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.
7 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
8 * new platform: experimental support for SBCL x86/Darwin, including
9 MacOS X 10.4.5 on Intel.
10 * new feature: Unicode character names are now known to the system
11 (through CHAR-NAME and NAME-CHAR).
12 * new feature: the filesystem location of SBCL's core file is
13 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
14 * minor incompatible change: the contrib modules SB-POSIX and
15 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
16 this change is to make it easier to distribute
17 location-independent binaries.
18 * bug fix: as implied by AMOP, standardized classes no longer have
19 slots named by external symbols of public packages. (reported by
21 * optimization: calling functions via constant symbols -- as in
22 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
23 via the function object as in (FUNCALL #'FOO).
24 * optimization: on x86 and x86-64, the calling convention for the
25 case of dealing with an unknown number of values has been altered
26 to be friendlier to the prediction heuristics implemented,
27 particularly with reference to CALL and RET pairing. (thanks to
29 * optimization: on x86, the code for access to thread-local symbol
30 values and binding/unbinding of thread-local symbols is smaller.
31 (thanks to Alastair Bridgewater)
32 * enhancement: CONSTANTP is now able to determine constantness of
33 more complex forms, including calls to constant-foldable standardized
34 functions and some special forms beyond QUOTE.
35 * fixed bug: occasional GC crashes on Solaris/x86
36 * optimization: x86-64 supports stack allocation of results of simple
37 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
38 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
39 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
40 (thanks to James Knight)
41 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
42 by the standard (thanks to Stephen Compall)
44 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
45 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
46 be used for bundling the runtime and the core file into one
47 executable binary. This feature is not currently supported on all SBCL
48 platforms. (thanks to James Bielman and NIIMI Satoshi)
49 * new feature: a generational or ephemeral garbage collector is now
50 the default on the PowerPC platform (both Linux and Darwin). The
51 old Cheney (stop and copy) collector is a build-time option.
52 (thanks to Cyrus Harmon, after Raymond Toy)
53 * minor incompatible change: the method by which SBCL finds its
54 contributed modules has changed; it no longer relies on symbolic
55 links from an $SBCL_HOME/systems directory, but searches directly
56 in the subdirectories of $SBCL_HOME.
57 * enhancement: the dynamic heap size on the Linux/PPC platform is
58 markedly larger, even using the older Cheney garbage collector.
59 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
60 certain complicated string types. (reported by Gary King)
61 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
62 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
63 * fixed bug: DOCUMENTATION on structures no longer signals an
64 error if no structure type of the right name exists. (reported by
66 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
67 applicable non-standard primary or :AROUND method on
68 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
69 no longer cause unbound variable errors. (reported by Kevin Reid)
70 * fixed bug: metacircle resolution in cases where methods have slots
71 added before the slots from STANDARD-METHOD. (reported by Jean
73 * fixed bug: the Power PC assembly code for calling into lisp
74 sometimes computed its return address wrongly (depending on the
75 15th bit of the address, fixed at link time). This used to
76 manifest itself in a segmentation violation while building PCL.
77 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
78 many others over the years)
79 * fixed bug: ROOM no longer reports silly numbers for stack usage if
80 the stack pointer (treated as a signed integer) is negative.
81 (thanks to Peter van Eynde)
83 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
84 * new platform: experimental support for the Windows operating
85 system has been added. (thanks to Alastair Bridgewater)
86 * fixed several bugs in and robustified the PPC FFI (including
87 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
88 * bug fix: allow non-simple string symbol names (reported by Paul
90 * bug fix: interrupt handling on NetBSD (thanks to Richard M
92 * bug fix: saving a core corrupted callbacks on x86/x86-64
93 * bug fix: closed a loophole in metacircularity detection and
94 grounding in the PCL implementation of CLOS.
95 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
96 (reported by tomppa on #lisp)
97 * bug fix: clear the direction flag on Lisp -> C transitions, as
98 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
99 (reported by Andrej Grozin and Hendrik Maryns)
100 * optimization: major improvements to GC efficiency on GENCGC platforms
101 * optimization: faster implementation of EQUAL
102 * optimization: emit more efficient opcodes for some common
103 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
104 * optimization: several other minor code-generation improvements on x86-64
105 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
107 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
108 * minor incompatible change: (SETF CLASS-NAME) and (SETF
109 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
110 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
111 by ANSI as a generic function, and for consistency (SETF
112 GENERIC-FUNCTION-NAME) is treated likewise.
113 * fixed bug #233.b: make constraint propagation notice when a variable
114 value is changed after it is referenced but before it is used
115 * fixed bug #296: no more arbitrary behaviour from filenames with
116 odd characters as --load arguments. (reported by Adam Warner)
117 * fixed bug #390: :CHARACTER-SET pathname components now work as
118 expected. (reported by Tim Daly Jr)
119 * fixed bug #391: complicated :TYPE intersections in slot
120 definitions no longer cause an error in PCL internals.
121 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
122 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
123 returning the number of octets which would be written to the
124 file-stream. (thanks to Robert J. Macomber)
125 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
126 arguments to the internal stream functions. (thanks to David
128 * bug fix: stack exhaustion is now handled on the x86/SunOS
129 platform. (thanks to Daisuke Homma)
130 * bug fix: more accurate ROOM results on GENCGC platforms
131 * optimization: improved type inference for arithmetic-for
132 index variables in LOOP
133 * optimization: faster floating-point SQRT on x86-64
134 * fixed some bugs revealed by Paul Dietz' test suite:
135 ** DOCUMENTATION returns NIL instead of "" for method combinations
136 that don't have a docstring
138 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
139 * minor incompatible change: (SETF CLASS-NAME) and (SETF
140 GENERIC-FUNCTION-NAME) are no longer generic functions, and
141 therefore are not customizeable by user code (as seems to be at
142 least permitted and maybe required by AMOP). As a consolation,
143 however, the SBCL implementation of these functions now calls
144 REINITIALIZE-INSTANCE as specified by AMOP.
145 * bug fix: slot-definition documentation is propagated as per ANSI
146 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
147 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
148 argument, as do eslotd initialization methods. (from Pascal
149 Costanza's "Closer" project)
150 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
151 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
153 * bug fix: it is now possible to have more than one subclass of
154 STANDARD-GENERIC-FUNCTION without causing stack overflow.
155 (reported by Bruno Haible, Pascal Costanza and others)
156 * bug fix: the dependent update protocol now works for generic
157 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
159 * bug fix: condition-class instances corresponding to
160 DEFINE-CONDITION forms are now created eagerly. (reported by
161 Kalle Olavi Niemitalo on comp.lang.lisp)
162 * bug fix: floating point printing is more accurate in some
163 circumstances. (thanks to Simon Alexander)
164 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
165 merged with *DEFAULT-PATHNAME-DEFAULTS*.
166 * bug fix: callbacks on OS X now preserve stack-alignment.
167 (thanks to Cyrus Harmon)
168 * enhancement: the x86-64 disassembler is much better at
169 disassembling SSE instructions. (thanks to Lutz Euler)
170 * enhancement: CHECK-TYPE on a variable now allows type-inference
171 (thanks to Vincent Arkesteijn)
172 * optimization: improved performance of EUC-JP external format.
173 (thanks to NIIMI Satoshi)
174 * optimization: performance improvements to IO on file streams of
175 :ELEMENT-TYPE CHARACTER
176 * optimization: much faster memory allocation on x86-64
177 * optimization: faster garbage collections (latency and throughput)
179 * optimization: faster bignum subtraction and division on x86 and x86-64
180 (thanks to Lutz Euler)
181 * optimization: more accurate type-inference for loops that use a
182 floating point index variable or a negative step.
184 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
185 * bug fix: add a workaround to SBCL looping infinitely at startup on
186 Linux kernels with apparently buggy implementations of personality().
187 (thanks to Svein Ove Aas)
188 * bug fix: Unicode symbols are correctly printed in LDB backtraces
189 (thanks to David Lichteblau)
190 * bug fix: local bindings shadow global symbol macros for
191 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
192 * optimization: non-open coded uses of numeric comparison operators
193 (e.g. >) no longer cons when called with more than one parameter
194 on platforms supporting dynamic-extent allocation.
195 * enhancement: saving cores with foreign code loaded is now supported
196 on MIPS/Linux in addition to the previously supported platforms.
197 * bug fix: FIND-RESTART now tests for activity, not applicability when given
198 a restart object as identifier. (reported by Helmut Eller for CMUCL)
199 * bug fix: division by zero in sb-sprof when no samples were collected
200 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
201 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
202 (thanks to Alexander Kjeldaas)
203 * new feature: ignoring the return values of destructive functions
204 like NREVERSE signals a compile-time style-warning.
205 (thanks to Kevin Reid)
206 * new feature: the alignment of alien structure fields can be
207 explicitly specified. (thanks to Cyrus Harmon)
208 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
209 and dump core on SIGQUIT
211 ** incompatible change: threads do not inherit values of specials
212 from their parents (see manual)
213 ** bug fix: threads stacks belonging to dead threads are freed by the
214 next exiting thread, no need to gc to collect thread stacks anymore
215 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
216 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
217 ** bug fix: dynamic variable and thread start related gc lossage
218 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
220 * fixed some bugs revealed by Paul Dietz' test suite:
221 ** SUBTYPEP is slightly more accurate on heinously complicated
222 CONS types where some of the members have uncertain (in the
223 NIL, NIL sense) type relationships to each other.
225 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
226 no longer purified unless :PURIFY T is explicitly specified.
227 ** Non-purified cores are significantly smaller than before
229 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
230 * new feature: timers based on Zach Beane's excellent timer package
231 * added support for the following external formats: koi8-u,
232 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
233 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
234 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
235 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
236 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
237 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
238 * incompatible change: a threaded SBCL will no longer revert to
239 non-threaded mode on non-NPTL systems, but refuse to start entirely.
240 * bug fix: interrupts are disabled until startup is complete; no
241 more sigsegvs when receiving a signal to soon
242 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
244 * bug fix: add a workaround for the memory randomization features in
245 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
246 workaround will only be in effect on systems with the proc filesystem
248 * bug fix: printing objects of type HASH-TABLE signals a
249 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
251 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
252 have been running for over 50 days. (reported by Gilbert Baumann)
253 * bug fix: the logic for getting names of functions gets less
254 confused when confronded with alternate-metaclass
255 funcallable-instances. (reported by Cyrus Harmon)
256 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
257 consistent, even on internal alternate-metaclass objects.
258 * bug fix: flush closure information collected by physical
259 environment analysis prepass before the main pass. (bug reported
261 * bug fix: compiler pack phase does not modify a hashtable, which is
262 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
263 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
264 may be expanded. (reported by Alexander <malishev> on c.l.l)
265 * bug fix: no more style-warnings when compiling code using LAST
266 under high-SPEED low-SPACE settings. (reported by David Wallin
268 * fixed some incompatibilities between SBCL's MOP and the MOP
270 ** the METAOBJECT class is now implemented;
271 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
272 STANDARD-OBJECT, as required;
273 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
274 now compatible (as required by VALIDATE-SUPERCLASS); there
275 remains a consistency requirement over the presence or absence
276 of the FUNCTION class in the superclass list of finalized
277 classes; see the manual for more details;
278 ** the :AROUND method for COMPUTE-SLOTS on
279 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
280 requested slot ordering.
282 ** bug fix: parent thread now can be gc'ed even with a live
284 ** bug fix: binding a special with PROGV to no value is not
285 the same as that symbol not having been bound (thanks to
287 ** bug fix: binding specials is thread safe (thanks to
289 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
290 and SB-SPROF do not die with 'no handler for signal XX in
291 interrupt_handle_now(..)' anymore
292 ** bug fix: WITH-TIMEOUT works with multiple running threads
293 * fixed some bugs revealed by Paul Dietz' test suite:
294 ** ENSURE-GENERIC-FUNCTION should take a method class object for
295 the :method-class keyword argument.
297 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
298 * new port: the Solaris operating system on x86 processors is now
299 mostly supported, though some rough edges in the environment
300 remain. (thanks to Daisuke Homma)
301 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
302 space for its heap. (thanks to Thiemo Seufer)
303 * optimization: SBCL on MIPS platforms now supports dynamic-extent
304 closures. (thanks to Thiemo Seufer)
305 * minor incompatible change: eof selects abort in the debugger.
306 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
307 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
309 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
310 deprecated, and will go away in a future revision of SBCL.
311 * minor incompatible change: GC-ON and GC-OFF are no longer
312 implemented with a counter, it does not matter how many times gc
313 is switched on or off
314 * bug fix: discriminating functions for generic function classes
315 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
316 make invalid assumptions about method precedence order. (reported
318 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
319 platforms. (thanks to Thiemo Seufer)
320 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
321 * bug fix: correct pathname printing with printer escaping is on.
322 (thanks to Kevin Reid)
323 * bug fix: complex VOP definitions in "user-space" no longer trigger
324 package locks. (reported by Zach Beane)
325 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
326 for funcallable-instance functions, and is no different from
328 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
329 characters in the current readtable. (reported by Nicholas Neuss)
330 * bug fix: SUBTYPEP on various CONS types returns more a more
331 accurate acknowledgment of its certainty.
332 * optimizations: REMOVE-DUPLICATES now runs in linear time on
333 lists in some cases. This partially fixes bug 384.
334 * flush all standard streams before prompting in the REPL and the
336 * bug fix: signal handling and triggering gc do not conflict
337 directly anymore, in particular a high frequency sb-sprof does
338 not prevent gc from running
339 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
340 approximation for timezone and DST information between the
341 universal time epoch and the smallest negative 32-bit time_t.
342 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
343 year 1899 so long as the encoded time is non-negative.
344 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
345 the correct answer for non-constant rotate amounts when performing
346 an inline 32-bit rotation.
348 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
349 there is only one thread in the session
350 ** bug fix: memory leak for streams created in one thread and
351 written to in another
352 ** bug fix: lockup when compiled with gcc4
353 ** bug fix: race that allows the gc to be triggered when gc is
355 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
356 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
358 ** bug fix: finished threads are gc'ed properly
359 * fixed some bugs revealed by Paul Dietz' test suite:
360 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
361 has arguments with a different set of applicable methods from
362 the orignal arguments.
363 ** The type error thrown by MAP now has a correct expected-type
365 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
366 name a compiled function.
367 ** Three MISC tests where a large bignum was improperly coerced to
368 a float that couldn't represent that bignum during type
369 derivation were fixed.
370 ** SUBTYPEP can now handle types involving the negation of a
371 list-form FUNCTION type.
372 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
373 as (COMPLEX (AND RATIO (NOT FIXNUM))).
374 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
376 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
377 * New feature: Experimental support for bivalent streams: streams
378 opened with :element-type :default now allow character and binary
379 (unsigned-byte 8) I/O
380 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
381 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
383 * bug fix: no more highly sporadic "couldn't check whether ~S is
384 readable" when reading a stream and an interrupt hits in the middle
385 of a select system call
386 * compiler better recognizes complex arrays (reported by Eduardo
388 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
389 for variables on SBCL built with linkage-tables. (reported by Luis
391 * various error reporting improvements.
392 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
393 (Thanks to Thiemo Seufer)
394 * bug fix: nested reader invokations work correctly
395 * bug fix: it is possible to have simultaneous references to foreign
396 code and foreign data with the same name.
398 ** added x86-64 support
399 ** incompatible change: the threading api now works with thread
400 objects instead of thread ids
401 ** bug fix: threads are protected from signals and interruption when
402 starting up or going down
403 ** bug fix: a race where an exiting thread could lose its stack to gc
404 ** bug fix: don't halt on infinite error in threads if possible
405 ** fixed numerous gc deadlocks introduced in the pthread merge
406 ** bug fix: fixed thread safety issues in read and print
407 ** bug fix: debugger doesn't hang on session lock if interrupted at
408 an inappropriate moment
409 ** bug fix: run-program is now thread safe(r)
410 ** bug fix: inner with-recursive-lock no longer releases the mutex
411 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
412 * fixed some bugs revealed by Paul Dietz' test suite:
413 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
414 EXPECTED-TYPE slots filled.
415 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
417 ** MAKE-CONDITION accepts classes as type-designators.
418 ** COMPILE may never return NIL.
419 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
420 range before calling Unix time functions
422 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
423 * numerous signal handling fixes to increase stability
424 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
425 * minor incompatible change: we now correctly canonize default
426 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
427 as per the MOP, rather than the historical (INITARG INITFUNCTION
428 INITFORM). (reported by Bruno Haible)
429 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
430 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
431 for more information.
432 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
433 pathname is a directory pathname.
434 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
435 * dynamic space size on PPC has been increased to 768Mb. (thanks to
437 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
438 :METACLASS argument in addition to a class name. (reported by
439 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
440 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
441 (reported by Bruno Haible; patch from Gerd Moellmann)
442 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
444 * bug fix: more cleanups to the floating point exception handling on
445 x86-64 (thanks to James Knight)
446 * bug fix: the compiler does not try to constant fold calls of
447 COERCE and BIT-* functions when they return freshly constructed
448 objects. (reported by Nikodemus Siivola and Paul Dietz)
449 * optimization: improved the MIPS versions of generic (in the
450 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
451 * optimization: direct conversion of (unsigned-byte 32) to floats on
452 the PowerPC platform.
