error arguments for undefined functions are now
2511 computed correctly on the PPC/Linux platform.
2512 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2513 Raymond Toy's patch for CMU CL)
2514 * Support for the Solaris 9 operating environment has been included
2515 (thanks to Daniel Merritt)
2516 * A very ugly but hopefully complete draft of the missing FFI chapter
2517 of the manual has been created by reformatting the corresponding
2518 CMU CL manual chapter into (currently very ugly and incoherent)
2519 DocBook and bringing it up to date for SBCL behavior. Thus, the
2520 manual is now essentially complete, at least by my extreme
2521 once-and-only-once standards, whereby it's acceptable to refer to
2522 the doc strings of SB-EXT functions as the primary documentation.
2523 * The fasl file version number has changed again, due to cleanup of
2524 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2525 is no longer a static symbol.)
2527 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2528 * SBCL now builds with OpenMCL (version 0.12) as the
2529 cross-compilation host; also, more progress has been made toward
2530 bootstrapping under CLISP.
2531 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2533 * bug 158 fixed: The compiler can now deal with integer loop
2534 increments different from 1; fixing this turned out also to fix
2536 * bug 169 fixed: no more bogus warnings about using lexical bindings
2537 despite the presence of perfectly good SPECIAL declarations (thanks
2538 to David Lichteblau)
2539 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2540 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2541 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2543 * bug 180 fixed: Method combination specifications no longer ignore
2544 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2545 * bug fix: Structure type predicate functions now check their argument
2546 count as they should.
2547 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2548 correctly. (thanks to Pierre Mai)
2549 * minor incompatible change: The --noprogrammer option is deprecated
2550 in favor of the new --disable-debugger option, which is very similar.
2551 (The major difference is that it takes effect at a slightly different
2552 time at startup, causing handling of errors in --sysinit and
2553 --userinit files will be affected differently.) The
2554 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2555 been added to allow this functionality to be controlled from ordinary
2556 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2557 maintainers, who might want to run non-interactive scripts to
2558 build SBCL cores which will later be used interactively.)
2559 * minor incompatible change: The LOAD function no longer, when given
2560 a wild pathname to load, loads all files matching that pathname.
2561 Instead, an error of type FILE-ERROR is signalled.
2563 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2564 * bug 147 fixed: The compiler preserves its block link/count
2565 invariants more correctly now so that it doesn't crash. (thanks
2567 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2569 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2570 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2571 no-value-for-this-character as NIL (as specified by ANSI).
2572 * HOST-NAMESTRING on physical pathnames now returns a string that is
2573 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2574 (thanks to Christophe Rhodes)
2575 * The Alpha port handles icache flushing more correctly. (thanks to
2577 * More progress has been made toward bootstrapping under CLISP. (thanks
2578 to Christophe Rhodes)
2579 * The fasl file format has changed again, because dynamic loading
2580 on OpenBSD (which has non-ELF object files) motivated some cleanups
2581 in the way that foreign symbols are transformed and passed around.
2582 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2583 is no longer treated as whitespace by the reader, but instead as
2584 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
2585 |A
\x7fB|, instead of A as it used to.
2587 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2588 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2589 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2590 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2591 this as of 0.7.1.45, but was left out of the previous news section
2592 (thanks to Dan Barlow)
2593 * SBCL now runs on the Solaris operating system on SPARC architectures
2594 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2595 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2596 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2597 * SPARC backend cleanups, allowing builds of cores optimized for V8
2598 and V9 SPARCS, and also emission of code targeted to a particular
2599 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2601 * SBCL is closer to bootstrapping under CLISP, thanks to various
2602 fixes by Christophe Rhodes.
2603 * The fasl file format has changed again, to allow the compiler's
2604 INFO database to support symbol macros.
2605 * The user manual (in doc/) is formatted into HTML more nicely.
2606 (thanks to coreythomas)
2607 * The system is smarter about SUBTYPEP relationships, especially
2608 those involving NOT types (including types such as ATOM which are
2609 represented internally using NOT types). Thus SUBTYPEP is less
2610 likely to return (VALUES NIL NIL) in general, and in particular
2611 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2613 * The fasl file format has changed again, because the internal
2614 representation of types now includes a new slot to support the new
2615 SUBTYPEP-of-NOT-types logic.
2616 * (not a change in the main branch of SBCL, but a related prototype
2617 which can hopefully be merged into the main branch of SBCL in the
2618 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2619 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2620 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2621 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2622 dispatch macro character combinations. (thanks to Alexey Dejneka)
2623 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2624 that superficially look like logical namestrings correctly.
2625 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2627 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2628 default supplied-p) for &optional and &key arguments. (thanks to
2631 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2632 * incompatible change: The compiler is now less aggressive about
2633 tail call optimization, doing it only when (> SPACE DEBUG) or
2634 (> SPEED DEBUG). (This is an incompatible change because there are
2635 programs which relied on the old CMU-CL-style behavior to optimize
2636 away their unbounded recursion which will now die of stack overflow.)
2637 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2638 for non-GENCGC systems has been increased to 20M (since that
2639 seems much closer to the likely performance optimum for modern
2640 systems than the old 4M value was)
2641 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2642 and *DEBUG-PRINT-LEVEL*
2643 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2644 of CMU CL's support for SPARC, and various endianness and other
2645 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2646 * new syntactic sugar for the Unix command line: --load foo.bar is now
2647 an alternate notation for --eval '(load "foo.bar")'.