453 * optimization: structure instances with raw slots now use less
454 memory, and probably show better memory locality. (thanks to
456 * optimization: DYNAMIC-EXTENT declarations for lists and closures
457 are treated as requests for stack allocation on the x86-64,
458 Alpha-32, PPC and SPARC platforms.
459 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
460 the wrong file descriptor; implementation of SOCKET-OPEN-P.
461 (thanks to Tony Martinez)
463 ** gcing a dead thread can no longer lead to lockups
464 ** threads block signals until they are set up properly
465 ** errno is no longer shared by threads
466 ** interrupt-thread restores the eflags register on x86
467 ** fixed some lockups due to gc/thread interaction
468 * fixed some bugs revealed by Paul Dietz' test suite:
469 ** invalid dotted lists no longer raise a read error when
471 ** signal an error if a symbol that names a declaration is used
472 as the name of a type, or vice versa
473 ** allow using the (declare (typespec var*)) abbreviation for
474 (declare (type typespec var*)) with all type specifiers
475 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
476 ** always return NIL from PPRINT-POP when OBJECT is NIL
477 ** don't signal errors when pretty-printing malformed LABELS,
478 FLET or MACROLET forms
479 ** declarations in a DOLIST body are also in scope for the
481 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
483 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
486 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
487 * fixed cross-compiler leakages that prevented building a 32-bit
488 target with a 64-bit host compiler.
489 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
490 opened with :IF-EXISTS :OVERWRITE.
491 * fixed bug 281: error for an invalid qualifier in a short-form method
492 combination method is not signalled until the faulty method is called.
493 * bug fix: iteration variable type inferrer failed to deal with open
494 intervals. (reported by Alan Shields)
495 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
496 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
497 or more runtime options were provided to the sbcl binary.
498 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
499 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
500 * optimization: there's now a fast-path for fixnum arguments in the
501 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
502 * optimization: the code generated on x86-64 is more compact thanks
503 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
504 * bug fix: floating-point exception handling now partly works on
505 x86-64. (Thanks to James Knight)
506 * improvement to the MIPS backend from Thiemo Seufer: C-style
507 64-bit long long arguments and return values to alien functions
509 * fixed some bugs revealed by Paul Dietz' test suite:
510 ** the type-error signalled from WARN has a filled-in DATUM slot.
511 ** the type-error required when a stream is not associated with
512 a file has the stream as its datum.
513 ** type-errors on single-floats on x86-64 no longer have
514 :INVALID-OBJECT as the datum
515 ** the type-errors signalled for invalid function names now have
516 a correct expected type
517 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
518 for structures defined with a :TYPE. Documentation strings for
519 typed structures are no longer immediately discarded
520 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
521 broadcast streams and synonym streams. FILE-LENGTH now also works
522 on broadcast streams.
524 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
525 * incompatible change: the --noprogrammer option, deprecated since
526 version 0.7.5, has been removed. Please use the equivalent
527 --disable-debugger option instead.
528 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
530 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
531 inoperational for a while now) has been completely removed.
532 * null lexical environments are now printed as #<NULL-LEXENV>,
533 significantly reducing the amount of clutter in typical backtraces.
534 * documentation on weak pointers, finalization, and after GC hooks
535 has been added to the manual.
536 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
537 as other specialized array types, is much faster. SUBSEQ and
538 COPY-SEQ on such arrays have also been sped up.
539 * optimization: EQL is now more efficient when at least other argument
540 is known to be of type (OR FIXNUM (NOT NUMBER)).
541 * fixed bug: the runtime is now less vulnerable to changes in the
542 size of the SBCL object on OS X, and virtual memory is reserved for
543 all spaces that need to be at a fixed address.
544 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
545 * fixed bug: finalizers and after GC hooks that cause consing are now
547 * fixed bug: compiler error messages and summaries are now printed to
548 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
549 * fixed inference of the upper bound of an iteration variable.
550 (reported by Rajat Datta).
551 * fixed bug 211e: calling local functions with duplicated constant
552 keyword argument no longer causes a bogus style warning about an
554 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
555 ftype declaration to be disregarded. (reported by Dave Roberts)
556 * fixed bug 373: caused by erronous compilation of references to alien
557 variables in the runtime on ppc/darwin.
558 * fixed bug 376: CONJUGATE type deriver.
559 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
560 (reported by Baughn on #lisp)
561 * a cross-compiler bug on non-x86 platforms has been identified and
562 fixed. (thanks to Bruno Haible)
563 * improvements to the MIPS runtime code for increased stability.
564 (thanks to Thiemo Seufer)
565 * increased the maximimum compact environment size to allow
566 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
567 * partial workaround for bug 135: don't name the function we're
568 calling for hairy cases of EVAL, so as not to accumulate one environment
569 entry per call to EVAL. (reported by Kevin Reid)
570 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
571 * optimization: type testing for non-vector arrays should be faster.
572 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
574 * bug fix: setting 31st element of a bit vector to zero did not work
576 * bug fix: redefining a class definition which failed due to a
577 previous accessor / function clash now works (but see BUGS entry
578 #380 for more problems in this area). (thanks to Zach Beane)
579 * the long form of DEFINE-METHOD-COMBINATION disables method group
580 checking when given a single method group with pattern *. (thanks
582 * on x86 compiler supports stack allocation of results of simple
583 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
584 * support for building and running on Mac OS X 10.4 "Tiger" has been added
585 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
586 * fixed some bugs related to Unicode integration:
587 ** the restarts for recovering from input and output encoding
588 errors only appear when there is in fact such an error to
590 * fixed some bugs revealed by Paul Dietz' test suite:
591 ** MISC.549 and similar: late transformation of unsafe type
592 assertions into derived types caused unexpected code
594 ** SCALE-FLOAT type deriver is less wrong.
595 ** type derivers for EXP, LOG and similar functions compute result
596 types for complex arguments better.
597 ** (MISC.563) CONJUGATE type deriver works for very restricted
599 ** out-of-line type testers for character strings are available.
600 ** EQUAL compiler transform understands specialness of objects
602 ** accessing double-floats stored on the stack now works on x86-64.
603 ** debugger internals could sometimes create invalid lispobjs,
604 resulting in GC crashes.
605 ** MISC.548: type check weakening can convert required type into
607 ** initialization forms for bindings are not in scope of free special
610 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
611 * incompatible change: thread support for non-NPTL systems has
612 been removed - locking is buggy and unreliable. A threaded
613 SBCL build will now warn at startup and refuse to create
614 new threads, unless futex support is detected in the kernel
615 * incompatible change: the top level REPL now has only an ABORT
616 restart associated with it, not TOPLEVEL and ABORT as it used to.
617 TOP and TOPLEVEL are now available as debugger commands for
618 returning to the top level.
619 * incompatible change: forms evaluated in the REPL now use the
620 global optimization policy.
621 * incompatible change: user- and system-initialization files are
622 no longer processed with LOAD, but by READ and EVAL; hence the
623 global optimization policy, startup package, readtable, etc,
625 * The .fasl file format number has been incremented because of
626 various incompatible changes.
627 * internal entry point details and argument counts no longer appear
628 in backtraces unless explicitly requested by setting
629 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
630 * built-in and standard functions no longer have names like "top
631 level local call to FOO".
632 * fixed bug 32: functions defined in non-null lexical environments
633 now have more legible printed representation
634 * fixed bug 33: functions defined in non-null lexical environemnts
635 are now more amenable to inspection by INSPECT.
636 * workaround for bug 354: XEPs no longer appear in backtraces unless
637 explicitly requested.
638 * fixed bug: receiving the signal which results from attempting to
639 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
640 SIGBUS on other *BSDs) on architectures where the C stack is also
641 the Lisp stack (x86 and x86-64) from foreign code no longer leads
642 to debugger confusion or wild instability. (reported by Cheuksan
644 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
645 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
646 (reported by Lutz Euler)
647 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
648 compiled code. (reported by Paul Dietz)
649 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
650 specialized parameter to a method to something that is not TYPEP
651 the specializer is now possible.
652 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
653 face of package deletion.
654 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
655 pathname. (reported by David Tolpin on c.l.l)
656 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
657 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
658 STRING-STREAM and FILE-STREAM, respectively.
659 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
660 than previously; in addition, its test suite is now run on build.
661 (thanks to Luke Gorrie)
662 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
663 * optimization: added a immediate representation for single-floats
665 * fixed some bugs revealed by Paul Dietz' test suite:
666 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
667 ** attempting to create a package with a colliding nickname causes
668 correctable errors to be signalled.
669 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
670 ** several x86-64 backend bugs related to sign-extension of immediate
673 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
674 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
675 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
677 * fixed loading of multiply forward-referenced layouts.
678 (thanks to Cheuksan Wang)
679 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
680 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
681 (see COMPILE-FILE documentation for details of :PRINT option)
682 * fixed bugs 19 and 317: fixed-format floating point printing is
683 more accurate. This also fixes a bug reported by Adam Warner
684 related to the ~@F format directive.
685 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
687 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
688 dispatch table if the from-char is a dispatch macro character.
689 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
690 by the wordsize no longer ignore the last word. (reported by Lutz
692 * fixed bug in type checking of dynamic-extent variables. (reported
694 * optimization: sequence traversal functions use their freedom to
695 coerce function designators to functions.
696 * optimization: code with many calls to CLOS methods specialized on
697 CLOS classes has had redundant type checks removed.
698 * optimization: added declarations to speed up operations that access
699 the internal character database (for example STRING-UPCASE)
700 * optimization: comparison operations between floats and sufficiently small
701 fixnums no longer create extra rationals
702 * fixed some bugs related to Unicode integration:
703 ** portions of multibyte characters at the end of buffers for
704 character-based file input are correctly transferred to the
705 start of the buffer at the next read.
706 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
707 passing it through to OPEN.
708 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
709 argument given to internal calls to OPEN.
710 ** the built-in comment readers (introduced by character sequences
711 ";" and "#|") are more forgiving to encoding errors; they will
712 STYLE-WARN and then attempt to resync the stream at a character
713 boundary. (thanks to Teemu Kalvas)
714 * fixed some bugs in the x86-64 port:
715 ** Negative short int return values from c-calls are sign-extended
717 ** The stack is aligned to 16-bytes for c-calls, as required by
718 the ABI. (thanks to Cheuksan Wang)
719 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
720 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
721 for enums. (thanks to Vincent Arkesteijn)
722 ** Multiple small optimizations and bugfixes for floating point
724 * fixed some bugs revealed by Paul Dietz' test suite:
725 ** Space, Tab, Linefeed, Return and Page have the invalid
726 secondary constituent character trait.
727 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
729 ** WITH-INPUT-FROM-STRING should only update the index place on
731 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
733 ** Bit-array operations (BIT-AND and similar) worked incorrectly
734 with one-dimensional arrays with fill pointers.
735 ** TYPE-OF failed on a complex with an integer realpart and a
737 ** compiler failure during type inference for the code of form
738 (IF (EQL X (THE ...)) ...) (MISC.535).
740 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
741 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
742 platform. The port supports 61-bit fixnums, large memory spaces
743 and reloading shared object files.
744 * enhancement: saving cores with foreign code loaded is now
745 supported on ppc/Darwin in addition to the previously supported
747 * enhancement: the statistical profiler now walks deeper into the
748 call stack for more meaningful call-graphs and accrued time
749 reports (x86/x86-64 only). It also now reports time spent in
751 * enhancement: it is now possible to trace most individual methods
752 of a generic function in addition to tracing the generic function
754 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
756 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
757 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
758 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
759 produces an error. (thanks to Vincent Arkesteijn)
760 * bug fix: NAMESTRING on pathnames with :WILD components in their
761 directories works correctly. (thanks to Artem V. Andreev)
762 * fixed bug 125: compiler preserves identity of closures. (reported
764 * bug fixed: functions with &REST arguments sometimes failed with
765 "Undefined function" when compiled with (DEBUG 3). (reported by
767 * bug fix: overflow during compiling of setting element of a bit
768 vector with constant index and value. (reported by Timmy Douglas)
769 * build fix: fixed the dependence on *LOAD-PATHNAME* and
770 *COMPILE-FILE-PATHNAME* being absolute pathnames.
771 * on x86 compiler partially supports stack allocation of dynamic-extent
773 * GO and RETURN-FROM do not check the extent of their exit points
774 when compiled with SAFETY 0.
775 * fixed some bugs related to Unicode integration:
776 ** encoding and decoding errors are now much more robustly
777 handled; it should now be possible to recover even from invalid
778 input or output to the terminal. (thanks to Teemu Kalvas)
779 ** provided a first cut at implementing STRING-TO-OCTETS and
780 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
781 ** altered the SB-MD5 contributed module slightly, changing the
782 interface just enough for it to be supportable for builds where
783 lisp characters are not eight bits.
784 * fixed some bugs revealed by Paul Dietz' test suite:
785 ** the FORMATTER-generated functions for ~V[ conditionals require
786 the correct number of arguments.
787 ** READ-FROM-STRING returns the mandated second value when applied
788 to displaced strings.
789 ** the #\Rubout and #\Backspace characters are treated as invalid
790 constituent characters by the tokenizer.
792 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
793 * new feature: reloading changed shared object files with
794 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
795 * new feature: references to foreign variables and functions
796 can now be compiled and loaded before the corresponding shared
797 object file is loaded, as long as the foreign definitions are
798 available at runtime.
799 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
800 supported. (thanks to Dan Debertin)
801 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
802 just keywords. (thanks to Vincent Arkesteijn)
803 * fixed bug #331: structure-class instances corresponding to
804 DEFSTRUCT forms are now created eagerly.
805 * fixed bug #345: backtraces from calls to undefined functions work
806 on x86 as well. Related bug #61 is now also partially fixed on x86
807 (backtraces from throws to unknown catch tags.)
808 * bug fix: lambda-list parsing is now stricter vrt. order and number
809 of lambda-list keywords.
810 * bug fix: as specified by AMOP, an error is signalled if a
811 class-option appears multiple times in a DEFCLASS form. (reported
813 * bug fix: RANDOM can be compiled when the compiler derives the type
814 of its numeric argument as a disjoint set of small integers.
815 (reported by Paul Dietz)
816 * bug fix: starting a core saved with shared objects loaded when
817 those objects are not available no longer causes threaded SBCL to
818 hang. (reported by Sean Ross)
819 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
821 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
822 argument, which must always be NIL. (reported by Kalle Niemitalo)
823 * bug fix: printing 1.0d+23 no longer results in an error.
824 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
825 * bug fix: structure slot setters preserve evaluation order. (thanks
827 * fixed some bugs related to Unicode integration:
828 ** RUN-PROGRAM can allow its child to take input from a Lisp
829 stream. (reported by Stefan Scholl)
830 ** ASDF-INSTALL successfully downloads .tar.gz files in all
831 locales. (reported by Ken Causey)
832 * fixed some bugs revealed by Paul Dietz' test suite:
833 ** INCF, DECF and REMF evaluate their place form as specified in
835 ** FORMATTER expands ~{ ~} iteration directives with V or #
836 parameters correctly.
837 ** FORMATTER deals with the ~@[ ~] conditional directive where the
838 consequent uses no arguments correctly.
839 ** the system has a partial understanding of the (COMPLEX RATIO)
841 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
844 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
845 * new feature: a build-time option (controlled by the :SB-UNICODE
846 keyword feature, enabled by default) for building the system with
847 support for the entire 21-bit character space defined by the
849 * new feature: the system now has rudimentary external-format
850 support; the primary user-visible change at this time is that
851 characters with the high bit set (such as lower-case-e-acute) will
852 print correctly to a terminal in a UTF-8 environment.
853 * minor incompatible change: BASE-CHAR no longer names a class;
854 however, CHARACTER continues to do so, as required by ANSI.
855 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
856 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
857 has been moved to the SB-EXT package (temporarily re-exported from
859 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
860 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
862 * fixed bug #351: better error-handlind and reporting for malformed
864 * fixed bug #350: bignum-printing is now more memory-efficient,
865 allowing printing of very large bignums, eg. (expt 2 10000000).
866 (reported by Bruno Haible)
867 * fixed bug #302: better primitive-type selection for intersection
869 * fixed bug #308: non-graphic characters now all have names, as
870 required. (reported by Bruno Haible)
871 * bug fix: redefining a class with different superclasses now correctly
872 removes it from the direct-subclasses of its previous superclasses.
873 (reported by David Morse)
874 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
875 new value now works. (reported by Bruno Haible)
876 * bug fix: correct canonicalization of multiple non-standard slot
877 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
878 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
879 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
880 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
881 now exists, an signals an error.