2649 ** The system now detects stack overflow and handles it gracefully,
2650 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2651 optimization settings. (This is a good thing in general, and
2652 its introduction in this version should be particularly timely
2653 for anyone whose code fails because of suppression of tail
2655 ** The system now hunts for the C variable "environ" in a more
2656 devious way, to avoid segfaults when the C library version differs
2657 between compile time and run time. (thanks to Christophe Rhodes)
2658 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2659 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2660 ** The compiler no longer issues bogus style warnings for undefined
2661 classes in the same source file as the DEFCLASSes which defined
2662 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2664 ** fixes in CONDITION class precedence list for undefined function
2665 errors (thanks to Alexei Dejneka)
2666 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2667 correctly. (thanks to Dan Barlow)
2668 ** portability fixes aiming at bootstrapping under CLISP (thanks
2669 to Dave McDonald and Christophe Rhodes)
2670 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2671 ** fixes in type translation and and type inference (thanks to
2673 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2674 ** various fixes in the new ports (thanks to Dan Barlow)
2675 * several changes related to debugging:
2676 ** suppression of tail recursion, as noted above
2677 ** stack overflow detection, as noted above
2678 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2679 is now the default. (For some time encapsulation has been more
2680 reliable than the breakpoint-based :ENCAPSULATE NIL
2681 implementation, at least on X86 systems; and I just noticed that
2682 encapsulation also seems closer to the spirit of the ANSI
2685 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2687 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2688 up properly again. (There was a packaging bug in 0.7.0 which
2689 left their definitions in SB-SYS::LOAD-FOREIGN and
2690 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2691 vital for most things which interface to C-level interfaces,
2692 like extensions working with sockets or databases or
2693 Perl-compatible regexes or whatever, and the need to fix
2694 this bug is the main reason that 0.7.1 was released so
2696 ** DEFGENERIC is now choosier about the methods it redefines, so that
2697 reLOADing a previously-LOADed file containing DEFGENERICs does
2698 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2699 cycle now works as it should. (thanks to Alexey Dejneka)
2700 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2701 fixed by Christophe Rhodes. (He actually submitted this patch
2702 months ago, and I delayed until after 0.7.0.)
2703 ** Bug 111 (internal compiler confusion about runtime checks on
2704 FUNCTION types) was fixed by Alexey Dejneka.
2705 * Some internal cleanups (getting rid of variables which aren't
2706 needed now that the byte interpreter is gone) caused the fasl
2707 file format number to change again.
2709 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2710 * major incompatible change: The default fasl file extension, i.e. the
2711 default extension for files produced by COMPILE-FILE, has changed
2712 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2714 ** There are many changes in the implementation of the compiler.
2715 SBCL is now essentially a compiler-only implementation of ANSI
2716 Common Lisp. EVAL still "interprets" a few special cases, but
2717 almost all the interesting cases are handled by creating
2718 a LAMBDA expression, calling COMPILE on it, then calling
2719 FUNCALL on the result.
2720 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2721 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2722 Since the code is newer, there might still be some new bugs
2723 (though not as many as before Martin Atzmueller's fixes:-). But
2724 the new code is substantially simpler and clearer, and hopefully
2725 any remaining bugs will be simpler, less fundamental, and more
2726 fixable then the bugs in the old code.
2727 ** The revised compiler is still a little unsteady on its feet.
2729 *** The debugging information it produces (particularly the names
2730 of FUNCTION objects) is sometimes much less useful than what
2731 the old compiler produced.
2732 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2733 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2734 within a MACROLET) has been temporarily weakened.
2735 ** There are new compiler optimizations for various functions:
2736 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2737 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2738 *** the math functions TRUNCATE, FLOOR, and CEILING
2739 *** the function-of-all-trades COERCE
2740 Mostly these should be transparent, but there's one
2741 potentially-annoying problem (bug 117): when the compiler
2742 inline-expands a function and does type analysis on the result,
2743 it can create control paths which have type mismatches, and
2744 when it can't prove that those control paths aren't taken,
2745 it will issue WARNINGs about the type mismatches. This is
2746 a particular problem in practice for the new sequence functions.
2747 It's not clear how this should be fixed, and for now, a
2748 workaround is given in the entry for 117 in the BUGS file.
2749 ** (Because of the interaction between the two previous items --
2750 occasional inlining problems and new inline expansions -- some
2751 of the new sequence function optimizations won't really kick in
2752 completely until debugging information, and then inlining, are
2753 straightened out in some future version.)
2754 * minor incompatible changes:
2755 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2756 DIRECTORY no longer implicitly promotes NIL slots of its
2757 pathname argument to :WILD. In particular, when you ask for the
2758 contents of a directory (which you used to be able to do without
2759 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2760 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2761 ** changes in behavior that ANSI explicitly defines to be
2762 implementation dependent:
2763 *** The new compiler-only implementation still conforms with ANSI,
2764 but acts a little different than before. Besides the obvious
2765 changes in performance tradeoffs (that the cost per form passed
2766 to EVAL has gone up, and the cost per form executed by EVAL
2767 has gone down), the behavior of the system changes a little
2768 because there are no longer any interpreted function objects.
2769 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2770 e.g. doing COMPILE on the output of interactive DEFUN is
2772 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2774 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2775 has changed from (:CL) to NIL.
2776 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2777 e.g. #\Newline and #\Tab, don't have names specified in the
2778 ANSI Common Lisp standard, is now based on their ASCII symbolic
2779 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2780 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2781 are no longer used for output.
2782 ** changes in internal implementation constants:
2783 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2784 4 million. (If your application spends a lot of time GCing and
2785 you have a lot of RAM, you might want to experiment with
2786 increasing it even more.)
2787 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2788 However, almost all old code should still continue to work without
2789 immediate update, as SB-C-CALL is now a (deprecated) nickname
2791 ** Old operator names in the style DEF-FOO are now deprecated in
2792 favor of new corresponding names DEFINE-FOO, for consistency with
2793 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2794 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2795 internal symbols, but a few supported extensions like
2796 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2797 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2798 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2799 etc. as you get deeper into recursive calls to the debugger
2800 command loop, instead of the old "5]", "5]]", "5]]]"
2801 sequence. (I was motivated to do this when squabbles between
2802 ILISP and SBCL left me very deeply nested in the debugger. In the
2803 short term, this change will probably provoke more ILISP/SBCL
2804 squabbles, but hopefully it will be an improvement in the long run.)
2805 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2806 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2807 function, since declaiming return types involving aliens is
2808 (1) annoyingly messy to do by hand and (2) vital to efficient
2809 compilation of code which calls such functions.
2810 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2811 longer reexported by the SB-EXT package. They're solely useful
2812 for alien code, so it seems more logical that you should get
2813 them from the SB-ALIEN package, not in SB-EXT.
2814 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2815 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2816 features. Instead, the code that they used to control is always
2817 built into the system.
2818 * many other bug fixes
2819 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2820 advantage of the new EVAL-WHEN stuff and to clean them up in
2821 general, and they are now more ANSI-compliant in a number of
2822 ways. Martin Atzmueller is responsible for a lot of this.
2823 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2825 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2826 *** correct ERROR type for various file operations
2827 *** some fixes for Lisp streams
2828 *** DEFMETHOD syntax checking
2829 *** changing old weird representation of debug information as
2830 strings (which, among their other deficiencies, don't transform
2831 correctly when you rename packages, and don't change their
2832 print representation when you change things like *PACKAGE*
2833 and *PRINT-LENGTH*) to symbols and lists of symbols
2834 He also made several improvements and fixed several bugs in DESCRIBE.
2835 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2837 *** misbehavior of WRITE-STRING/WRITE-LINE
2838 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2839 and several other LOOP problems as well
2840 *** DIRECTORY when similar filenames are present
2841 *** DEFGENERIC with :METHOD options
2842 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2843 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2844 *** argument ordering in FIND with :TEST option
2845 *** mishandled package designator argument in APROPOS-LIST
2846 *** various problems in the backquote readmacro
2847 *** a bug in APROPOS
2848 *** probably some others that I'm not describing very well here,
2849 since the CVS log documents them by reference to sbcl-devel
2850 messages, and the SourceForge archives aren't working well.:-(
2851 ** Dan Barlow improved the Alpha port (and is making progress on the
2852 PPC port, for those of you who think different).
2853 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2854 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2855 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2856 continues to work on the SPARC port (for those of you in a position
2857 to look down upon our little PC-compatible boxes from a great height).
2858 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2859 and uses that copy, rather than the current dynamic value, when
2860 it's trying to decide whether to truncate output. Thus e.g.
2861 (let ((*print-lines* 50))
2862 (pprint-logical-block (stream nil)
2864 (let ((*print-lines* 8))
2865 (print (aref possiblybigthings i) stream)))))
2866 should now truncate the logical block only at 50 lines, instead of
2867 often truncating it at 8 lines, as it did before.
2868 * The doc/cmucl/ directory, containing old CMU CL documentation
2869 from the time of the fork, is no longer part of the base system.
2870 SourceForge has shut down its anonymous FTP service, and with it
2871 my original plan for distributing the old CMU CL documentation
2872 there. For now, if you need these files you can download an old
2873 SBCL source release and extract them from it.
2874 * The fasl file version number changed again, for dozens of reasons,
2875 some of which are apparent above.
2877 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2878 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2879 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2880 patch, so that DISASSEMBLE output is much nicer.
2881 * The code in the SB-PROFILE package now seems reasonably stable.
2882 I still haven't decided what the final interface should look like
2883 (I'd like PROFILE to interact cleanly with TRACE, since both
2884 facilities use function encapsulation) but if you have a need
2885 for profiling now, you can probably use it successfully with
2886 the current CMU-CL-style interface.
2887 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2888 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2889 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2890 link now returns the dangling link itself, and for similar
2891 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2892 link itself. (In these cases the old code signalled an error and
2893 looped endlessly, respectively.) Thus, DIRECTORY now works even
2894 in the presence of dangling and cyclic symbolic links.
2895 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2896 is now a supported extension again, since the consensus on
2897 sbcl-devel was that it can be useful for ordinary development
2898 work, not just for debugging SBCL itself.
2899 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2900 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2901 that currently-defined functions might be redefined later with
2902 different return types.
2903 * Hash tables can be printed readably, as inspired by CMU CL code
2904 of Eric Marsden and SBCL code of Martin Atzmueller.
2905 * better error handling in CLOS method combination, thanks to
2906 Martin Atzmueller porting Pierre Mai's CMU CL patches
2907 * more overflow fixes for >16Mbyte I/O buffers
2908 * A bug in READ has been fixed, so that now a single Ctrl-D
2909 character suffices to cause end-of-file on character streams.
2910 In particular, now you only need one Ctrl-D at the command
2911 line (not two) to exit SBCL.
2912 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2914 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2915 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2916 does the right thing.
2917 * The implementation of some type tests, especially for CONDITION
2918 types, is now tidier and maybe faster, due to CMU CL code
2919 originally by Douglas Crosher, ported by Martin Atzmueller.
2920 * Some math functions have been fixed, and there are new
2921 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2922 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2923 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2924 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2925 likely to be built into sbcl-0.7.0.)
2926 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2927 CL, ported to SBCL by Martin Atzmueller)
2928 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2929 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2930 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2931 * There's a new slam.sh hack to shorten the edit/compile/debug
2932 cycle for low-level changes to SBCL itself, and a new
2933 :SB-AFTER-XC-CORE target feature to control the generation of
2934 the after-xc.core file needed by slam.sh.
2935 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2936 COMPILE-FILE is no longer supported, so that now every function
2937 gets an entry point, so that block compilation looks a little
2938 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2939 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2940 returns the number of bytes consed since the system started,
2941 rather than the number consed since the first time the function
2942 was called. (The new definition parallels ANSI functions like
2943 CL:GET-INTERNAL-RUN-TIME.)
2944 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2945 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2946 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2947 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2948 * Fasl file version numbers are now independent of the target CPU,
2949 since historically most system changes which required version
2950 number changes have affected all CPUs equally. Similarly,
2951 the byte fasl file version is now equal to the ordinary
2954 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2955 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2956 is no longer recognized. I apologize for this, because it was
2957 listed in SB-EXT as a supported extension, but I found that
2958 its existing behavior was poorly specified, as well as incorrectly
2959 specified, and it looked like too much of a mess to straighten it
2960 out. I have enough on my hands trying to get ANSI stuff to work..
2961 * many patches ported from CMU CL by Martin Atzmueller, with
2962 half a dozen bug fixes in pretty-printing and the debugger, and
2963 half a dozen others elsewhere
2964 * fixed bug 13: Floating point infinities are now supported again.
2965 They might still be a little bit flaky, but thanks to bug reports
2966 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2967 as flaky as they were.
2968 * The --noprogrammer command line option is now supported. (Its
2969 behavior is slightly different in detail from what the old man
2970 page claimed it would do, but it's still appropriate under the
2971 same circumstances that the man page talks about.)