882 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
883 during macroexpansion for non-iteration variables as well. (reported
884 by Bruno Haible for CMUCL)
885 * bug fix: Cyclic structures and unprintable objects in compiler
886 messages no longer cause errors. (reported by Bruno Haible)
887 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
888 and READ-FROM-STRING all now return a primary value of NIL if
889 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
890 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
891 (reported by Bruno Haible for CMUCL)
892 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
893 do not designate a condition. (reported by Bruno Haible for
895 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
896 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
897 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
898 permitting resolution in favour of any of the conflicting symbols.
899 (reported by Bruno Haible for CMUCL)
900 * bug fix: EQUAL compiler optimizations is less aggressive on
901 strings which can potentially compare true despite having distinct
902 specialized array element types.
903 * bug fix: unit enumerations can be defined without dividing by
904 zero. (thanks to Vincent Arkesteijn)
905 * FORMAT compile-time argument count checking has been enhanced.
906 (report from Bruno Haible for CMUCL)
907 * a partial workaround for the bug 262: the compiler does not try to
908 inline-expand a local function doing RETURN-FROM from a deleted
909 BLOCK. (thanks to Peter Denno for the bug report and to David
910 Wragg for the simple test case)
911 * fixed some bugs revealed by Paul Dietz' test suite:
912 ** RENAME-PACKAGE allows all package designators as new package
914 ** constraint propagation and lambda variable substitution are
915 more cautious in dealing with partially deleted code.
916 ** compiler sometimes lost reoptimization passes.
917 ** CERROR, given a condition as condition designator, passes its
918 remaining arguments to the continue format control without
920 ** Case-altering FORMAT directives work correctly on non-ASCII
922 ** The REAL type specifier handles bounds outside the floating
923 point ranges without signalling FLOATING-POINT-OVERFLOW.
924 ** Functions with IR1-transformations can create intercomponent
925 references to global functions.
926 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
928 ** FORMAT ~R treats a nil value for its first parameter correctly.
930 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
931 * enhancement: saving cores with foreign code loaded is now
932 supported on x86/NetBSD and sparc/Linux in addition to the previously
934 * bug fix: on some platforms repeated installations caused multiple
935 copies of HTML documentation to be installed -- should not happen
936 any more. (reported by Stefan Scholl)
937 * bug fix: parsing self-recursive alien record types multiple times
938 no longer causes infinite recursion. (reported by Thomas F. Burdick,
939 original patch by Helmut Eller for CMUCL)
940 * bug fix: stack-exhaustion detection works now on NetBSD as well.
941 (thanks to Richard Kreuter)
942 * bug fix: defining classes whose accessors are methods on existing
943 generic functions in other (locked) packages no longer signals
944 bogus package lock violations. (reported by François-René Rideau)
945 * bug fix: special variables as DEFMETHOD parameters no longer have
946 associated bogus type declarations. (reported by David Wragg and
948 * bug fix: read-write consistency on streams of element-type
949 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
950 * bug fix: redefiniton of the only method of a generic function with
951 no DEFGENERIC no longer emits a full WARNING. In addition,
952 redefinition of generic functions with no DEFGENERIC to an
953 incompatible lambda list now signals an error. (thanks to Zach
955 * bug fix: DEFGENERIC now works even when there's a function of the
956 same name in an enclosing lexical environment. (thanks to Zach
958 * fixed compiler failure, caused by instrumenting code during
959 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
960 * optimization: added loop analysis and improved register allocation
961 * fixed some bugs revealed by Paul Dietz' test suite:
962 ** POSITION on displaced vectors with non-zero displacement
963 returns the right answer.
964 ** (SIMPLE-STRING) is a valid type specifier for sequence
966 ** *PRINT-LEVEL* handling for slotless structures is pedantically
968 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
969 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
971 ** The justification version of the FORMAT ~< directive treats
972 non-zero minpad parameter correctly.
974 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
975 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
976 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
977 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
978 the supported interface.
979 * new feature: Single-stepping of code compiled with DEBUG 2 or
980 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
981 * new feature: saving cores with foreign code loaded is now
982 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
983 Timothy Moore's work for CMUCL)
984 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
985 parameters to * instead of NIL if no initform is supplied.
986 (reported by Johan Bockgaard)
987 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
988 correctly when FUNCALL appears as the car of the form. Note:
989 despite this FUNCALL forms are not currently subject to
990 compiler-macro expansion. (port of Raymond Toy's fix for the
991 same from CMUCL, reported by Johan Bockgaard)
992 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
993 (thanks for Teemu Kalvas)
994 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
995 case instead of terminating the iteration (thanks for Julian
996 Squires, Sean Champ and Raymond Toy)
997 * bug fix: incorrect expansion of defgeneric that caused a style
998 warning. (thanks for Zach Beane)
999 * bug fix: slot accessor effective method computation works properly
1000 for classes with multiple non-standard applicable methods on
1001 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1002 * on x86 compiler supports stack allocation of results of LIST and
1003 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1004 CMUCL implementation by Gerd Moellmann)
1005 * fixed some bugs revealed by Paul Dietz' test suite:
1006 ** FORMAT strings with both the ~<~:;~> form of the justification
1007 directive and pretty-printing directives cause an error.
1009 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1010 * incompatible change: the internal functions
1011 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1012 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1013 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1014 instead of the old functions.
1015 * new feature: on platforms where "dladdr" is available foreign
1016 function names now appear in backtraces. (based on Helmut Eller's
1018 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1019 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1021 * documentation: ASDF manual is now distributed with SBCL.
1022 * bug fix: SBCL can now load its contributed modules with REQUIRE
1023 even if the system-provided entries have been removed from
1024 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1026 * bug fix: backtraces involving undefined functions or assembly
1027 routines are more informative. (thanks to Brian Downing)
1028 * bug fix: mutually referent alien structures now work correctly.
1029 (reported by Rick Taube)
1030 * bug fix: structures defined by WITH-ALIEN can be referred to
1031 within other definitions in the same WITH-ALIEN.
1032 * bug fix: division operators (MOD, TRUNCATE and the like) with
1033 constant zero divisors and integer dividends no longer generate
1035 * bug fix: provide default methods for INPUT-STREAM-P and
1036 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1037 * bug fix: improve the detection and resolution of MOP metacycles
1038 (where effective method computation depends on the generic function
1039 having its effective method computed). (reported by Bruno Haible)
1040 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1041 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1042 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1043 * bug fix: pathnames with relative directory components can be
1044 represented relative to default pathnames.
1045 * optimization: in taking the GCD of bignums, reduce the two bignums
1046 to approximately the same size (using Euclid's algorithm) before
1047 applying the more sophisticated binary GCD. (thanks to Juho
1049 * optimization: COUNT on bitvectors now operates word-at-a-time.
1050 * optimization: ASH with a positive, but not necessarily constant,
1051 (leftwards) shift, when in a modular context, is compiled to a
1053 * fixed some bugs revealed by Paul Dietz' test suite:
1054 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1055 if the corresponding argument is NIL.
1056 ** FORMAT directives accepting a minpad parameter treat negative
1058 ** literal commas as character parameters to FORMAT directives are
1060 ** literal spaces directly after ~<Newline> directives within a
1061 format-logical-block (~:< ~@:>) do not induce :FILL-style
1062 conditional newlines.
1063 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1064 as specified: it no longer includes conditional newlines).
1065 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1067 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1068 :PER-LINE-PREFIX argument does not evaluate to a string.
1070 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1071 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1072 the manual for details; for now, package locks can be disabled by
1073 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1074 if no major problems are found then it is likely that they will be
1075 compiled in unconditionally.
1076 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1077 unsupported operators on all platforms. To load a shared library
1078 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1079 object file, link it into a shared library outside of SBCL and
1080 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1082 * bug workaround: ROOM T and the GC were not working together
1083 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1084 reporting facility aren't preserved. That reporting has been
1085 disabled, so now until and unless someone figures out how to make it
1086 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1087 * minor incompatible change: as threatened around sbcl-0.8.0, the
1088 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1089 The symbols are also exported from SB-PCL for backwards
1090 compatibility, but more so than before SB-PCL should be treated as
1091 an implementation-internal package.
1092 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1094 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1095 It is known as of this release not to work on the Alpha, however.
1096 * fixed bug #167: errors signalled due to illegal syntax in method
1097 bodies are now more legible.
1098 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1099 designators and can hence be used with TYPEP.
1100 * fixed bug #333: CHECK-TYPE now ensures that the type error
1101 signalled, if any, has the right object to be accessed by
1102 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1103 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1104 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1106 * fixed bug #334: programmatic addition of slots using specialized
1107 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1108 and :ALLOCATION :CLASS slots.
1109 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1110 float exponents. (rereported by Peter Seibel)
1111 * fixed a bug: #\Space (and other whitespace characters) are no
1112 longer considered to be macro characters in standard syntax by
1113 GET-MACRO-CHARACTER.
1114 * fixed bug: initialization of condition class metaobjects no longer
1115 causes an instance of the condition to be created. (reported by
1117 * fixed bug: it is now possible to have slots such that
1118 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1119 description is neither :INSTANCE nor :CLASS.
1120 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1121 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1122 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1123 non-standard SLOT-DEFINITION-ALLOCATION values.
1124 * fixed bugs: various race conditions handling exiting threads.
1125 CL-PPCRE's thread torture test now passes.
1126 * fixed arguably-a-bug: GC time proportional to number of threads in
1127 system even when most of them are idle
1128 * optimization: improved performance of BIT and SBIT on bit-vectors.
1129 * .fasl file incompatibility: The fasl file version number has
1130 been incremented because of changes associated with package locks.
1132 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1133 * minor incompatible change: the system no longer provides
1134 optimization hints (conditions which are TYPEP
1135 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1136 mode; these hints are still emitted when higher SPEED optimization
1138 * new contrib module: a sampling profiler (profiling by statistical
1139 sampling, rather than by instrumenting functions) is available as
1140 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1141 * the behaviour of the standard function ED is now customizeable by
1142 third parties through a hook variable: see ED's documentation
1143 string for information on the protocol.
1144 * the compiler no longer emits efficiency notes for (FUNCALL X)
1145 when the type of X is uncertain under default optimization
1147 * fixed bug 276: mutating a binding of a specialized parameter to a
1148 method to something that is not TYPEP the specializer is now
1150 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1151 exhibit the required behaviour on the x86 platform. (thanks to
1152 Peter van Eynde, Eric Marsden and Bruno Haible)
1153 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1154 even for difficult arguments. (reported by Peter Graves)
1155 * fixed bug 141a: the backquote printer now descends quoted
1157 * fixed another bug in backquote printing: no more destructive
1158 modification of the form's list structure. (reported by Brian
1160 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1161 move between its address being taken and the call to
1162 interrupt_thread, fixing a crashing race condition.
1163 * the SB-POSIX contrib implementation has been adjusted so that it
1164 no longer exhibits ridiculously poor performance when constructing
1165 instances corresponding to C structs.
1167 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1168 * minor incompatible change: the sb-grovel contrib now treats C
1169 structures as alien (in the sense of SB-ALIEN) objects rather than
1170 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1171 has implications for memory management of client code
1172 (sb-grovel-returned objects must now be manually managed) and for
1173 type safety (alien objects now have full types).
1174 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1175 used to control emission of compiler diagnostics, rather than the
1176 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1177 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1178 quality should be considered deprecated.
1179 * install.sh now installs the user manual as well
1180 * (not quite a new documentable feature, but worth considering in
1181 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1182 beginnings of a semantically meaningful condition hierarchy is
1183 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1185 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1186 designator as the defaults argument.
1187 * fixed bug: Displaced arrays whose displaced-to array has become
1188 too small now cause ARRAY-DIMENSION to signal an error, providing
1189 for safer bounds-checking. (reported by Bruno Haible)
1190 * fixed bug: DEFCLASS slot definitions with identical :READER and
1191 :WRITER names now signal a reasonable error. (reported by Thomas
1193 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1195 * fixed bug: Invalid slot specification errors now print correctly.
1196 (thanks to Zach Beane)
1197 * fixed bug 320: Shared to local slot value transfers in class
1198 redefinitions now happen corrently from superclasses as
1199 well. (reported by Bruno Haible)
1200 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1202 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1203 improper lists correctly. (reported by Bruno Haible)
1204 * fixed bug 313: source-transform for <fun-name> was erroneously
1205 applied to a call of a value of a variable with name <fun-name>.
1206 (reported by Antonio Menezes Leitao)
1207 * fixed bug 307: The obsolete instance protocol ensures that
1208 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1209 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1210 remove dead unknown-values globs from the middle of the stack.
1211 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1212 easier to resume long computations after using *BREAK-ON-SIGNALS*
1213 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1214 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1215 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1216 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1217 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1218 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1220 * fixed a (fairly theoretical) bug in string printing: if
1221 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1222 does not have array-element-type equal to the most general string
1224 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1225 function name. (thanks to Nikodemus Siivola)
1226 * fixed bug: compile-time format string checker failed on
1227 non-closed ~{. (reported by Thomas F Burdick)
1228 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1229 don't warn on legal loop code involving more than one aggregate
1230 boolean. (thanks to Nikodemus Siivola)
1231 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1232 in keyword position to MAKE-PACKAGE was not being reported as
1234 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1235 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1237 * fixed bugs 280 and 312: the checking for multiple definitions in a
1238 file is less likely to become confused by uses of inline
1240 * fixed bug: the #S reader macro performs the keyword coercion
1241 specified for slot names. (reported by Kalle Niemitalo)
1242 * fixed bug: lambda lists may contain symbols whose names start with
1243 & but are not lambda-list-keywords; their occurrence triggers a
1245 * fixed bug 321: define-method-combination argument lambda lists do
1246 not cause invalid code to be generated when &optional or &aux
1247 variables have default values. (reported by Bruno Haible)
1248 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1249 have CLOS classes; this appears also to have fixed reported
1250 instability in stack exhaustion detection.
1251 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1252 MUFFLE-WARNING when no associated restart is present is now
1254 * optimization: rearranged the expansion of various defining macros
1255 so that each expands into only one top-level form in a
1256 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1258 * optimization: used a previously little-used slot in symbols to
1259 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1261 * fixed some bugs revealed by Paul Dietz' test suite:
1262 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1264 ** Fixed an optimization bug related to inheritance of initforms
1265 from local to shared slots.
1266 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1267 ** CAST optimizer forgot to flush argument derived type.
1268 ** print/read consistency on floats is now orders of magnitude
1269 more likely. (thanks also to Bruno Haible for a similar report
1271 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1272 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1273 for objects of type REAL. Make it so.
1274 ** SXHASH is specified (infelicitously) to respect similarity,
1275 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1276 it so. (thanks to Markus Ziegler)
1277 ** on the Alpha, the compiler succeeds in compiling functions
1278 returning a known number of arguments greater than 63.
1279 ** fixed handling of invalid NIL arguments in keyword position
1281 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1283 ** provided a readably-printable representation for RANDOM-STATE
1285 ** ensured that pathnames, where they have a namestring, always
1286 print using #P"..." syntax.
1288 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1289 * Support for the forthcoming 2.0 version of the NetBSD kernel
1290 running on x86 hardware has been added. (thanks to Perry
1291 E. Metzger most immediately, and others for their past work)
1292 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1293 current release version; SBCL's OpenBSD support had been broken
1294 since about the time of OpenBSD's switch to ELF binary format.)
1295 * [placeholder for DX summary]
1296 ** user code with &REST lists declared dynamic-extent, under high
1297 speed or space and low safety and debug optimization policy.
1298 * The manual has been converted to Texinfo format and the debugger
1299 chapter from the cmucl manual has been added.
1300 * A facility has been added to extract documentation strings from
1301 sbcl and store them as Texinfo-formatted snippets for inclusion in
1302 the manual (via Texinfo's @include directive)
1303 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1304 the test case to Dave Roberts)
1305 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1306 pointers. (reported by Sean Ross)
1307 * bug fix: PROFILE output is printed nicely even for large numerical
1308 values. (thanks to Zach Beane)
1309 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1310 greater than 32 handle EOF correctly.
1311 * bug fix: on X86 an immediate argument of the IMUL instruction is
1312 correctly printed in disassembly. (thanks to Lutz Euler)
1313 * bug fix: class slots in redefined classes preserve their old
1314 values. (thanks to Bruno Haible and Nikodemus Siivola)
1315 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1316 succeeds. (reported by Marco Baringer)
1317 * fixed some bugs revealed by Paul Dietz' test suite:
1318 ** READ-SEQUENCE now works on ECHO-STREAMs.
1319 ** RATIONALIZE works more according to its specification. (thanks
1322 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1323 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1324 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1325 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1326 should matter to you only if you rebind the printer control
1327 variables and then find you want different bindings in the
1328 debugger than in the ordinary execution of your program.)
1329 * The runtime build system has been tweaked to support building
1330 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1331 assembler and linker. (thanks to Nikodemus Siivola)
1332 * Unbound, undefined, undeclared variables now trigger full
1333 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1334 more often programmer error than deliberate exploitation of undefined
1336 * optimization: the hash algorithm for strings has changed to one
1337 that is less vulnerable to spurious collisions. (thanks to Juho
1339 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1340 needless bounds checking. (thanks to Juho Snellman)
1341 * optimization: implemented multiplication as a modular
1342 (UNSIGNED-BYTE 32) operation on the PPC backend.
1343 * fixed some bugs revealed by Paul Dietz' test suite:
1344 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1346 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1347 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1348 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1350 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1352 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1354 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1355 their output stream on EOF from read.
1356 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1357 have been read to end-of-file.
1358 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1360 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1361 description of determination of which consecutive characters
1363 ** Printing the "Space" character with escaping on now yields "#\\ ",
1364 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1365 ** Reading floating-point numbers with *READ-BASE* set to a number
1366 less than 10 works correctly.
1367 ** Reading floating-point numbers with *READ-BASE* set to a number
1368 more than 10 works correctly.
1369 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1370 the readtable currently in effect.
1372 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1373 * minor incompatible change: parsing of namestrings on a physical
1374 (Unix) host has changed; numbers after the final #\. in a
1375 namestring are no longer interpreted as a version field. This is
1376 intented to be largely invisible to the user, except that the
1377 meaning of the namestring "*.*.*" has changed: it now refers to a
1378 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1379 should usually be replaced by
1380 (make-pathname :name :wild :type :wild :version :wild)
1381 with the added benefit that this is more likely to be portable.
1382 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1383 option to OPEN now signals an error if the file being opened
1384 exists; this may have an impact on existing code.
1385 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1386 signals received, on the PowerPC platforms (both Linux and
1387 Darwin). (thanks to Pierre Mai for pointing out the location of
1389 * several fixes on OS X: The system now builds and runs cleanly on
1390 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1391 (thanks to Brian Mastenbrook)
1392 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1393 second-resolution: integer multiples of 1/3600 between -24 and 24.
1394 (thanks to Vincent Arkesteijn)
1395 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1396 type when called with 1 argument; PEEK-CHAR checked type of
1397 PEEK-TYPE only after having read first character from a
1398 stream. (reported by Peter Graves)
1399 * bug fix: the garbage collector now has much better locality
1400 behaviour, and in particular no longer treats all memory as being
1401 exhausted when many small objects point to each other in a deeply
1403 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1404 recognized as being TYPEP their class.
1405 * bug fix: the PUSHNEW documentation string has been corrected.
1406 (thanks to Vincent Arkesteijn)
1407 * bug fix: defaulting of the value for the last of an atypically
1408 large number of multiple values being bound was not being
1409 performed correctly on the Alpha or PPC platforms
1410 * optimization: implemented multiplication as a modular
1411 (UNSIGNED-BYTE 32) operation on the x86 backend.
1412 * optimization: SEARCH on simple-base-strings can now be open-coded.
1413 (see also contrib/compiler-extras.lisp for inspiration for
1414 teaching the compiler about the Boyer-Moore algorithm).
1415 * value, returned by MAX (and MIN) called with several EQUALP, but
1416 not EQL, arguments now does not depend on compiler settings.
1417 * fixed some bugs revealed by Paul Dietz' test suite:
1418 ** in stack analysis liveness information is propagated from
1419 non-local entry points.
1420 ** pathwise CAST removing failed when the CAST node did not start
1422 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1423 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1425 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1426 host is already defined.
1427 ** RENAME-FILE works on streams instead of signalling an internal
1429 ** PEEK-CHAR uses the current readtable when determining whether
1430 or not a character is whitespace.
1431 ** MERGE-PATHNAMES handles the case when the pathname does not
1432 specify a name while the default-pathname specifies a version.
1433 ** Pathnames now stand a better chance of respecting print/read
1435 ** Attempting to use standardized file system operators with a
1436 pathname with invalid :DIRECTORY components signals a
1438 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1440 ** DIRECTORY on logical pathnames is more correct.
1441 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1442 signal a TYPE-ERROR if their argument is not a stream.
1443 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1444 designator argument does not designate a stream.
1445 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1446 examining the synonym.
1447 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1449 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1450 element-type larger than ([UN]SIGNED-BYTE 32).
1452 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1453 * When built with the :SB-FUTEX feature, threaded builds now take
1454 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1455 for faster/more reliable mutex and condition variable support.
1456 * Incompatible change (but one you probably shouldn't have been using
1457 anyway): the interface and code for arbitrating between multiple
1458 threads in the same user session has been redesigned.
1459 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1460 error when called without an explicit environment argument.
1461 (thanks to Nikodemus Siivola)
1462 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1463 confused about their position. (thanks to Adam Warner and Gerd
1465 * bug fix: Pretty printing backquoted forms with unquotations in the
1466 argument list position of various code constructs such as LAMBDA
1467 now works correctly. (reported by Paul Dietz)
1468 * bug fix: Pretty printing unquotations no longer loses all
1469 stream position information.
1470 * optimization: performance of string output streams is now less
1471 poor for multiple small sequence writes.
1472 * optimization: performance of CSUBTYPEP in the presence of complex
1473 expressions involving CONS and NOT many times has been improved.
1474 (reported by Paul Dietz)
1475 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1477 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1478 streams for the REPL, for the debugger, and so on. (thanks to
1480 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1481 * fixed some bugs revealed by Paul Dietz' test suite:
1482 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1483 is ignored if it is too small, rather than propagating through
1484 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1485 ** extremely complex negations of CONS types were not being
1486 sufficiently canonicalized, leading to inconsistencies in
1488 ** VALUES tranformer lost derived type.
1490 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1491 * fixed a bootstrapping bug: the build process no longer assumes
1492 that the various BOOLE-related constants have the same value in
1493 host and target lisps. (noted by Paul Dietz' test suite on an
1494 SBCL binary built from CLISP)
1495 * The system can now be dynamically linked on the MIPS platform,
1496 which enables dynamic loading of foreign code from Lisp. (thanks
1497 to Ralf Baechle for discussions on the MIPS ABI)
1498 * The system now records debugging information for its own source
1499 files in a filesystem-position-independent manner, relative to
1500 the "SYS" logical host.
1501 * fixed a compiler bug: MV-LET convertion did not check references
1502 to the "max args" entry point. (reported by Brian Downing)
1503 * tweaked disassembly notes to be less confident about proclaiming
1504 some instruction as an LRA. (thanks to Brian Downing)
1505 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1506 now each have their own history, command character, and other
1507 characteristics. (thanks to David Lichteblau)
1508 * fixed some bugs revealed by Paul Dietz' test suite:
1509 ** compiler failure in compiling LOGAND expressions including a
1511 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1512 shift greater than 32.
1513 ** FLUSH-DEST did not mark blocks for type check regeneration.
1514 ** HANDLER-CASE failed to accept declarations in handler clauses
1515 in some circumstances.
1517 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1518 * New code in contrib/sb-introspect (still probably not entirely
1519 stable yet) provides some support for smart Lisp development
1520 environments like SLIME.
1521 * The conditions signalled for errors occurring when loading .fasl
1522 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1523 in a way which should help ASDF recover gracefully.
1524 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1525 stuff has been cleaned up. If you code contrib/ stuff, this might
1526 affect you, and you can look at contrib/README, contrib/STANDARDS,
1527 and/or the 0.8.4.27 diff to check.
1528 * In full calls the compiler now does not generate checks for declared
1529 argument types for all arguments.
1530 * various threading fixes
1531 ** and some experimental patches which didn't make it into
1532 the main tree for this release, but which are shipped in
1533 contrib/experimental-thread.patch as a possible fix for some
1534 failures (deadlock, spinning...) in GC-intensive multithreaded
1536 * fixed PPC build problem (source code incompatibility of different
1537 library versions): added offsetof() hackery which attempts to divine
1538 where glibc maintainers put uc_mcontext today
1539 * fixed bug 282: compiler does not trust type assertions while passing
1540 arguments to a full call.
1541 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1542 &OPTIONAL VALUES type parameter.
1543 * fix bug 214: algorithm for noting rejected templates is now more
1544 similar to that of template seletion. (also reported by rydis on
1546 * fixed bug 141b: printing backquoted information readably and prettily
1547 inserts a space where necessary.
1548 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1549 (CHARACTER 10) are now reported as errors, rather than propagated
1550 as unknown types. (reported by piso on #lisp)
1551 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1552 with values NIL and :ERROR. (thanks to Milan Zamazal)
1553 * fixed bug 191c: CLOS now does proper keyword argument checking as
1554 described in CLHS 7.6.5 and 7.6.5.1.
1555 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1556 longer raise an error; further, using a list as a for-as-arithmetic
1557 counter now raises a meaningful error.
1558 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1559 now understood by sequence creation functions such as MAKE-SEQUENCE
1561 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1562 a STRING-INPUT-STREAM.
1563 * compiler enhancement: SIGNUM is now better able to derive the type
1565 * type declarations inside WITH-SLOTS are checked. (reported by
1567 * fixed some bugs revealed by Paul Dietz' test suite:
1568 ** incorrect optimization of TRUNCATE for a positive first
1569 argument and negative second.
1570 ** compiler failure in let-convertion during flushing dead code.
1571 ** compiler failure while deriving type of TRUNCATE on an
1572 interval, containing 0.
1573 ** ASH of a negative bignum by a negative bignum count now returns
1575 ** intersection of CONS types now canonicalizes properly, fixing
1576 inconsistencies in SUBTYPEP.
1578 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1579 * incompatible change: The --disable-debugger command line
1580 option now clobbers the debugger at a more fundamental
1581 level, by redefining #'INVOKE-DEBUGGER instead of by
1582 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1583 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1584 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1585 enter the debugger REPL and then suffer recursive errors
1586 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1587 while under the new system, BREAK in --disable-debugger mode
1588 terminates the system just as an unhandled error would.
1589 * fixed compiler performance when processing loops with a step >1;
1590 * bug fix: DOCUMENTATION now retrieves generic function
1591 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1592 support has been systematized, and now supports the methods
1593 specified by ANSI, along with a default method and a method for
1594 slot documentation. (reported by Nathan Froyd)
1595 * bug fix: effective methods associated with a generic function are
1596 no longer cached over a change of that generic function's method
1597 combination. (reported by Andreas Fuchs)
1598 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1599 names a non-existent directory. (thanks to Andreas Fuchs)
1600 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1601 the PPC platform, lead to stack corruption; this has been fixed.
1602 (reported by Rainer Joswig)
1603 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1604 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1605 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1606 arithmetic in general, is now much more reliable on the PPC
1608 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1609 platform now returns the right answer.
1610 * optimization: restored some effective method precomputation in
1611 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1612 precomputation is now tunable.
1613 * optimization: compiler-internal data structure use has been
1614 reviewed, and changes have been made that should improve the
1615 performance of the compiler by about 20%.
1616 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1617 simple-base-strings and simple-bit-vectors is improved.
1618 * optimization: the optimization of 32-bit logical and arithmetic
1619 functions introduced in version 0.8.3 on the x86 has been
1620 implemented on the mips, ppc and sparc platforms; an
1621 implementation of the same facility, but for 64-bit arithmetic,
1622 has been added for the alpha.
1623 * microoptimization: the compiler is better able to make use of the
1624 x86 LEA instruction for multiplication by constants.
1625 * bug fix: in some situations compiler did not report usage of
1626 generic arithmetic in (SPEED 3) policy.
1627 * bug 145b fix: compiler used wrong type specifier while converting
1628 MEMBER-types to numeric.
1629 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1631 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1632 index argument now works properly on the Alpha platform.
1633 * bug fix: floating point exception treatment on the Alpha platform
1635 * bug fix: FILE-POSITION works much better on string input and
1636 output streams. (thanks to Nikodemus Siivola)
1637 * bug fix: many threading/garbage collection symptoms sorted.
1638 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1639 might be pseudo-atomic.
1640 * internal change: Stopping for GC is now done with signals not
1641 ptrace. GC is now done in whichever thread wanted it, instead of
1643 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1645 * bug fix: result form in DO is not contained in the implicit
1647 * incompatible change: ICR structure is changed; the value part of
1648 CONTINUATION is now called LVAR; corresponding functions are
1649 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1650 * added type deriver for ISQRT (thanks to Robert E. Brown).
1651 * bug fix: better support for loading from the command line when an
1652 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1654 * fixed some bugs revealed by Paul Dietz' test suite:
1655 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1656 ** ROUND and FROUND now give the right answer when given very
1657 small float arguments.
1658 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1660 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1661 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1662 ** LCM should return a non-negative integer.
1663 ** PARSE-INTEGER returned the index of a terminator instead of the
1664 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1665 ** PARSE-INTEGER returned an incorrect index being applied to a
1667 ** LCM with two arguments of 0 returns 0 rather than signalling
1669 ** unsigned addition of a 32-bit constant with the high bit set no
1670 longer causes an internal compiler error on the x86.
1671 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1672 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1673 with negative last argument.
1674 ** byte specifiers with zero size and position no longer cause
1675 an error during type derivation.
1676 ** bignum multiplication on the Alpha platform now returns the
1678 * porting: The system now builds on SuSE AMD64, although it still
1679 generates a 32-bit binary.
1680 * .fasl file incompatibility: The fasl file version number has
1681 been incremented (because of the changes to internal compiler
1682 data structures referred to above).
1684 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1685 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1686 more accurately, on the Darwin kernel running on PowerPC hardware.
1687 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1688 * Compiler code deletion notes now signal a condition of type
1689 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1690 an associated MUFFLE-WARNING restart.
1691 * The compiler now performs limited argument count validation of
1692 constant format strings in FORMAT, and where appropriate in ERROR,
1693 CERROR and WARN. (thanks to Gerd Moellmann)
1694 * New ASDF-INSTALL contrib can be used for automatic download and
1695 installation of third-party Lisp code from CCLAN or other sites
1697 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1698 which forces another thread to execute a function supplied by the
1700 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1701 now accept and act upon their :ELEMENT-TYPE keyword argument.
1702 (reported by Martin Atzmueller, Edi Weitz)
1703 * bug fix: FILE-POSITION now accepts position designators up to
1704 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1705 is the greater. (thanks to Patrik Nordebo)
1706 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1707 Kalle Olavi Niemitalo)
1708 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1709 printable. (reported by Eric Marsden)
1710 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1711 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1712 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1713 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1714 (thanks to Stig Sandoe)
1715 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1716 UNREAD-CHAR now correctly includes the unread character in the
1717 target sequence. (thanks to Gerd Moellmann)
1718 * bug fix in threaded builds: the system can now be suspended and
1719 resumed by shell job control with minimal disruption.
1720 * bug fixes in times and timezones >2038AD
1721 * better handling of "where is GNU make?" problem in build scripts
1722 (thanks to Nikodemus Siivola)
1723 * new optimization: inside a named function any reference to a
1724 function with the same name is considered to be a self-reference;
1725 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1726 optimization quality.
1727 * new optimization on x86: logical functions and + now have
1728 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1729 used when the result is truncated to 32 bits.
1730 * VALUES declaration is partially enabled.
1731 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1732 * fixed some bugs revealed by Paul Dietz' test suite:
1733 ** The system now obeys the constraint imposed by
1734 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1735 types form a lattice under type intersection.
1736 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1737 ** ASSOC now ignores NIL elements in an alist.
1738 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1739 and (1+ MOST-POSITIVE-FIXNUM) answers.
1740 ** The addition of a method with invalid qualifiers to a generic
1741 function does not cause an error to be signalled immediately;
1742 a warning is signalled, and the error is generated only on
1743 calling the generic function.
1744 * changed .fasl file version number, in part to add type codes for
1745 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1746 obscure ANSI requirements
1748 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1749 * fixed bug 148: failure to inline-expand a local function left
1750 garbage, confusing the compiler.
1751 * fixed bugs 3cd: structure slot readers perform type check if the
1752 slot can have an invalid value (i.e. it is either not initialized
1753 or can be written with a less specific slot writer).
1754 * bug fix: the compiler now traps array references to elements off
1755 the end of an array; previously, the bounds checking in some
1756 circumstances could go off-by-one.
1757 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1759 * type declarations for array element types now obey the description
1760 on the CLHS page "Declaration TYPE", as per discussions on
1761 sbcl-help around 2003-05-08. This means that a declaration
1762 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1763 declaration, all references to BAR will be asserted or assumed
1764 (with THE, so dependent on compiler policy) to involve objects of
1765 type FOO. Note that no such declaration is implied in
1766 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1767 * declared types of functions from the "Conditions"
1768 chapter. (reported by Paul Dietz)
1769 * bug fix: CERROR accepts a function as its first argument.
1770 * bug fix: NTH an NTHCDR accept a bignum as index
1771 arguments. (reported by Adam Warner)
1772 * optimization: character compare routines now optimize comparing
1773 against a constant character. (reported by Gilbert Baumann)
1774 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1775 argument now works properly on the MIPS platform.