2972 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2973 are now supported, and enabled by default. Thus, the compiler can
2974 handle many floating point and complex operations much less
2975 inefficiently. (Thus e.g. you can implement a complex FFT
2977 * The compiler now detects type mismatches between DECLAIM FTYPE
2978 and DEFUN better, and implements CHECK-TYPE more correctly, and
2979 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2980 thanks to patches from Martin Atzmueller.
2981 * various fixes to make the cross-compiler more portable to
2982 ANSI-conforming-but-different cross-compilation hosts (notably
2983 Lispworks for Windows, following bug reports from Arthur Lemmens)
2984 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2985 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2986 fixed thanks to Pierre Mai's CMU CL patches.
2987 * fixes to keep the system from overflowing internal counters when
2988 it tries to use i/o buffers larger than 16M bytes
2989 * fixed bug 45a: Various internal functions required to support
2990 complex special functions have been merged from CMU CL sources.
2991 (When I was first setting up SBCL, I misunderstood a compile-time
2992 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2993 * improved support for type intersection and union, fixing bug 12
2994 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2995 more obscure bugs as well
2996 * some steps toward byte-compiling non-performance-critical
2997 parts of the system, courtesy of patches from Martin Atzmueller
2998 * Christophe Rhodes has made some debian packages of sbcl at
2999 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3000 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3001 stable, but are nonetheless usable. When he's ready, I'd be happy
3002 to add them to the SourceForge "File Releases" section. (And if
3003 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3004 * new fasl file format version number (because of changes in
3005 internal representation of (OR ..) types to accommodate the new
3006 support for (AND ..) types, among other things)
3008 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3009 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3011 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3013 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3014 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3015 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3016 * DESCRIBE now works on structure objects again.
3017 * Most function call argument type mismatches are now handled as
3018 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3019 know whether the function will be redefined before the call is
3020 executed. (The compiler could flag local calls with full WARNINGs,
3021 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3022 it doesn't keep track of enough information to know whether calls
3023 are local in this sense.)
3024 * Compiler output is now more verbose, with messages truncated
3025 later than before. (There should be some supported way for users
3026 to override the default verbosity, but I haven't decided how to
3027 provide it yet, so this behavior is still controlled by the internal
3028 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3029 src/compiler/ir1util.lisp.)
3030 * Fasl file format version numbers have increased again, because
3031 support for the Gray streams extension changes the layout of the
3032 system's STREAM objects.
3033 * The Gray subclassable streams extension now works, thanks to a
3034 patch from Martin Atzmueller.
3035 * The full LOAD-FOREIGN extension (not just the primitive
3036 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3037 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3038 but like most other programs, it defaults to copying the Unix
3039 environment from the original process instead of starting the
3040 new process in an empty environment.
3041 * Extensions which manipulate the Unix environment now support
3042 an :ENVIRONMENT keyword option which doesn't smash case or
3043 do other bad things. The CMU-CL-style :ENV option is retained
3044 for porting convenience.
3045 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3046 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3048 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3050 * A patch from Martin Atzmueller seems to have solved the SIGINT
3051 problem, and as far as we know, signal-handling now works cleanly.
3052 (If you find any new bugs, please report them!)
3053 * The system no longer defaults Lisp source file names to types
3054 ".l", ".cl", or ".lsp", but only to ".lisp".
3055 * The compiler no longer uses special default file extensions for
3056 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3057 seems to expect a single default extension for all compiled code,
3058 and there's no compelling reason to try to stretch the standard
3059 to allow two different extensions.) Instead, byte-compiled files
3060 default to the same extension as native-compiled files.
3061 * Fasl file format version numbers have increased again, because
3062 a rearrangement of internal implementation packages made some
3063 dumped symbols in old fasl files unreadable in new cores.
3064 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3065 many fewer weird special cases.
3066 * Bug #17 (differing COMPILE-FILE behavior between logical and
3067 physical pathnames) has been fixed, and some related misbehavior too,
3068 thanks to a patch from Martin Atzmueller.
3069 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3070 by Tim Moore, ported to SBCL by Martin Atzmueller.
3071 * Martin Atzmueller fixed several filesystem-related problems,
3072 including bug #36, in part by porting CMU CL patches, which were
3073 written in part by Paul Werkowski.
3074 * More compiler warnings in src/runtime/ are gone, thanks to
3075 more patches from Martin Atzmueller.
3076 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3079 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3081 * DESCRIBE now works on CONDITION objects.
3082 * The debugger now handles errors which arise when trying to print
3083 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3085 * The build system now uses an additional file, customize-target-features.lisp,
3086 to allow local modifications to the target *FEATURES* list. (The point of
3087 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3088 debugging features enabled, without having to worry about propagating it
3089 into everyone's system when I do a "cvs update".) When no
3090 customize-target-features.lisp file exists, the target *FEATURES* list
3091 should be constructed the same way as before.
3092 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3093 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3094 warn about behavior which is undefined under the ANSI spec. Note: This
3095 is specified by ANSI, but it's not very popular with programmers.
3096 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3097 macro in the SBCL sources for an example of a workaround which you
3098 might use to make portable ANSI-standard code which does what you want.
3099 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3100 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3101 when it's used as a non-toplevel form. (This is required in order
3102 to implement the DEFCONSTANT-EQX macro.)
3103 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3104 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3105 funny things, due to bugs in the implementation of EVAL-WHEN
3106 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3107 non-toplevel DEFCONSTANTs are uncommon.)
3108 * The core file version number and fasl file version number have been
3109 incremented, because the old noncompliant DEFCONSTANT behavior involved
3110 calling functions which no longer exist, and because I also took the
3111 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3112 * fixed bug 1 (error handling before read-eval-print loop starts), and
3113 redid debugger restarts and related debugger commands somewhat while
3115 ** The QUIT debugger command is gone, since it did something
3116 rather different than the SB-EXT:QUIT command, and since it never
3117 worked properly outside the main toplevel read/eval/print loop.
3118 Invoking the new TOPLEVEL restart provides the same functionality.
3119 ** The GO debugger command is also gone, since you can just invoke
3120 the CONTINUE restart directly instead.