1776 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1777 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1778 declared to have a functional type, does not cause an error
1780 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1781 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1782 * fixed bug: if last continuation of a deleted block has a
1783 destination, this destination should be deleted too. (reported by
1785 * fixed a bug in the bootstrap process: the host compiler's values
1786 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1787 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1788 test case from Patrik Nordebo)
1789 * improved the ability of the disassembler on the PPC platform to
1790 provide helpful disassembly notes.
1791 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1792 the class in more cases than previously.
1793 * bug fix: FILE-POSITION now understands :START and :END for
1794 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1795 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1796 8) data. (thanks to David Lichteblau)
1797 * bug fix: it is possible to add a method to a generic function
1798 without lambda list.
1799 * bug fix: reader failed to signal END-OF-FILE inside an
1800 object representation. (reported by Nikodemus Siivola)
1801 * fixed some bugs revealed by Paul Dietz' test suite:
1802 ** LAST and [N]BUTLAST should accept a bignum.
1803 ** condition slot accessors are methods.
1804 ** (VECTOR NIL) is a subtype of STRING.
1806 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1807 * minor incompatible change: some nonsensical specialized lambda
1808 lists (used in DEFMETHOD) which were previously ignored now signal
1810 * minor incompatible change: the system is now aware of the types of
1811 variables in the COMMON-LISP package, and will signal errors for
1812 most violations of these type constraints (where previously they
1813 were silently accepted).
1814 * minor incompatible change: COMPILE-FILE now uses the freedom
1815 afforded (ANSI 3.2.2.3) to use derived function types for
1816 functions defined in the same file. This also permits the system
1817 to warn on static type mismatches and function
1818 redefinition. (Currently it does not work with high DEBUG level.)
1819 * minor incompatible change: VALUES declaration is disabled.
1820 * When issuing notes, the compiler now signals a condition of type
1821 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1822 restart for use in user handlers. It is expected that the
1823 COMPILER-NOTE condition will eventually become a condition
1824 supertype to a hierarchy of note types, which will then be
1825 handleable in a similar fashion. However, at the moment, no such
1826 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1827 but it's an implementation detail, not a classification for the
1829 * Changes in type checking closed the following bugs:
1830 ** type checking of unused values (192b, 194d, 203);
1831 ** template selection based on unsafe type assertions (192c, 236);
1832 ** type checking in branches (194bc).
1833 * A short form of VALUES type specifier has ANSI meaning (it has
1834 increased the number of situations when SBCL cannot perform type
1836 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1837 keywords or constants is permissible.
1838 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1839 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1840 classes. (thanks to Antonio Martinez)
1841 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1842 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1843 argument to a pretty-print stream. Instead, it leaves any such
1844 operations to DESCRIBE-OBJECT methods.
1845 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1847 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1848 lambda lists. (thanks to David Lichteblau)
1849 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1850 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1851 (thanks to Gerd Moellmann)
1852 * fixed bug 246: increased compilation speed of long
1853 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1855 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1857 * DEFCONSTANT now throws a condition of type
1858 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1859 constant to a non-EQL value; CONTINUE and ABORT restarts
1860 respectively change and preserve the value.
1861 * fixed bug 63: The code walker, part of the implementation of CLOS,
1862 is now better at handling symbol macros.
1863 * bug fix: There is no longer an internal implementation type named
1864 CL:LENGTH. (reported by Raymond Toy)
1865 * bug fix: In macro-like defining macros/special operators the
1866 implicit block does not enclose the lambda list.
1867 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1868 enclosed in parentheses) are not suitable as type specifiers, and
1869 their use properly signals an error now.
1870 * bug fix: An argument count mismatch for a type specifier in code
1871 being compiled no longer causes an unhandled error at compile
1872 time, but signals a compile-time warning.
1873 * fixed simple vector readable printing
1874 * bug fix: DESCRIBE takes more care over whether the class
1875 precedence list slot of a class is bound before accessing it.
1876 (reported by Markus Krummenacker)
1877 * bug fix: FORMATTER can successfully compile pretty-printer format
1878 strings which use variants of the ~* directive inside.
1879 * bug fix: SEARCH now applies its TEST predicate to the elements of
1880 the arguments in the correct order. (thanks to Wolfhard Buss)
1881 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1882 in type inference. (thanks to Robert E. Brown)
1883 * bug fix: Reading in symbols with an explicit package name of ""
1884 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1885 * fixed some bugs revealed by Paul Dietz' test suite:
1886 ** NIL is now allowed as a structure slot name.
1887 ** Arbitrary numbers, not just REALs, are allowed in certain
1888 circumstances in LOOP for-as-arithmetic clauses.
1889 ** Multiple class redefinitions before slot access no longer
1890 causes a type error.
1891 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1892 association between the name and a class.
1893 ** Generic functions with non-standard method-combination and over
1894 five methods all of which return constants no longer return NIL
1895 after the first few invocations. (thanks to Gerd Moellmann)
1896 ** CALL-NEXT-METHOD with no arguments now passes the original
1897 values of the arguments, even in the presence of assignment.
1898 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1900 ** DISASSEMBLE works with closures and funcallable instances.
1901 ** ADD-METHOD now returns the generic function, not the new method.
1902 ** FIND-METHOD signals an error if the lengths of the specializers
1903 is incompatible with the generic function, even if the ERRORP
1905 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1906 which its argument is a member.
1907 ** DEFCLASS only redefines the class named by its class-name
1908 argument if that name is the proper name of the class;
1909 otherwise, it creates a new class.
1910 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1911 of the UNBOUND-SLOT condition to the name of the slot.
1912 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1913 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1914 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1915 SLOT-MAKUNBOUND in the specified fashion.
1917 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1918 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1919 cross-compilation host. As a consequence, we can now bootstrap our
1920 way up to SBCL starting with a bare gcc toolchain and human-readable
1921 source code (first the source to CLISP, then the source to SBCL).
1922 * A contributed module containing a partial implementation of the
1923 simple-streams interface has been included. (thanks to Rudi
1925 * A contributed module implementing the RFC1321 Message Digest
1926 Algorithm, known as MD5, has been included.
1927 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1928 no longer has any effect, as the code controlled by this feature
1929 has been deleted. (As far as we know, no-one has ever built using
1930 this feature, and its semantics were confused in any case).
1931 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1932 work on conditions (as required by the ANSI specification),
1933 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1934 expected behaviour on conditions. Users should note, however,
1935 that such behaviour is not required by the ANSI specification,
1936 and so use of this behaviour may render their code unportable.
1937 * fixed some bugs revealed by Paul Dietz' test suite:
1938 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1940 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1941 ** SXHASH on condition objects no longer returns NIL.
1942 ** :ALLOCATION :CLASS slots are better treated; their values are
1943 updated on class redefinition, and initforms inherited from
1944 superclasses are applied.
1945 ** REMOVE-METHOD returns its generic function argument even when
1946 no method was removed.
1947 ** SHARED-INITIALIZE now initializes the values of the requested
1948 slots, including those with :ALLOCATION :CLASS.
1949 ** ALLOCATE-INSTANCE now works on structure classes defined via
1950 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1952 ** SLOT-EXISTS-P now works on conditions, as well as structures
1954 ** MAKE-LOAD-FORM now has the required methods on
1955 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1956 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1957 keyword, and now implements the SLOT-NAMES argument.
1958 ** methods with &OPTIONAL arguments no longer allow too many
1959 arguments to be passed in the call without error.
1960 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1961 option is consistent with the required arguments of the generic
1962 function lambda list.
1963 * bug fix: REQUIRE accepts a string designator. (Thanks to
1965 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1966 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1967 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1968 * bug fix: The compiler now checks for duplicated variables in macro
1970 * bug fix: SETQ on globals returns the correct value.
1971 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1972 not a COMPILER-ERROR (followed by some other strange error on
1973 choosing the CONTINUE restart).
1974 * bug fix: make.sh and friends are now more consistent in the way that
1975 they look for GNU "make".
1977 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1978 * experimental native threads support (on x86 Linux >=2.4 only).
1979 This is not compiled in by default: you need to add :SB-THREAD to
1980 the target features. See the "Beyond ANSI" chapter of the manual
1982 * fix for longstanding nonANSIism: The old distinction between
1983 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1984 The return value from CL:FIND-CLASS is now a CLOS class, and
1985 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1986 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1987 * An interface to the MetaObject Protocol, as described in Kiczales,
1988 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1989 MIT Press, 1991, is available from the SB-MOP package.
1990 * incompatible change: the SB-PCL package should now be considered
1991 a private implementation detail, and no longer a semi-private MOP
1993 * minor incompatible change: due to rearrangement for threads, the
1994 control stack and binding stack are now allocated at arbitrary
1995 addresses instead of being hardcoded per-port. Users affected by
1996 this probably have to be doing advanced things with shared
1997 libraries, and will know who they are.
1998 * minor incompatible change: Previously, all --eval forms used were
1999 processed with READ before any of them were processed with EVAL.
2000 Now each --eval form is processed with both READ and EVAL before
2001 the next --eval form is processed. (Thus package operations like
2002 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2003 work as the user might reasonably expect.)
2004 * minor incompatible change: *STANDARD-INPUT* is now only an
2005 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2007 * minor incompatible change: Y-OR-N-P is now character-oriented,
2008 not line oriented. Also, YES-OR-NO-P now works without errors.
2009 (thanks to Antonio Martinez)
2010 * sb-aclrepl module improvements: an integrated inspector, added
2011 repl features, and a bug fix to :trace command.
2012 * Known functions, which cannot be open coded by the backend, are
2013 considered to be able to check types of their arguments. (fixing
2014 a bug report by Nathan J. Froyd)
2015 * fixed a bug in computing method discriminating functions: It is
2016 now possible to define methods specialized on classes which have
2017 forward-referenced superclasses. (thanks to Gerd Moellmann)
2018 * fixed evaluation order in optional entries (reported by Gilbert
2020 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2021 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2022 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2023 required-by-AMOP NAME argument, as well as CLASS and
2024 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2025 * fixed bug 20: DEFMETHOD can define methods using names that are
2026 not the proper names of classes to designate class specializers.
2027 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2028 Unix file descriptors, instead of blowing up. (thanks to Antonio
2030 * Garbage collection refactoring: user-visible change is that a
2031 call to the GC function during WITHOUT-GCING will not do garbage
2032 collection until the end of the WITHOUT-GCING. If you were doing
2033 this you were probably losing anyway.
2034 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2035 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2036 TYPEP the latter but not the former.
2037 * The compiler issues a full WARNING for calls to undefined functions
2038 with names from the CL package.
2039 * MAP-INTO for a vector destination is open coded. (reported by
2040 Brian Downing on c.l.l)
2041 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2042 documentation string.
2043 * fixed some bugs revealed by Paul Dietz' test suite:
2044 ** COPY-ALIST now signals an error if its argument is a dotted
2046 ** Condition slots are now accessed more correctly in the presence
2047 of multiple initargs for a given slot.
2048 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2049 exclude restarts of the same name associated with a different
2051 ** DEFCLASS of forward-referenced classes with another
2052 forward-referenced class in the superclasses list no longer
2054 ** Condition slots are now initialized once each, not multiple
2055 times. (thanks to Gerd Moellmann)
2056 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2058 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2059 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2060 arguments contain duplicated elements.
2061 ** RESTART-CASE understands local macros.
2062 ** RESTART-CASE associates exactly its own restarts with a condition.
2063 ** ENDP in safe mode checks its argument to be of type LIST.
2064 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2065 in question is unbound.
2066 ** Optimized MAKE-INSTANCE functions no longer cause internal
2067 assertion failures in the presence of duplicate initargs.
2068 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2069 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2070 propagated even in the ctor optimized implementation of
2072 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2074 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2075 * a better implementation of SXHASH on (simple) bit vectors,
2076 measured both in execution speed and in distribution of results
2077 over the positive fixnums, has been installed. Likewise, a better
2078 implementation of EQUAL for simple bit vectors is now available.
2079 * fixed CEILING optimization for a divisor of form 2^k.
2080 * fixed bug 240 (emitting extra style warnings "using the lexical
2081 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2082 by Antonio Martinez)
2083 * fixed SXHASH, giving different results for NIL depending on type
2084 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2085 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2086 variable. (found by Rolf Wester)
2087 * SBCL does not ignore type declarations for special
2088 variables. (reported by rif on c.l.l 2003-03-05)
2089 * some bug fixes in contrib/sb-aclrepl/
2090 * fixed some bugs revealed by Paul Dietz' test suite:
2091 ** a bug in the CONS type specifier, whereby the CAR and CDR
2092 types got intertwined, has been fixed;
2093 ** the type system is now able to reason about the interaction
2094 between INTEGER and RATIO types more completely;
2095 ** APPEND, [N]REVERSE and NRECONC check that those their
2096 arguments, which must be proper lists, are really so;
2097 ** An array specialized to be unable to hold elements has been
2098 implemented, as required -- yes, really -- by ANSI;
2099 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2100 on malformed property lists;
2102 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2103 * incompatible packaging change: in line with Unix convention,
2104 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2105 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2106 * REQUIRE and PROVIDE are now optionally capable of doing something
2107 useful. See the documentation string for REQUIRE.
2108 * infrastructure for a managed SBCL contrib system: contributed
2109 modules in this release include:
2110 ** the ASDF system definition facility;
2111 ** an interface to the BSD Sockets API;
2112 ** an ACL-like convenience interface to the repl;
2113 (thanks to Kevin Rosenberg)
2114 ** an implementation of ROTATE-BYTE, with efficient implementation
2116 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2117 gives a closer approximation to the right answer than previously.
2118 (thanks to Raymond Toy)
2119 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2120 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2121 optional environment arguments, as required by ANSI.
2122 * fixed bugs in other functions taking environment objects, allowing
2123 calls with an explicit NIL environment argument to be compiled
2125 * fixed bug 228: primary return values from
2126 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2127 COMPILE or FUNCTION.
2128 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2129 :TYPE will no longer signal errors on innocuous objects.
2130 * fixed bug 231b: SETQ is better at respecting type declarations in
2131 the lexical environment.
2132 * fixed a bug in DEFCLASS: classes named by symbols with no or
2133 unprintable packages can now be defined.
2134 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2135 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2136 * fixed compiler failure related to checking types of functions.
2137 (reported by Robert E. Brown)
2138 * the compiler is now much more consistent in its error-checking
2139 treatment of bounding index arguments to sequence functions: in
2140 (SAFETY 3) code, errors will be signalled in almost all cases if
2141 invalid sequence bounding indices are passed to functions defined
2142 by ANSI to operate on sequences.
2143 * fixed a bug in the build procedure: documentation of SBCL-specific
2144 packages is now preserved and available in the final Lisp image.
2145 * lifted FDEFINITION lookup out of loops in the implementation of
2146 many list operations. (thanks to Robert E. Brown)
2147 * fixed a bug in the reader: the #n# reader macro now works for
2148 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2149 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2150 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2151 E. Brown and Tony Martinez respectively)
2152 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2153 SLOT-BOUNDP: in particular, it is now able to optimize them much
2154 better, and is now not vulnerable to having packages renamed.
2155 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2156 always called when a slot is not present in an instance. (thanks
2158 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2159 longer signal an error on unreading a character following EOF on
2160 the previous constituent stream. (thanks to Tony Martinez)
2161 * fixed some bugs revealed by Paul Dietz' test suite:
2162 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2163 not just nonnegative fixnums;
2164 ** the logical bit-array operators such as BIT-AND now accept an
2165 explicit NIL for their "opt-arg" argument (to indicate a
2166 freshly-consed result bit-array);
2167 ** ELT now signals an error on an invalid sequence index in safe
2169 ** the type system is now cleverer about negations of numeric
2170 types, and consequently understands the BIGNUM and RATIO types
2172 ** the type system is now cleverer about the interaction between
2173 INTEGER and RATIO types: while bugs still remain, many more
2174 cases are accurately computed;
2175 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2176 if it is in the last clause;
2177 ** CONSTANTLY now correctly returns a side-effect-free function in
2179 ** DECLARE is no longer treated as a special-operator; in
2180 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2181 * incremented fasl file version number due to the change in the
2182 DEFSTRUCT-SLOT-DESCRIPTION structure.
2184 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2185 * minor incompatible change: code processed by the "interpreter" or
2186 EVAL now has a compilation optimization policy of (DEBUG 2)
2187 (changed from (DEBUG 1)) to improve debuggability of interactive
2188 development, and to allow the use of the debug RETURN command in
2190 * an experimental implementation of the RETURN command for the
2191 debugger has been included. (thanks to Frederik Kuivinen)
2192 * fixed bug 62: constraints were not propagated into a loop.
2193 * fixed bug in embedded calls of SORT (reported and investigated by
2195 * fixed some bugs revealed by Paul Dietz' test suite:
2196 ** printing and reading of arrays with some dimensions having
2197 length 0 (thanks to Gerd Moellmann);
2198 ** BOA constructor with &AUX argument without a default value does
2199 not cause a type error;
2200 ** CONSTANTP now returns true for all self-evaluating objects.
2202 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2203 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2204 accessors that are related by inheritance, as specified in the
2205 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2206 Valtteri Vuorikoski)
2207 * The compiler is now able to inline functions that were defined in
2208 a complex lexical environment (e.g. inside a MACROLET).
2209 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2211 * Support for the upcoming FreeBSD-5.0 release has been included.
2212 (thanks to Dag-Erling Smorgrav)
2213 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2214 effect when it is not in a toplevel context.
2215 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2216 stand a better chance of being correct. (thanks to Gerd
2218 * fixed bug in COERCE, which now signals an error on coercing a
2219 rational to a bounded real type which excludes the expected
2221 * The compiler is now able to derive types more accurately from the
2222 COERCE and COMPILE functions.
2223 * fixed bug 223: functional binding is considered to be constant
2224 only for symbols in the CL package.
2225 * fixed bug 231: SETQ did not check the type of a variable being set
2226 (reported by Robert E. Brown)
2227 * A new optimization for MAKE-INSTANCE has been included, fixing
2228 various bugs (including relating to :ALLOCATION :CLASS slots and
2229 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2231 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2232 ** As required by ANSI, LOOP now disallows anonymous collection
2233 clauses such as COLLECT I in conjunction with aggregate boolean
2234 clauses such as THEREIS (= I 1);
2235 ** LOOP now signals an error when any variable is reused in the
2236 same loop (including the potentially useful construct analogous
2237 to WITH A = 1 WITH A = (1+ A);
2238 ** IT is only a special loop symbol within the first clause of a
2239 conditional loop clause;
2240 ** LOOP with a typed iteration variable over a hashtable now
2241 signals a type error iff it should.
2242 * fixed some other bugs revealed by Paul Dietz' test suite:
2243 ** FILE-STREAM now names the class previously known as FD-STREAM;
2244 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2245 argument) no longer signals an error;
2246 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2247 of the slot symbol, rather than using the current package
2248 ((:CONC-NAME "") continues to intern the slot's name in the
2250 * incremented fasl file version number, because of the incompatible
2251 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2252 of the new implementation of DEFINE-COMPILER-MACRO.
2254 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2255 * Support for building SBCL for MIPS platforms running in
2256 little-endian mode has now been checked in, and basic
2257 functionality on said platforms verified.
2258 * minor incompatible change: PCL now records the pathname of a file
2259 in which methods and the like are defined, rather than its
2261 * minor incompatible change: TRUENAME now considers the truename of
2262 a file naming a directory to be the pathname with :DIRECTORY
2263 component indicating that directory.
2264 * minor incompatible change: a NAMED clause in the extended form of
2265 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2266 reason for the previous behaviour is unclear.
2267 * more systematization and improvement of CLOS and MOP conformance
2268 in PCL (thanks to Gerd Moellman and Pierre Mai):
2269 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2271 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2272 primary methods with no specializers;
2273 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2275 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2276 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2277 FORWARD-REFERENCED-CLASSes; error reporting on
2278 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2280 ** SXHASH on CLOS instances now uses a slot internal to the
2281 instance to return different numbers on distinct instances,
2282 while preserving the same return value through invocations of
2284 ** DEFMETHOD signals errors when methods with longer incongruent
2285 lambda lists are added to generic functions;
2286 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2287 CLASS, as specified in AMOP;
2288 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2289 on the order returned by the primary method for classes of
2290 class STANDARD-CLASS;
2291 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2292 * fixed some bugs shown by Paul Dietz' test suite:
2293 ** DOLIST puts its body in TAGBODY;
2294 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2296 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2297 value producing form;
2298 ** if more variables are given to PROGV than values, extra
2299 variables are bound and made to have no value;
2300 ** NSUBSTITUTE on list arguments gets the right answer with
2302 ** ELT signals an error of type TYPE-ERROR when the index argument
2303 is not a valid sequence index;
2304 ** LOOP signals (at macroexpansion time) an error of type
2305 PROGRAM-ERROR when duplicate variable names are found;
2306 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2307 ** FUNCALL of special-operators now cause an error of type
2309 ** PSETQ now works as required in the presence of side-effecting
2310 symbol-macro places;
2311 ** NCONC accepts any object as its last argument;
2312 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2314 ** loop-for-as-package does not require a package to be explicitely
2316 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2318 * fixed bug 166: compiler preserves "there is a way to go"
2319 invariant when deleting code.
2320 * fixed bug 172: macro lambda lists with required arguments after
2321 &REST arguments now cause an error to be signalled. (thanks to
2323 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2324 bug. (thanks to Matthew Danish)
2325 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2327 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2328 arguments are lexically rebound. (thanks to Gerd Moellmann and
2330 * fixed bug 194: error messages are now more informative when there
2331 is no primary method applicable in a call to a generic
2332 function. (thanks to Gerd Moellmann)
2333 * fixed bug in command line argument checking (thanks to Julian
2335 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2336 (thanks to Matthew Danish)
2337 * incremented fasl file version number, because of the
2338 SXHASH-related changes in the layout of CLOS data structures
2340 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2341 * minor incompatible change: The runtime (the Unix executable named
2342 "sbcl") is now much pickier about the .core files it will load.
2343 Essentially it now requires .core files to descend from the same
2344 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2345 as the runtime does. (The intent is to prevent the crashes which
2346 can occur, and which can even be reported as mysterious failures,
2347 when people patch the sources or change the build parameters
2348 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2349 sbcl and .core files.)
2350 * fixed bug: VALUES-LIST is no longer optimized away.
2351 * fixed bug 142: The FFI conversion of C string values to Lisp
2352 string values no longer conses excessively. (thanks to Nathan
2353 Froyd porting Raymond Toy's fix to CMU CL)
2354 * began to systematize and improve MOP conformance in PCL (thanks to
2355 Nathan Froyd, Gerd Moellman and Pierre Mai):
2356 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2358 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2359 ** FINALIZE-INHERITANCE is now called on class finalization;
2360 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2361 argument precedence order.
2362 * fixed bug 202: The compiler no longer fails on functions whose
2363 derived types contradict their declared type.
2364 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2365 so it can be non-toplevel.
2366 * The fasl file version number has changed (because of the new
2367 implementation of DEFMACRO).
2368 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2369 safe code, that any length requirement by their type-specifier
2370 argument is valid. The exceptions to this are described in bug
2372 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2373 functions now check the types of their inputs as required by ANSI.
2374 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2375 introduced symbol is DECLAREd to be SPECIAL.
2376 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2377 * fixed bug: :COUNT argument to sequence functions may be negative
2378 * fixed bug: body of DO-SYMBOLS may contain declarations
2379 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2380 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2381 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2382 symbol macro only once
2383 * fixed printing of call frame when argument list is unavailable
2384 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2385 * compiler no longer signals WARNING on unknown keyword
2388 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2389 * A beta-quality port to the mips architecture running Linux,
2390 based on the old CMUCL backend, has been made. It has been tested
2391 on a big-endian kernel, and works sufficiently well to be able to
2392 rebuild itself; it has not been tested in little-endian mode.
2393 * fixed an inconsistency between gencgc.c and purify.c which made
2394 dumping/loading .core files unreliable
2395 * fixed bug 120a: The compiler now deals correctly with IFs where
2396 the consequent is the same as the alternative, instead of
2397 misderiving the return type. (thanks to Alexey Dejneka)
2398 * fixed bug 113: Logical pathnames are now dumpable (the logical
2399 host is resolved at load-time, throwing an error if it is not
2401 * fixed bug 174: FORMAT's error message is slightly clearer when a
2402 non-printing character is used in a format directive.
2403 * fixed several bugs in compiler checking of type declarations, i.e.
2404 violations of the Python "declarations are assertions" principle
2405 (thanks to Alexey Dejneka)
2406 * fixed several bugs in PCL's error checking (thanks to Gerd
2408 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2410 * fixed bug in compilation of functions as first class values
2411 (thanks to Antonio Martinez-Shotton)
2412 * The compiler's handling TYPE-ERRORs which it can prove will
2413 inevitably happen at runtime has been cleaned up and corrected
2414 in several ways. (thanks to Alexey Dejneka)
2415 * improved argument type checking for various basic arithmetic
2416 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2417 lavished on them in the past that they can be compiled in many
2418 ways in different special cases
2419 * fixed bug 181: compiler checks validity of user supplied type
2421 * cleaned up code flushing in optimization: Function calls which
2422 should signal errors for safety purposes (e.g. which ANSI says
2423 should signal errors when their arguments are of incorrect type)
2424 are no longer optimized away.
2425 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2426 * incremented fasl file version number, because changes in the
2427 implementation of sequence functions like COERCE caused
2428 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2429 in old inline expansions) to become undefined. (Actually these
2430 changes were later undone, so we might very well be binary
2431 compatible with 0.7.7 after all, but leaving the version number
2432 incremented seemed like the simplest and most conservative
2435 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2436 * An alpha-quality port to the parisc architecture running Linux,
2437 based on the old CMUCL backend, has been made. This, even more so
2438 than the other backends, should be considered still a work in
2439 progress; known problems include that the Linux kernel in 64-bit
2440 mode does not propagate the correct sigcontext structure to
2441 userspace, and consequently SBCL on a parisc64 kernel will not
2443 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2444 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2445 suppressed.) Also now that the compiler is looking at declarations
2446 in the environment, it checks optimization declarations as well,
2447 and suppresses inlining when (> DEBUG SPEED).
2448 * More fixes have been made to treatment of floating point exception
2449 treatment and other Unix signals. In particular, floating point
2450 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2451 * The detection and handling of control stack exhaustion (infinite
2452 or very deeply nested recursion) has changed. Stack exhaustion
2453 detection is now done by write-protecting pages at the OS level
2454 and applies at all optimization settings; when found, a
2455 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2456 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2457 be caught using IGNORE-ERRORS.
2458 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2459 correctly when transforming recursive LABELS functions to LETs.
2460 (thanks to Alexey Dejneka)
2461 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2462 that are names of constants or global variables.
2463 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2464 alien routines with docstrings.
2465 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2466 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2468 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2469 (thanks to Eric Marsden)
2470 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2471 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2472 * Bug fix: The SPARC backend can now compile functions involving
2473 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2474 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2475 to a functions expecting a PATHNAME-DESIGNATOR.
2476 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2477 lambda lists. (thanks to Alexey Dejneka)
2478 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2479 functions so that EXPORTing the name of the class doesn't cause
2480 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2481 to Antonio Martinez for reporting this)
2482 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2483 constant-folded without failing an assertion. (thanks to Einar
2484 Floystad Dorum for reporting this)
2485 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2486 types are now dealt with more correctly by the compiler.
2487 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2488 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2489 behaviour analogous to RENAME-FILE. This puts its behaviour more
2490 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2492 * The fasl file version number has changed again. (because of the
2493 bug fix involving the names of PCL MAKE-INSTANCE functions)
2495 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2496 * bug fix: Floating point exceptions are treated much more
2497 consistently on the x86/Linux and PPC/Linux platforms.
2498 * Array initialization with :INITIAL-ELEMENT is now much faster for
2499 cases when the compiler cannot open code the array creation, but
2500 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2501 array accesses have also seen a speed increase.
2502 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2503 specified. (This was at the root of some bad interactions between
2504 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2506 * bug fix: Internal error arguments for undefined functions are now
2507 computed correctly on the PPC/Linux platform.
2508 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2509 Raymond Toy's patch for CMU CL)
2510 * Support for the Solaris 9 operating environment has been included
2511 (thanks to Daniel Merritt)
2512 * A very ugly but hopefully complete draft of the missing FFI chapter
2513 of the manual has been created by reformatting the corresponding
2514 CMU CL manual chapter into (currently very ugly and incoherent)
2515 DocBook and bringing it up to date for SBCL behavior. Thus, the
2516 manual is now essentially complete, at least by my extreme
2517 once-and-only-once standards, whereby it's acceptable to refer to
2518 the doc strings of SB-EXT functions as the primary documentation.
2519 * The fasl file version number has changed again, due to cleanup of
2520 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2521 is no longer a static symbol.)
2523 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2524 * SBCL now builds with OpenMCL (version 0.12) as the
2525 cross-compilation host; also, more progress has been made toward
2526 bootstrapping under CLISP.
2527 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2529 * bug 158 fixed: The compiler can now deal with integer loop
2530 increments different from 1; fixing this turned out also to fix
2532 * bug 169 fixed: no more bogus warnings about using lexical bindings
2533 despite the presence of perfectly good SPECIAL declarations (thanks
2534 to David Lichteblau)
2535 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2536 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2537 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2539 * bug 180 fixed: Method combination specifications no longer ignore
2540 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2541 * bug fix: Structure type predicate functions now check their argument
2542 count as they should.
2543 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2544 correctly. (thanks to Pierre Mai)
2545 * minor incompatible change: The --noprogrammer option is deprecated
2546 in favor of the new --disable-debugger option, which is very similar.
2547 (The major difference is that it takes effect at a slightly different
2548 time at startup, causing handling of errors in --sysinit and
2549 --userinit files will be affected differently.) The
2550 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2551 been added to allow this functionality to be controlled from ordinary
2552 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2553 maintainers, who might want to run non-interactive scripts to
2554 build SBCL cores which will later be used interactively.)
2555 * minor incompatible change: The LOAD function no longer, when given
2556 a wild pathname to load, loads all files matching that pathname.
2557 Instead, an error of type FILE-ERROR is signalled.
2559 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2560 * bug 147 fixed: The compiler preserves its block link/count
2561 invariants more correctly now so that it doesn't crash. (thanks
2563 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2565 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2566 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2567 no-value-for-this-character as NIL (as specified by ANSI).
2568 * HOST-NAMESTRING on physical pathnames now returns a string that is
2569 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2570 (thanks to Christophe Rhodes)
2571 * The Alpha port handles icache flushing more correctly. (thanks to
2573 * More progress has been made toward bootstrapping under CLISP. (thanks
2574 to Christophe Rhodes)
2575 * The fasl file format has changed again, because dynamic loading
2576 on OpenBSD (which has non-ELF object files) motivated some cleanups
2577 in the way that foreign symbols are transformed and passed around.
2578 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2579 is no longer treated as whitespace by the reader, but instead as
2580 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2581 |A
\7fB|, instead of A as it used to.
2583 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2584 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2585 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2586 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2587 this as of 0.7.1.45, but was left out of the previous news section
2588 (thanks to Dan Barlow)
2589 * SBCL now runs on the Solaris operating system on SPARC architectures
2590 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2591 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2592 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2593 * SPARC backend cleanups, allowing builds of cores optimized for V8
2594 and V9 SPARCS, and also emission of code targeted to a particular
2595 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2597 * SBCL is closer to bootstrapping under CLISP, thanks to various
2598 fixes by Christophe Rhodes.
2599 * The fasl file format has changed again, to allow the compiler's
2600 INFO database to support symbol macros.
2601 * The user manual (in doc/) is formatted into HTML more nicely.
2602 (thanks to coreythomas)
2603 * The system is smarter about SUBTYPEP relationships, especially
2604 those involving NOT types (including types such as ATOM which are
2605 represented internally using NOT types). Thus SUBTYPEP is less
2606 likely to return (VALUES NIL NIL) in general, and in particular
2607 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2609 * The fasl file format has changed again, because the internal
2610 representation of types now includes a new slot to support the new
2611 SUBTYPEP-of-NOT-types logic.
2612 * (not a change in the main branch of SBCL, but a related prototype
2613 which can hopefully be merged into the main branch of SBCL in the
2614 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2615 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2616 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2617 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2618 dispatch macro character combinations. (thanks to Alexey Dejneka)
2619 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2620 that superficially look like logical namestrings correctly.
2621 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2623 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2624 default supplied-p) for &optional and &key arguments. (thanks to
2627 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2628 * incompatible change: The compiler is now less aggressive about
2629 tail call optimization, doing it only when (> SPACE DEBUG) or
2630 (> SPEED DEBUG). (This is an incompatible change because there are
2631 programs which relied on the old CMU-CL-style behavior to optimize
2632 away their unbounded recursion which will now die of stack overflow.)
2633 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2634 for non-GENCGC systems has been increased to 20M (since that
2635 seems much closer to the likely performance optimum for modern
2636 systems than the old 4M value was)
2637 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2638 and *DEBUG-PRINT-LEVEL*
2639 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2640 of CMU CL's support for SPARC, and various endianness and other
2641 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2642 * new syntactic sugar for the Unix command line: --load foo.bar is now
2643 an alternate notation for --eval '(load "foo.bar")'.