3121 ** The TOP debugger command is also gone, since it's redundant with the
3122 FRAME 0 command, and since it interfered with abbreviations for the
3124 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3126 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3127 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3128 some porting work by Martin Atzmueller)
3129 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3130 with a boatload of patches which he ported to SBCL. Now that those
3132 ** The system tries to make sure that its low-priority messages
3133 are prefixed by semicolons, to help people who like to use
3134 syntax highlighting in their ILISP buffer. (This patch
3135 was originally due to Raymond Toy.)
3136 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3137 patches originally written by Raymond Toy.
3138 ** The compiler understands coercion between single-value and
3139 multiple-VALUES type expressions better, getting rid of some very
3140 weird behavior, thanks to patches originally by Robert MacLachlan
3141 and Douglas Crosher.
3142 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3143 lambda lists, thanks to a patch originally by Pierre Mai.
3144 ** The system no longer bogusly warns about "abbreviated type
3146 ** The compiler gets less confused by inlining and RETURN-FROM,
3147 thanks to some patches originally by Tim Moore.
3148 ** The system no longer hangs when dumping circular lists to fasl
3149 files, thanks to a patch originally from Douglas Crosher.
3150 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3151 undefined function error.
3152 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3153 in structure slot initforms) for now, documented workaround instead:-|
3154 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3155 * fixed bug 5: added stubs for various Gray stream functions called
3156 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3157 installed, at least appropriate type errors are generated
3158 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3159 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3160 consistently in DEFMETHOD forms.
3161 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3162 it doesn't seem to affect SBCL after all
3163 * The C runtime system now builds with better optimization and many
3164 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3166 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3168 * The system is now under CVS at SourceForge (instead of the
3169 CVS repository on my home machine).
3170 * The new signal handling code has been tweaked to treat register
3171 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3172 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3173 downstream, e.g. in debug-int.lisp, has implicit dependencies
3174 on the unsignedness of integer representation of machine words,
3175 and that caused the system to bomb out with infinite regress
3176 when trying to recover from type errors involving signed values,
3177 e.g. (BUTLAST '(1 2 3) -1).)
3178 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3179 (This was one of the bugs Peter Van Eynde reported back in July.)
3180 * The system now uses code inspired by Colin Walters' O(N)
3181 implementation of MAP (from the cmucl-imp@cons.org mailing
3182 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3183 inline the MAP operation, and there is more than one
3184 sequence argument to the MAP call (so that it can't just
3185 do ETYPECASE once and for all based on the type of the
3186 single sequence argument). (The old non-inline implementation
3187 of the general M-argument sequence-of-length-N case required
3188 O(M*N*N) time when any of the sequence arguments were LISTs.)
3189 * The QUIT :UNIX-CODE keyword argument has been renamed to
3190 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3191 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3193 * Raymond Wiker's patches to port dynamic loading from Linux to
3194 FreeBSD have been added.
3195 * The BUGS file is now more nearly up to date, thanks in large part
3196 to Martin Atzmueller's review of it.
3197 * The debugger now flushes standard output streams before it begins
3198 its output ("debugger invoked" and so forth).
3199 * The core version number and fasl file version number have both
3200 been incremented, because of incompatible changes in the layout
3202 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3203 used to not be called for a saved Lisp image.)
3204 * Martin Atzmueller's version of a patch to fix a compiler crash,
3205 as posted on sbcl-devel 13 September 2000, has been installed.
3206 * Instead of installing Martin Atzmueller's patch for the
3207 compiler transform for SUBSEQ, I deleted the compiler transform,
3208 and transforms for some similar consing operations.
3209 * A bug in signal handling which kept TRACE from working on OpenBSD
3211 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3212 away by constant folding
3213 * The system now defines its address space constants in one place
3214 (in the Lisp sources), and propagates them automatically elsewhere
3215 (through GENESIS and the sbcl.h file). Therefore, patching the
3216 address map is less unnecessarily tedious and error-prone. The
3217 Lisp names of address space constants have also been systematized.
3218 * CVS tags like dollar-Header-dollar have been removed from
3219 the sources, because they have never saved me trouble and
3220 they've been source of trouble working with patches and other
3221 diff-related operations.
3222 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3223 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3225 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3227 * The system has been ported to OpenBSD.
3228 * The system now compiles with a simple "sh make.sh" on the systems
3229 that it's supported on. I.e., now you no longer need to tweak
3230 text in the target-features.lisp-expr and symlinks in src/runtime/
3231 by hand, the make.sh takes care of it for you.
3232 * The system is no longer so grossly inefficient when compiling code
3233 involving vectors implemented as general (not simple) vectors (VECTOR T),
3234 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3235 which dares to use the various sequence functions on non-simple
3236 vectors, takes less of a performance hit.
3237 * There is now a primitive type predicate VECTOR-T-P
3238 to test for the (VECTOR T) type, so that e.g.
3239 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3240 can now be compiled with some semblance of efficiency. (The old code
3241 turned the type declaration into a full call to %TYPEP at runtime!)
3242 * AREF on (VECTOR T) is still not fast, since it's still compiled
3243 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3244 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3245 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3246 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3247 without making full calls to SUBTYPEP at runtime.
3248 (Some analogous efficiency issues for non-simple vectors specialized to
3249 element types other than T, or for non-simple multidimensional arrays,
3250 have not been addressed. They could almost certainly be handled the
3251 same way if anyone is motivated to do so.)
3252 * The changes in array handling break binary compatibility, so
3253 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3254 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3255 instead of (VALUES T T).
3256 * By following the instructions that Dan Barlow posted to sbcl-devel
3257 on 2 July 2000, I was able to enable primitive dynamic object
3258 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3259 functionality is not implemented (since it calls ld to resolve
3260 library references automatically, requiring RUN-PROGRAM for its
3261 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3262 try to resolve library references) is now supported.
3263 * The system now flushes the standard output streams when it terminates,
3264 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3265 them at several other probably-convenient times, e.g. in each pass of
3266 the toplevel read-eval-print loop, and after evaluating a form given
3267 as an "--eval" command-line option. (These changes were motivated by a
3268 discussion of stream flushing issues on cmucl-imp in August 2000.)
3269 * The source transform for TYPEP of array types no longer assumes
3270 that an array whose element type is a not-yet-defined type
3271 is implemented as an array of T, but instead punts, so that the
3272 type will be interpreted at runtime.
3273 * There is now some support for cross-compiling in make.sh: each of
3274 the phases of make.sh has its own script. (This should be transparent
3275 to people doing ordinary, non-cross-compile builds.)
3276 * Since my laptop doesn't have hundreds of megabytes of memory like
3277 my desktop machine, I became more motivated to do some items on
3278 my to-do list in order to reduce the size of the system a little:
3279 ** Arrange for various needed-only-at-cold-init things to be
3280 uninterned after cold init. To support this, those things have
3281 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3282 symbols with such names are now uninterned after cold init).
3283 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3284 fasl files for cold load.
3285 ** Remove the old compiler structure pooling code (which used to
3286 be conditional on the target feature :SB-ALLOC) completely.
3287 ** Redo the representation of some data in cold init to be more compact.
3288 (I also looked into supporting byte compiled code at bootstrap time,
3289 which would probably reduce the size of the system a lot, but that
3290 looked too complicated, so I punted for now.)
3291 * The maximum signal nesting depth in the src/runtime/ support code has
3292 been reduced from 4096 to 256. (I don't know any reason for the very
3293 large old value. If the new smaller value turns out to break something,
3294 I'll probably just bump it back up.)
3295 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3297 * Many, many bugs reported by Peter Van Eynde have been added to
3298 the BUGS list; some have even been fixed.
3299 * While enabling dynamic object file loading, I tried to make the
3300 code easier to understand, renaming various functions and variables
3301 with less ambiguous names, and changing some function calling
3302 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3303 * While trying to figure out how to do the OpenBSD port, I tried to
3304 clean up some of the code in src/runtime/. In particular, I dropped
3305 support for non-POSIX signal handling, added various comments,
3306 tweaked the code to reduce the number of compilation warnings, and
3307 renamed some files to increase consistency.
3308 * To support the new automatic configuration functionality in make.sh,
3309 the source file target-features.lisp-expr has been replaced with the
3310 source file base-target-features.lisp-expr and the machine-generated
3311 file local-target-features.lisp-expr.
3312 * fixed a stupid quoting error in make.sh so that using CMU CL
3313 "lisp -batch" as cross-compilation host works again
3315 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3317 * DESCRIBE no longer tries to call itself recursively to describe
3318 bound/fbound values, so that it no longer fails on symbols which are
3319 bound to themselves (like keywords, T, and NIL).
3320 * DESCRIBE now works on generic functions.
3321 * The printer now prints less-screwed-up representations of closures
3322 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3323 * A private symbol is used instead of the :EMPTY keyword previously
3324 used to mark empty slots in hash tables. Thus
3325 (DEFVAR *HT* (MAKE-HASH-TABLE))
3326 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3327 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3328 now does what ANSI says that it should. (You can still get
3329 similar noncompliant behavior if bang on the hash table
3330 implementation with all the symbols you get back from
3331 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3332 This breaks binary compatibility, since tests for equality to
3333 :EMPTY are wired into things like the macroexpansion of
3334 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3336 * There's now a minimal placeholder implementation for CL:STEP,
3337 as required by ANSI.
3338 * An obscure bug in the interaction of the normal compiler, the byte
3339 compiler, inlining, and structure predicates has been patched
3340 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3341 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3343 * Missing ordinary arguments in a macro call are now detected even
3344 when the macro lambda list contains &KEY or &REST.
3345 * The debugger no longer complains about encountering the top of the
3346 stack when you type "FRAME 0" to explicitly instruct it to go to
3347 the top of the stack. And it now prints the frame you request even
3348 if it's the current frame (instead of saying "You are here.").
3349 * As specified by ANSI, the system now always prints keywords
3350 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3352 * The default initial SIZE of HASH-TABLEs is now smaller.
3353 * Type information from CLOS class dispatch is now propagated
3354 into DEFMETHOD bodies, so that e.g.
3355 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3357 is now basically equivalent to
3358 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3359 (DECLARE (TYPE SINGLE-FLOAT X))
3361 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3362 operation, without having to do run-time type dispatch.
3363 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3364 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3365 * Since I don't seem to be making big file reorganizations very often
3366 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3367 is overflowing my ability to conveniently back them up), I've finally
3368 checked the system into CVS. (The CVS repository is on my home system,
3369 not at SourceForge -- putting it on SourceForge might come later.)
3370 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3371 high-level GC-NOTIFY-FOO functions send their output. (There's
3372 still very little control of where low-level verbose GC functions
3373 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3374 less than it used to -- the GC-NOTIFY-FOO functions are now under
3375 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3376 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3377 in only one place in the source code, and propagates it automatically
3378 everywhere that it's needed. Thus e.g. when I bump the version from
3379 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3380 * The C source files now include boilerplate legalese and documentation
3381 at the head of each file (just as the Lisp source files already did).
3382 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3383 to his page will be replaced with a link to his new CLiki service.
3385 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3387 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3388 * The build process now looks for GNU make under the default name "gmake",
3389 instead of "make" as it used to. If GNU make is not available as "gmake"
3390 on your system, you can change this default behavior by setting the
3391 GNUMAKE environment variable.
3392 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3393 can build without error under CMU CL.
3395 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3397 * There is now a partial SBCL user manual (with some new text and some
3398 text cribbed from the CMU CL manual).
3399 * The beginnings of a profiler have been added (starting with the
3400 CMU CL profiler and simplifying and cleaning up). Eventually the
3401 main interface should be through the TRACE macro, but for now,
3402 it's still accessed through vaguely CMU-CL-style functions and macros
3403 exported from the package SB-PROFILE.
3404 * Some problems left over from porting CMU CL to the new
3405 cross-compilation bootstrap process have been cleaned up:
3406 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3407 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3408 part when getting rid of the compiler *BACKEND* stuff.)
3409 ** The value of *NULL-TYPE* was screwed up, because it was
3410 being initialized before the type system knew the final
3411 definition of the 'NULL type. This screwed up several key
3412 optimizations in the compiler, causing inefficiency in all sorts
3413 of places. (I found it because I wanted to understand why
3414 GET-INTERNAL-RUN-TIME was consing.)
3415 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3416 PROCLAIM FTYPE information. Unfortunately this broke binary
3417 compatibility again, since now the forms output by DEFGENERIC
3418 to refer to functions which didn't exist in 0.6.3.