2645 ** The system now detects stack overflow and handles it gracefully,
2646 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2647 optimization settings. (This is a good thing in general, and
2648 its introduction in this version should be particularly timely
2649 for anyone whose code fails because of suppression of tail
2651 ** The system now hunts for the C variable "environ" in a more
2652 devious way, to avoid segfaults when the C library version differs
2653 between compile time and run time. (thanks to Christophe Rhodes)
2654 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2655 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2656 ** The compiler no longer issues bogus style warnings for undefined
2657 classes in the same source file as the DEFCLASSes which defined
2658 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2660 ** fixes in CONDITION class precedence list for undefined function
2661 errors (thanks to Alexei Dejneka)
2662 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2663 correctly. (thanks to Dan Barlow)
2664 ** portability fixes aiming at bootstrapping under CLISP (thanks
2665 to Dave McDonald and Christophe Rhodes)
2666 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2667 ** fixes in type translation and and type inference (thanks to
2669 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2670 ** various fixes in the new ports (thanks to Dan Barlow)
2671 * several changes related to debugging:
2672 ** suppression of tail recursion, as noted above
2673 ** stack overflow detection, as noted above
2674 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2675 is now the default. (For some time encapsulation has been more
2676 reliable than the breakpoint-based :ENCAPSULATE NIL
2677 implementation, at least on X86 systems; and I just noticed that
2678 encapsulation also seems closer to the spirit of the ANSI
2681 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2683 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2684 up properly again. (There was a packaging bug in 0.7.0 which
2685 left their definitions in SB-SYS::LOAD-FOREIGN and
2686 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2687 vital for most things which interface to C-level interfaces,
2688 like extensions working with sockets or databases or
2689 Perl-compatible regexes or whatever, and the need to fix
2690 this bug is the main reason that 0.7.1 was released so
2692 ** DEFGENERIC is now choosier about the methods it redefines, so that
2693 reLOADing a previously-LOADed file containing DEFGENERICs does
2694 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2695 cycle now works as it should. (thanks to Alexey Dejneka)
2696 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2697 fixed by Christophe Rhodes. (He actually submitted this patch
2698 months ago, and I delayed until after 0.7.0.)
2699 ** Bug 111 (internal compiler confusion about runtime checks on
2700 FUNCTION types) was fixed by Alexey Dejneka.
2701 * Some internal cleanups (getting rid of variables which aren't
2702 needed now that the byte interpreter is gone) caused the fasl
2703 file format number to change again.
2705 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2706 * major incompatible change: The default fasl file extension, i.e. the
2707 default extension for files produced by COMPILE-FILE, has changed
2708 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2710 ** There are many changes in the implementation of the compiler.
2711 SBCL is now essentially a compiler-only implementation of ANSI
2712 Common Lisp. EVAL still "interprets" a few special cases, but
2713 almost all the interesting cases are handled by creating
2714 a LAMBDA expression, calling COMPILE on it, then calling
2715 FUNCALL on the result.
2716 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2717 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2718 Since the code is newer, there might still be some new bugs
2719 (though not as many as before Martin Atzmueller's fixes:-). But
2720 the new code is substantially simpler and clearer, and hopefully
2721 any remaining bugs will be simpler, less fundamental, and more
2722 fixable then the bugs in the old code.
2723 ** The revised compiler is still a little unsteady on its feet.
2725 *** The debugging information it produces (particularly the names
2726 of FUNCTION objects) is sometimes much less useful than what
2727 the old compiler produced.
2728 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2729 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2730 within a MACROLET) has been temporarily weakened.
2731 ** There are new compiler optimizations for various functions:
2732 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2733 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2734 *** the math functions TRUNCATE, FLOOR, and CEILING
2735 *** the function-of-all-trades COERCE
2736 Mostly these should be transparent, but there's one
2737 potentially-annoying problem (bug 117): when the compiler
2738 inline-expands a function and does type analysis on the result,
2739 it can create control paths which have type mismatches, and
2740 when it can't prove that those control paths aren't taken,
2741 it will issue WARNINGs about the type mismatches. This is
2742 a particular problem in practice for the new sequence functions.
2743 It's not clear how this should be fixed, and for now, a
2744 workaround is given in the entry for 117 in the BUGS file.
2745 ** (Because of the interaction between the two previous items --
2746 occasional inlining problems and new inline expansions -- some
2747 of the new sequence function optimizations won't really kick in
2748 completely until debugging information, and then inlining, are
2749 straightened out in some future version.)
2750 * minor incompatible changes:
2751 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2752 DIRECTORY no longer implicitly promotes NIL slots of its
2753 pathname argument to :WILD. In particular, when you ask for the
2754 contents of a directory (which you used to be able to do without
2755 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2756 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2757 ** changes in behavior that ANSI explicitly defines to be
2758 implementation dependent:
2759 *** The new compiler-only implementation still conforms with ANSI,
2760 but acts a little different than before. Besides the obvious
2761 changes in performance tradeoffs (that the cost per form passed
2762 to EVAL has gone up, and the cost per form executed by EVAL
2763 has gone down), the behavior of the system changes a little
2764 because there are no longer any interpreted function objects.
2765 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2766 e.g. doing COMPILE on the output of interactive DEFUN is
2768 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2770 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2771 has changed from (:CL) to NIL.
2772 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2773 e.g. #\Newline and #\Tab, don't have names specified in the
2774 ANSI Common Lisp standard, is now based on their ASCII symbolic
2775 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2776 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2777 are no longer used for output.
2778 ** changes in internal implementation constants:
2779 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2780 4 million. (If your application spends a lot of time GCing and
2781 you have a lot of RAM, you might want to experiment with
2782 increasing it even more.)
2783 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2784 However, almost all old code should still continue to work without
2785 immediate update, as SB-C-CALL is now a (deprecated) nickname
2787 ** Old operator names in the style DEF-FOO are now deprecated in
2788 favor of new corresponding names DEFINE-FOO, for consistency with
2789 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2790 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2791 internal symbols, but a few supported extensions like
2792 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2793 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2794 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2795 etc. as you get deeper into recursive calls to the debugger
2796 command loop, instead of the old "5]", "5]]", "5]]]"
2797 sequence. (I was motivated to do this when squabbles between
2798 ILISP and SBCL left me very deeply nested in the debugger. In the
2799 short term, this change will probably provoke more ILISP/SBCL
2800 squabbles, but hopefully it will be an improvement in the long run.)
2801 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2802 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2803 function, since declaiming return types involving aliens is
2804 (1) annoyingly messy to do by hand and (2) vital to efficient
2805 compilation of code which calls such functions.
2806 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2807 longer reexported by the SB-EXT package. They're solely useful
2808 for alien code, so it seems more logical that you should get
2809 them from the SB-ALIEN package, not in SB-EXT.
2810 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2811 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2812 features. Instead, the code that they used to control is always
2813 built into the system.
2814 * many other bug fixes
2815 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2816 advantage of the new EVAL-WHEN stuff and to clean them up in
2817 general, and they are now more ANSI-compliant in a number of
2818 ways. Martin Atzmueller is responsible for a lot of this.
2819 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2821 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2822 *** correct ERROR type for various file operations
2823 *** some fixes for Lisp streams
2824 *** DEFMETHOD syntax checking
2825 *** changing old weird representation of debug information as
2826 strings (which, among their other deficiencies, don't transform
2827 correctly when you rename packages, and don't change their
2828 print representation when you change things like *PACKAGE*
2829 and *PRINT-LENGTH*) to symbols and lists of symbols
2830 He also made several improvements and fixed several bugs in DESCRIBE.
2831 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2833 *** misbehavior of WRITE-STRING/WRITE-LINE
2834 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2835 and several other LOOP problems as well
2836 *** DIRECTORY when similar filenames are present
2837 *** DEFGENERIC with :METHOD options
2838 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2839 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2840 *** argument ordering in FIND with :TEST option
2841 *** mishandled package designator argument in APROPOS-LIST
2842 *** various problems in the backquote readmacro
2843 *** a bug in APROPOS
2844 *** probably some others that I'm not describing very well here,
2845 since the CVS log documents them by reference to sbcl-devel
2846 messages, and the SourceForge archives aren't working well.:-(
2847 ** Dan Barlow improved the Alpha port (and is making progress on the
2848 PPC port, for those of you who think different).
2849 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2850 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2851 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2852 continues to work on the SPARC port (for those of you in a position
2853 to look down upon our little PC-compatible boxes from a great height).
2854 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2855 and uses that copy, rather than the current dynamic value, when
2856 it's trying to decide whether to truncate output. Thus e.g.
2857 (let ((*print-lines* 50))
2858 (pprint-logical-block (stream nil)
2860 (let ((*print-lines* 8))
2861 (print (aref possiblybigthings i) stream)))))
2862 should now truncate the logical block only at 50 lines, instead of
2863 often truncating it at 8 lines, as it did before.
2864 * The doc/cmucl/ directory, containing old CMU CL documentation
2865 from the time of the fork, is no longer part of the base system.
2866 SourceForge has shut down its anonymous FTP service, and with it
2867 my original plan for distributing the old CMU CL documentation
2868 there. For now, if you need these files you can download an old
2869 SBCL source release and extract them from it.
2870 * The fasl file version number changed again, for dozens of reasons,
2871 some of which are apparent above.
2873 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2874 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2875 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2876 patch, so that DISASSEMBLE output is much nicer.
2877 * The code in the SB-PROFILE package now seems reasonably stable.
2878 I still haven't decided what the final interface should look like
2879 (I'd like PROFILE to interact cleanly with TRACE, since both
2880 facilities use function encapsulation) but if you have a need
2881 for profiling now, you can probably use it successfully with
2882 the current CMU-CL-style interface.
2883 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2884 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2885 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2886 link now returns the dangling link itself, and for similar
2887 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2888 link itself. (In these cases the old code signalled an error and
2889 looped endlessly, respectively.) Thus, DIRECTORY now works even
2890 in the presence of dangling and cyclic symbolic links.
2891 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2892 is now a supported extension again, since the consensus on
2893 sbcl-devel was that it can be useful for ordinary development
2894 work, not just for debugging SBCL itself.
2895 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2896 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2897 that currently-defined functions might be redefined later with
2898 different return types.
2899 * Hash tables can be printed readably, as inspired by CMU CL code
2900 of Eric Marsden and SBCL code of Martin Atzmueller.
2901 * better error handling in CLOS method combination, thanks to
2902 Martin Atzmueller porting Pierre Mai's CMU CL patches
2903 * more overflow fixes for >16Mbyte I/O buffers
2904 * A bug in READ has been fixed, so that now a single Ctrl-D
2905 character suffices to cause end-of-file on character streams.
2906 In particular, now you only need one Ctrl-D at the command
2907 line (not two) to exit SBCL.
2908 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2910 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2911 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2912 does the right thing.
2913 * The implementation of some type tests, especially for CONDITION
2914 types, is now tidier and maybe faster, due to CMU CL code
2915 originally by Douglas Crosher, ported by Martin Atzmueller.
2916 * Some math functions have been fixed, and there are new
2917 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2918 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2919 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2920 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2921 likely to be built into sbcl-0.7.0.)
2922 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2923 CL, ported to SBCL by Martin Atzmueller)
2924 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2925 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2926 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2927 * There's a new slam.sh hack to shorten the edit/compile/debug
2928 cycle for low-level changes to SBCL itself, and a new
2929 :SB-AFTER-XC-CORE target feature to control the generation of
2930 the after-xc.core file needed by slam.sh.
2931 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2932 COMPILE-FILE is no longer supported, so that now every function
2933 gets an entry point, so that block compilation looks a little
2934 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2935 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2936 returns the number of bytes consed since the system started,
2937 rather than the number consed since the first time the function
2938 was called. (The new definition parallels ANSI functions like
2939 CL:GET-INTERNAL-RUN-TIME.)
2940 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2941 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2942 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2943 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2944 * Fasl file version numbers are now independent of the target CPU,
2945 since historically most system changes which required version
2946 number changes have affected all CPUs equally. Similarly,
2947 the byte fasl file version is now equal to the ordinary
2950 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2951 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2952 is no longer recognized. I apologize for this, because it was
2953 listed in SB-EXT as a supported extension, but I found that
2954 its existing behavior was poorly specified, as well as incorrectly
2955 specified, and it looked like too much of a mess to straighten it
2956 out. I have enough on my hands trying to get ANSI stuff to work..
2957 * many patches ported from CMU CL by Martin Atzmueller, with
2958 half a dozen bug fixes in pretty-printing and the debugger, and
2959 half a dozen others elsewhere
2960 * fixed bug 13: Floating point infinities are now supported again.
2961 They might still be a little bit flaky, but thanks to bug reports
2962 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2963 as flaky as they were.
2964 * The --noprogrammer command line option is now supported. (Its
2965 behavior is slightly different in detail from what the old man
2966 page claimed it would do, but it's still appropriate under the
2967 same circumstances that the man page talks about.)
2968 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2969 are now supported, and enabled by default. Thus, the compiler can
2970 handle many floating point and complex operations much less
2971 inefficiently. (Thus e.g. you can implement a complex FFT
2973 * The compiler now detects type mismatches between DECLAIM FTYPE
2974 and DEFUN better, and implements CHECK-TYPE more correctly, and
2975 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2976 thanks to patches from Martin Atzmueller.
2977 * various fixes to make the cross-compiler more portable to
2978 ANSI-conforming-but-different cross-compilation hosts (notably
2979 Lispworks for Windows, following bug reports from Arthur Lemmens)
2980 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2981 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2982 fixed thanks to Pierre Mai's CMU CL patches.
2983 * fixes to keep the system from overflowing internal counters when
2984 it tries to use i/o buffers larger than 16M bytes
2985 * fixed bug 45a: Various internal functions required to support
2986 complex special functions have been merged from CMU CL sources.
2987 (When I was first setting up SBCL, I misunderstood a compile-time
2988 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2989 * improved support for type intersection and union, fixing bug 12
2990 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2991 more obscure bugs as well
2992 * some steps toward byte-compiling non-performance-critical
2993 parts of the system, courtesy of patches from Martin Atzmueller
2994 * Christophe Rhodes has made some debian packages of sbcl at
2995 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2996 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2997 stable, but are nonetheless usable. When he's ready, I'd be happy
2998 to add them to the SourceForge "File Releases" section. (And if
2999 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3000 * new fasl file format version number (because of changes in
3001 internal representation of (OR ..) types to accommodate the new
3002 support for (AND ..) types, among other things)
3004 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3005 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3007 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3009 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3010 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3011 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3012 * DESCRIBE now works on structure objects again.
3013 * Most function call argument type mismatches are now handled as
3014 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3015 know whether the function will be redefined before the call is
3016 executed. (The compiler could flag local calls with full WARNINGs,
3017 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3018 it doesn't keep track of enough information to know whether calls
3019 are local in this sense.)
3020 * Compiler output is now more verbose, with messages truncated
3021 later than before. (There should be some supported way for users
3022 to override the default verbosity, but I haven't decided how to
3023 provide it yet, so this behavior is still controlled by the internal
3024 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3025 src/compiler/ir1util.lisp.)
3026 * Fasl file format version numbers have increased again, because
3027 support for the Gray streams extension changes the layout of the
3028 system's STREAM objects.
3029 * The Gray subclassable streams extension now works, thanks to a
3030 patch from Martin Atzmueller.
3031 * The full LOAD-FOREIGN extension (not just the primitive
3032 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3033 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3034 but like most other programs, it defaults to copying the Unix
3035 environment from the original process instead of starting the
3036 new process in an empty environment.
3037 * Extensions which manipulate the Unix environment now support
3038 an :ENVIRONMENT keyword option which doesn't smash case or
3039 do other bad things. The CMU-CL-style :ENV option is retained
3040 for porting convenience.
3041 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3042 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3044 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3046 * A patch from Martin Atzmueller seems to have solved the SIGINT
3047 problem, and as far as we know, signal-handling now works cleanly.
3048 (If you find any new bugs, please report them!)
3049 * The system no longer defaults Lisp source file names to types
3050 ".l", ".cl", or ".lsp", but only to ".lisp".
3051 * The compiler no longer uses special default file extensions for
3052 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3053 seems to expect a single default extension for all compiled code,
3054 and there's no compelling reason to try to stretch the standard
3055 to allow two different extensions.) Instead, byte-compiled files
3056 default to the same extension as native-compiled files.
3057 * Fasl file format version numbers have increased again, because
3058 a rearrangement of internal implementation packages made some
3059 dumped symbols in old fasl files unreadable in new cores.
3060 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3061 many fewer weird special cases.
3062 * Bug #17 (differing COMPILE-FILE behavior between logical and
3063 physical pathnames) has been fixed, and some related misbehavior too,
3064 thanks to a patch from Martin Atzmueller.
3065 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3066 by Tim Moore, ported to SBCL by Martin Atzmueller.
3067 * Martin Atzmueller fixed several filesystem-related problems,
3068 including bug #36, in part by porting CMU CL patches, which were
3069 written in part by Paul Werkowski.
3070 * More compiler warnings in src/runtime/ are gone, thanks to
3071 more patches from Martin Atzmueller.
3072 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3075 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3077 * DESCRIBE now works on CONDITION objects.