3419 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3420 can use the new (as of 0.6.3) transform for SOME into MAP into
3422 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3423 values to (INTEGER 0 1000000), so that the time code will no longer
3424 occasionally get blown up by Linux returning 1000000 microseconds
3425 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3426 its output conform to the ANSI spec. (Alas, this makes its output
3427 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3428 * A full call to MAP NIL with a single sequence argument no longer conses.
3429 * fixes to problems pointed out by Martin Atzmueller:
3430 * The manual page no longer talks about multiprocessing as though
3431 it were currently supported.
3432 * The ILISP support patches have been removed from the distribution,
3433 because as of version 5.10.1, ILISP now supports SBCL without us
3434 having to maintain patches.
3435 * added a modified version of Raymond Toy's recent CMU CL patch for
3436 EQUALP comparison of HASH-TABLE
3438 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3440 * The system still can't cross-compile itself with
3441 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3442 entails), but at least it can compile more of itself that way
3443 than it used to be able to, and various buglets which were uncovered
3444 by trying to cross-compile itself that way have now been fixed.
3445 * This release breaks binary compatibility again. This time
3446 at least I've incremented the FASL file format version to 2, so that the
3447 problem can be detected reliably instead of just causing weird errors.
3448 * various new style warnings:
3449 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3450 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3451 ** using the lexical binding of a variable named in the *FOO* style
3452 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3453 as specified by ANSI.
3454 * *RANDOM-STATE* is no longer automatically initialized from
3455 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3456 default behavior of the system is to repeat its behavior every time
3457 it's run. If you'd like to change this behavior, you can always
3458 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3459 old convention there was no comparably easy way to get the system to
3460 repeat its behavior every time it was run.
3461 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3462 FTYPE declarations has been removed, in favor of their ANSI
3463 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3464 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3465 the types of their sequence arguments can be determined at compile time.
3466 This is done through a new open code expansion for MAP which eliminates
3467 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3468 argument types can be determined at compile time.
3469 * The optimizer now transforms COERCE into an identity operation when it
3470 can prove that the coerced object is already of the correct type. (This
3471 can be a win for machine generated code, including the output of other
3472 optimization transforms, such as the MAP transform above.)
3473 * Credit information has been moved from source file headers into CREDITS.
3474 * Source file headers have been made more standard.
3475 * The CASE macro now compiles without complaining even when it has
3478 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3480 * (Note that the way that the PCL macroexpansions were rewritten
3481 to accommodate the change in DEFGENERIC below breaks binary
3482 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3483 not run under sbcl-0.6.2. Once we get out of alpha releases,
3484 i.e. hit release 1.0.0, we'll probably try to maintain binary
3485 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3486 and sbcl-1.4.4. Until then, however, it might be fairly common
3487 for maintenance releases to break binary compatibility.)
3488 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3490 * The reporting of unhandled signals has been changed to print some
3491 explanatory text as well as the report form. (Previously only
3492 the report form was printed.)
3493 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3494 it defines, so that the compiler no longer issues undefined function
3495 warnings for compiled-but-not-yet-loaded generic functions.
3496 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3497 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3498 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3499 section of the ANSI spec are supported.
3500 * The "" nickname for the "KEYWORD" package has been removed.
3501 The reader still handles symbol tokens which begin with a package marker
3502 as keywords, but it doesn't expose its mechanism for doing so in the
3503 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3504 * The system now issues STYLE-WARNINGs for contradictory TYPE
3505 proclamations. (Warnings for contradictory FTYPE proclamations would
3506 be nice too, but those can't be done usefully unless the type system
3507 is made smarter about FUNCTION types.)
3508 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3509 systematized, so that "*target-*.lisp is supposed to exist only on the
3510 target and imply that there's a related file which exists on the
3511 host, and *host-*.lisp is supposed to exist only on the host and imply
3512 that there's a related file which exists on the target. This involves a
3513 lot of renaming. Hopefully the acute confusion caused by the renaming
3514 will be justified by the reduction in chronic confusion..
3515 ** runtime-type.lisp -> early-target-type.lisp
3516 ** target-type.lisp -> late-target-type.lisp
3517 ** early-host-format.lisp -> early-format.lisp
3518 ** late-host-format.lisp -> late-format.lisp
3519 ** host-error.lisp -> misc-error.lisp
3520 ** early-error.lisp -> early-target-error.lisp
3521 ** late-error.lisp -> late-target-error.lisp
3522 ** host-defboot.lisp -> early-defboot.lisp
3523 ** code/misc.lisp -> code/target-misc.lisp
3524 ** code/host-misc.lisp -> code/misc.lisp
3525 ** code/numbers.lisp -> code/target-numbers.lisp
3526 ** code/early-numbers.lisp -> numbers.lisp
3527 ** early-host-type.lisp -> early-type.lisp
3528 ** late-host-type.lisp -> late-type.lisp
3529 ** host-typep.lisp -> typep.lisp
3530 ** load.lisp -> target-load.lisp
3531 ** host-load.lisp -> load.lisp
3532 ** host-disassem.lisp -> disassem.lisp
3533 ** host-insts.lisp -> insts.lisp
3534 ** byte-comp.lisp -> target-byte-comp.lisp
3535 ** host-byte-comp.lisp -> byte-comp.lisp
3536 ** host-signal.lisp -> signal.lisp
3537 ** host-defstruct.lisp -> defstruct.lisp
3538 ** late-target-type.lisp -> deftypes-for-target.lisp
3539 Furthermore, several other previously target-only files foo.lisp (e.g.
3540 hash-table.lisp and random.lisp) have been split into a target-and-host
3541 foo.lisp file and a target-only target-foo.lisp file, with their key type
3542 definitions in the target-and-host part, so that the cross-compiler will
3543 know more about target types.
3544 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3545 gone away. In their place are various *BACKEND-FOO* variables
3546 corresponding to the slots of the old structure.
3547 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3548 into the target SBCL has been fixed.
3549 * The chill.lisp system for loading cold code into a running SBCL
3551 * Support for the CMU CL "scavenger hook" extension has been removed.
3552 (It was undocumented and unused in the CMU CL sources that SBCL was
3553 derived from, and stale in sbcl-0.6.1.)
3554 * Various errors in the cross-compiler type system were detected
3555 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3556 (enabling various consistency checks). Many of them were fixed,
3557 but some hard problems remain, so the compiler is back to
3558 running without *TYPE-SYSTEM-INITIALIZED* for now.
3559 * As part of the cross-compiler type system cleanup, I implemented
3560 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3561 * I have started adding UNCROSS calls throughout the type system
3562 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3563 on cross-compiler input files will be able to go away, and various
3565 * CONSTANTP now returns true for quoted forms (as explicitly required
3568 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3570 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3571 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3572 (MAKE-LIST -1). In the longer term, it ought to become true
3573 that declarations are assertions even at SAFETY 1. For now, it's not
3574 quite true even at SAFETY 3, but it's at least more nearly true..
3575 (Note that this change seems to increases the size of the system by
3576 O(5%) and to decrease the speed of the compiler by 20% or more.)
3577 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3578 for the problem of printing dozens of lines of distracting information
3579 about low-level system machinery as part of the top stack frame
3580 on entry to the debugger when an undefined function was called.
3581 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3582 is not reset to COMMON-LISP-USER.
3583 * Compilation of stuff related to dyncount.lisp has been made conditional
3584 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3585 smaller. The various dyncount-related symbols have been moved into
3586 a new "SB-DYNCOUNT" package.
3587 * tty-inspect.lisp has been renamed to inspect.lisp.
3588 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3589 feature has gone away. (When we eventually port to other flavors of
3590 libc and/or Unix, we'll try to make the differences between flavors
3591 invisible at the user level.)
3592 * Various other *FEATURES* tags, and/or their associated conditionals,
3593 have been removed if obsolescent, or given better documentation, or
3594 sometimes given more-mnemonic names.
3596 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3598 * tidied up "make.sh" script
3599 * tidied up system directory structure
3600 * better "clean.sh" behavior
3601 * added doc/FOR-CMUCL-DEVELOPERS
3602 * many many small tweaks to output format, e.g. removing possibly-confusing
3603 trailing #\. character in DESCRIBE-INSTANCE
3604 * (EQUALP #\A 'A) no longer signals an error.
3605 * new hashing code, including EQUALP hashing
3606 * tidied up Lisp initialization and toplevel
3607 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3608 * command line argument processing
3609 * added POSIX-GETENV function to deal with Unix-ish environment variables
3610 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3611 terminating SBCL on EOF
3612 * non-verbose GC by default
3613 * There is no more "sbcl" shell script; the sbcl file is now the C
3614 runtime executable (just like CMU CL).
3615 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3617 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3618 symbol and package information as Lisp native symbol and package objects
3619 instead of strings naming symbols and strings naming packages. This way,
3620 whenever packages are renamed (as in warm init), debug information is
3621 transformed along with everything else.
3622 * tweaked the optimization policy declarations which control the building
3623 of SBCL itself. Now, among other things, the system no longer saves
3624 source location debugging information. (This helps two problems at once
3625 by reducing SBCL size and by keeping SBCL from trying to look for its
3626 sources -- which may not exist -- when reporting errors.)
3627 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3628 debugging and testing purposes
3629 * cleaned up printing, making the printer call PRINT-OBJECT for
3630 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3631 methods, giving nearly-ANSI behavior
3632 * converted almost all special variables to use *FOO* naming convention
3633 * deleted PARSE-TIME functionality, since it can be done portably
3634 * moved some files out of cold init into warm init
3635 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3637 * regularized formatting of source files
3638 * added an install.sh script
3639 * fixed ridiculous memory usage of cross-compiler by making
3640 compiler/alloc.lisp not try to do pooling unless it can hook
3641 itself into the GC of the cross-compilation host. Now the system
3642 builds nicely on my old laptop.
3643 * added :SB-ALLOC in target-features.lisp-expr
3644 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3645 was not implemented)
3646 * re-did condition handling and note reporting in the compiler. Notes
3647 are no longer handled by signalling conditions. Style warnings
3648 and warnings are handled more correctly and reported in such a way
3649 that it's easy to find one or the other in your output (so that you
3650 can e.g. figure out which of many problems caused COMPILE-FILE to
3652 * changed the severity of several compiler warnings from full WARNING
3653 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3654 compiler note reporting so that it doesn't use the condition system
3655 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3656 COMPILE-FILE command)
3657 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3658 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3659 no longer supported, since their implementation was deeply intertwingled
3660 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3661 * removed bogus "support" for compiler macros named (SETF FOO), and
3662 removed the compiler macro for SETF INFO (but only after making a fool
3663 of myself on the cmucl-imp mailing list by posting a bogus patch for
3664 DEFINE-COMPILER-MACRO..)
3665 * Compiled files containing forms which have side effects on the Lisp
3666 reader (such as DEFPACKAGE forms) are now handled more correctly.
3667 (Compiler queuing of top level lambdas has been suppressed by setting
3668 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3669 * deleted various currently-unused source files, e.g. gengc.lisp. They
3670 may be added back at some point e.g. when porting to other architectures,
3671 but until they are it's distracting to distribute them and to try to
3673 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3674 there were so many of them they're just distractions, and UNCROSS is
3675 known to be able to handle the current sources
3676 * moved PROFILE functionality into TRACE, so that it will be clear
3677 how the wrapping and unwrapping of functions when you profile them
3678 interacts with the wrapping and unwrapping of functions when you
3679 trace them. (Actually, the functionality isn't there yet, but at least
3680 the interface specification is there. Hopefully, the functionality will
3681 arrive with some future maintenance release.)
3682 * removed host-oops.lisp
3683 * changed signature of QUIT function to allow UNIX-CODE argument
3684 * fixed READ-SEQUENCE bug
3685 * tweaked verbose GC output so that it looks more like the progress
3686 output that ANSI specifies for functions like LOAD
3687 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3688 * added <http://sbcl.sourceforge.com> to the banner information printed by