3078 * The debugger now handles errors which arise when trying to print
3079 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3081 * The build system now uses an additional file, customize-target-features.lisp,
3082 to allow local modifications to the target *FEATURES* list. (The point of
3083 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3084 debugging features enabled, without having to worry about propagating it
3085 into everyone's system when I do a "cvs update".) When no
3086 customize-target-features.lisp file exists, the target *FEATURES* list
3087 should be constructed the same way as before.
3088 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3089 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3090 warn about behavior which is undefined under the ANSI spec. Note: This
3091 is specified by ANSI, but it's not very popular with programmers.
3092 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3093 macro in the SBCL sources for an example of a workaround which you
3094 might use to make portable ANSI-standard code which does what you want.
3095 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3096 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3097 when it's used as a non-toplevel form. (This is required in order
3098 to implement the DEFCONSTANT-EQX macro.)
3099 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3100 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3101 funny things, due to bugs in the implementation of EVAL-WHEN
3102 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3103 non-toplevel DEFCONSTANTs are uncommon.)
3104 * The core file version number and fasl file version number have been
3105 incremented, because the old noncompliant DEFCONSTANT behavior involved
3106 calling functions which no longer exist, and because I also took the
3107 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3108 * fixed bug 1 (error handling before read-eval-print loop starts), and
3109 redid debugger restarts and related debugger commands somewhat while
3111 ** The QUIT debugger command is gone, since it did something
3112 rather different than the SB-EXT:QUIT command, and since it never
3113 worked properly outside the main toplevel read/eval/print loop.
3114 Invoking the new TOPLEVEL restart provides the same functionality.
3115 ** The GO debugger command is also gone, since you can just invoke
3116 the CONTINUE restart directly instead.
3117 ** The TOP debugger command is also gone, since it's redundant with the
3118 FRAME 0 command, and since it interfered with abbreviations for the
3120 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3122 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3123 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3124 some porting work by Martin Atzmueller)
3125 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3126 with a boatload of patches which he ported to SBCL. Now that those
3128 ** The system tries to make sure that its low-priority messages
3129 are prefixed by semicolons, to help people who like to use
3130 syntax highlighting in their ILISP buffer. (This patch
3131 was originally due to Raymond Toy.)
3132 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3133 patches originally written by Raymond Toy.
3134 ** The compiler understands coercion between single-value and
3135 multiple-VALUES type expressions better, getting rid of some very
3136 weird behavior, thanks to patches originally by Robert MacLachlan
3137 and Douglas Crosher.
3138 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3139 lambda lists, thanks to a patch originally by Pierre Mai.
3140 ** The system no longer bogusly warns about "abbreviated type
3142 ** The compiler gets less confused by inlining and RETURN-FROM,
3143 thanks to some patches originally by Tim Moore.
3144 ** The system no longer hangs when dumping circular lists to fasl
3145 files, thanks to a patch originally from Douglas Crosher.
3146 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3147 undefined function error.
3148 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3149 in structure slot initforms) for now, documented workaround instead:-|
3150 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3151 * fixed bug 5: added stubs for various Gray stream functions called
3152 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3153 installed, at least appropriate type errors are generated
3154 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3155 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3156 consistently in DEFMETHOD forms.
3157 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3158 it doesn't seem to affect SBCL after all
3159 * The C runtime system now builds with better optimization and many
3160 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3162 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3164 * The system is now under CVS at SourceForge (instead of the
3165 CVS repository on my home machine).
3166 * The new signal handling code has been tweaked to treat register
3167 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3168 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3169 downstream, e.g. in debug-int.lisp, has implicit dependencies
3170 on the unsignedness of integer representation of machine words,
3171 and that caused the system to bomb out with infinite regress
3172 when trying to recover from type errors involving signed values,
3173 e.g. (BUTLAST '(1 2 3) -1).)
3174 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3175 (This was one of the bugs Peter Van Eynde reported back in July.)
3176 * The system now uses code inspired by Colin Walters' O(N)
3177 implementation of MAP (from the cmucl-imp@cons.org mailing
3178 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3179 inline the MAP operation, and there is more than one
3180 sequence argument to the MAP call (so that it can't just
3181 do ETYPECASE once and for all based on the type of the
3182 single sequence argument). (The old non-inline implementation
3183 of the general M-argument sequence-of-length-N case required
3184 O(M*N*N) time when any of the sequence arguments were LISTs.)
3185 * The QUIT :UNIX-CODE keyword argument has been renamed to
3186 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3187 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3189 * Raymond Wiker's patches to port dynamic loading from Linux to
3190 FreeBSD have been added.
3191 * The BUGS file is now more nearly up to date, thanks in large part
3192 to Martin Atzmueller's review of it.
3193 * The debugger now flushes standard output streams before it begins
3194 its output ("debugger invoked" and so forth).
3195 * The core version number and fasl file version number have both
3196 been incremented, because of incompatible changes in the layout
3198 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3199 used to not be called for a saved Lisp image.)
3200 * Martin Atzmueller's version of a patch to fix a compiler crash,
3201 as posted on sbcl-devel 13 September 2000, has been installed.
3202 * Instead of installing Martin Atzmueller's patch for the
3203 compiler transform for SUBSEQ, I deleted the compiler transform,
3204 and transforms for some similar consing operations.
3205 * A bug in signal handling which kept TRACE from working on OpenBSD
3207 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3208 away by constant folding
3209 * The system now defines its address space constants in one place
3210 (in the Lisp sources), and propagates them automatically elsewhere
3211 (through GENESIS and the sbcl.h file). Therefore, patching the
3212 address map is less unnecessarily tedious and error-prone. The
3213 Lisp names of address space constants have also been systematized.
3214 * CVS tags like dollar-Header-dollar have been removed from
3215 the sources, because they have never saved me trouble and
3216 they've been source of trouble working with patches and other
3217 diff-related operations.
3218 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3219 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3221 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3223 * The system has been ported to OpenBSD.
3224 * The system now compiles with a simple "sh make.sh" on the systems
3225 that it's supported on. I.e., now you no longer need to tweak
3226 text in the target-features.lisp-expr and symlinks in src/runtime/
3227 by hand, the make.sh takes care of it for you.
3228 * The system is no longer so grossly inefficient when compiling code
3229 involving vectors implemented as general (not simple) vectors (VECTOR T),
3230 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3231 which dares to use the various sequence functions on non-simple
3232 vectors, takes less of a performance hit.
3233 * There is now a primitive type predicate VECTOR-T-P
3234 to test for the (VECTOR T) type, so that e.g.
3235 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3236 can now be compiled with some semblance of efficiency. (The old code
3237 turned the type declaration into a full call to %TYPEP at runtime!)
3238 * AREF on (VECTOR T) is still not fast, since it's still compiled
3239 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3240 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3241 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3242 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3243 without making full calls to SUBTYPEP at runtime.
3244 (Some analogous efficiency issues for non-simple vectors specialized to
3245 element types other than T, or for non-simple multidimensional arrays,
3246 have not been addressed. They could almost certainly be handled the
3247 same way if anyone is motivated to do so.)
3248 * The changes in array handling break binary compatibility, so
3249 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3250 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3251 instead of (VALUES T T).
3252 * By following the instructions that Dan Barlow posted to sbcl-devel
3253 on 2 July 2000, I was able to enable primitive dynamic object
3254 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3255 functionality is not implemented (since it calls ld to resolve
3256 library references automatically, requiring RUN-PROGRAM for its
3257 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3258 try to resolve library references) is now supported.
3259 * The system now flushes the standard output streams when it terminates,
3260 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3261 them at several other probably-convenient times, e.g. in each pass of
3262 the toplevel read-eval-print loop, and after evaluating a form given
3263 as an "--eval" command-line option. (These changes were motivated by a
3264 discussion of stream flushing issues on cmucl-imp in August 2000.)
3265 * The source transform for TYPEP of array types no longer assumes
3266 that an array whose element type is a not-yet-defined type
3267 is implemented as an array of T, but instead punts, so that the
3268 type will be interpreted at runtime.
3269 * There is now some support for cross-compiling in make.sh: each of
3270 the phases of make.sh has its own script. (This should be transparent
3271 to people doing ordinary, non-cross-compile builds.)
3272 * Since my laptop doesn't have hundreds of megabytes of memory like
3273 my desktop machine, I became more motivated to do some items on
3274 my to-do list in order to reduce the size of the system a little:
3275 ** Arrange for various needed-only-at-cold-init things to be
3276 uninterned after cold init. To support this, those things have
3277 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3278 symbols with such names are now uninterned after cold init).
3279 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3280 fasl files for cold load.
3281 ** Remove the old compiler structure pooling code (which used to
3282 be conditional on the target feature :SB-ALLOC) completely.
3283 ** Redo the representation of some data in cold init to be more compact.
3284 (I also looked into supporting byte compiled code at bootstrap time,
3285 which would probably reduce the size of the system a lot, but that
3286 looked too complicated, so I punted for now.)
3287 * The maximum signal nesting depth in the src/runtime/ support code has
3288 been reduced from 4096 to 256. (I don't know any reason for the very
3289 large old value. If the new smaller value turns out to break something,
3290 I'll probably just bump it back up.)
3291 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3293 * Many, many bugs reported by Peter Van Eynde have been added to
3294 the BUGS list; some have even been fixed.
3295 * While enabling dynamic object file loading, I tried to make the
3296 code easier to understand, renaming various functions and variables
3297 with less ambiguous names, and changing some function calling
3298 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3299 * While trying to figure out how to do the OpenBSD port, I tried to
3300 clean up some of the code in src/runtime/. In particular, I dropped
3301 support for non-POSIX signal handling, added various comments,
3302 tweaked the code to reduce the number of compilation warnings, and
3303 renamed some files to increase consistency.
3304 * To support the new automatic configuration functionality in make.sh,
3305 the source file target-features.lisp-expr has been replaced with the
3306 source file base-target-features.lisp-expr and the machine-generated
3307 file local-target-features.lisp-expr.
3308 * fixed a stupid quoting error in make.sh so that using CMU CL
3309 "lisp -batch" as cross-compilation host works again
3311 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3313 * DESCRIBE no longer tries to call itself recursively to describe
3314 bound/fbound values, so that it no longer fails on symbols which are
3315 bound to themselves (like keywords, T, and NIL).
3316 * DESCRIBE now works on generic functions.
3317 * The printer now prints less-screwed-up representations of closures
3318 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3319 * A private symbol is used instead of the :EMPTY keyword previously
3320 used to mark empty slots in hash tables. Thus
3321 (DEFVAR *HT* (MAKE-HASH-TABLE))
3322 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3323 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3324 now does what ANSI says that it should. (You can still get
3325 similar noncompliant behavior if bang on the hash table
3326 implementation with all the symbols you get back from
3327 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3328 This breaks binary compatibility, since tests for equality to
3329 :EMPTY are wired into things like the macroexpansion of
3330 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3332 * There's now a minimal placeholder implementation for CL:STEP,
3333 as required by ANSI.
3334 * An obscure bug in the interaction of the normal compiler, the byte
3335 compiler, inlining, and structure predicates has been patched
3336 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3337 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3339 * Missing ordinary arguments in a macro call are now detected even
3340 when the macro lambda list contains &KEY or &REST.
3341 * The debugger no longer complains about encountering the top of the
3342 stack when you type "FRAME 0" to explicitly instruct it to go to
3343 the top of the stack. And it now prints the frame you request even
3344 if it's the current frame (instead of saying "You are here.").
3345 * As specified by ANSI, the system now always prints keywords
3346 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3348 * The default initial SIZE of HASH-TABLEs is now smaller.
3349 * Type information from CLOS class dispatch is now propagated
3350 into DEFMETHOD bodies, so that e.g.
3351 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3353 is now basically equivalent to
3354 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3355 (DECLARE (TYPE SINGLE-FLOAT X))
3357 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3358 operation, without having to do run-time type dispatch.
3359 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3360 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3361 * Since I don't seem to be making big file reorganizations very often
3362 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3363 is overflowing my ability to conveniently back them up), I've finally
3364 checked the system into CVS. (The CVS repository is on my home system,
3365 not at SourceForge -- putting it on SourceForge might come later.)
3366 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3367 high-level GC-NOTIFY-FOO functions send their output. (There's
3368 still very little control of where low-level verbose GC functions
3369 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3370 less than it used to -- the GC-NOTIFY-FOO functions are now under
3371 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3372 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3373 in only one place in the source code, and propagates it automatically
3374 everywhere that it's needed. Thus e.g. when I bump the version from
3375 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3376 * The C source files now include boilerplate legalese and documentation
3377 at the head of each file (just as the Lisp source files already did).
3378 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3379 to his page will be replaced with a link to his new CLiki service.
3381 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3383 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3384 * The build process now looks for GNU make under the default name "gmake",
3385 instead of "make" as it used to. If GNU make is not available as "gmake"
3386 on your system, you can change this default behavior by setting the
3387 GNUMAKE environment variable.
3388 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3389 can build without error under CMU CL.
3391 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3393 * There is now a partial SBCL user manual (with some new text and some
3394 text cribbed from the CMU CL manual).
3395 * The beginnings of a profiler have been added (starting with the
3396 CMU CL profiler and simplifying and cleaning up). Eventually the
3397 main interface should be through the TRACE macro, but for now,
3398 it's still accessed through vaguely CMU-CL-style functions and macros
3399 exported from the package SB-PROFILE.
3400 * Some problems left over from porting CMU CL to the new
3401 cross-compilation bootstrap process have been cleaned up:
3402 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3403 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3404 part when getting rid of the compiler *BACKEND* stuff.)
3405 ** The value of *NULL-TYPE* was screwed up, because it was
3406 being initialized before the type system knew the final
3407 definition of the 'NULL type. This screwed up several key
3408 optimizations in the compiler, causing inefficiency in all sorts
3409 of places. (I found it because I wanted to understand why
3410 GET-INTERNAL-RUN-TIME was consing.)
3411 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3412 PROCLAIM FTYPE information. Unfortunately this broke binary
3413 compatibility again, since now the forms output by DEFGENERIC
3414 to refer to functions which didn't exist in 0.6.3.
3415 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3416 can use the new (as of 0.6.3) transform for SOME into MAP into
3418 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3419 values to (INTEGER 0 1000000), so that the time code will no longer
3420 occasionally get blown up by Linux returning 1000000 microseconds
3421 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3422 its output conform to the ANSI spec. (Alas, this makes its output
3423 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3424 * A full call to MAP NIL with a single sequence argument no longer conses.
3425 * fixes to problems pointed out by Martin Atzmueller:
3426 * The manual page no longer talks about multiprocessing as though
3427 it were currently supported.
3428 * The ILISP support patches have been removed from the distribution,
3429 because as of version 5.10.1, ILISP now supports SBCL without us
3430 having to maintain patches.
3431 * added a modified version of Raymond Toy's recent CMU CL patch for
3432 EQUALP comparison of HASH-TABLE
3434 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3436 * The system still can't cross-compile itself with
3437 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3438 entails), but at least it can compile more of itself that way
3439 than it used to be able to, and various buglets which were uncovered
3440 by trying to cross-compile itself that way have now been fixed.
3441 * This release breaks binary compatibility again. This time
3442 at least I've incremented the FASL file format version to 2, so that the
3443 problem can be detected reliably instead of just causing weird errors.
3444 * various new style warnings:
3445 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3446 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3447 ** using the lexical binding of a variable named in the *FOO* style
3448 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3449 as specified by ANSI.
3450 * *RANDOM-STATE* is no longer automatically initialized from
3451 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3452 default behavior of the system is to repeat its behavior every time
3453 it's run. If you'd like to change this behavior, you can always
3454 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3455 old convention there was no comparably easy way to get the system to
3456 repeat its behavior every time it was run.
3457 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3458 FTYPE declarations has been removed, in favor of their ANSI
3459 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3460 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3461 the types of their sequence arguments can be determined at compile time.
3462 This is done through a new open code expansion for MAP which eliminates
3463 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3464 argument types can be determined at compile time.
3465 * The optimizer now transforms COERCE into an identity operation when it
3466 can prove that the coerced object is already of the correct type. (This
3467 can be a win for machine generated code, including the output of other
3468 optimization transforms, such as the MAP transform above.)
3469 * Credit information has been moved from source file headers into CREDITS.
3470 * Source file headers have been made more standard.
3471 * The CASE macro now compiles without complaining even when it has
3474 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3476 * (Note that the way that the PCL macroexpansions were rewritten
3477 to accommodate the change in DEFGENERIC below breaks binary
3478 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3479 not run under sbcl-0.6.2. Once we get out of alpha releases,
3480 i.e. hit release 1.0.0, we'll probably try to maintain binary
3481 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3482 and sbcl-1.4.4. Until then, however, it might be fairly common
3483 for maintenance releases to break binary compatibility.)
3484 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3486 * The reporting of unhandled signals has been changed to print some
3487 explanatory text as well as the report form. (Previously only
3488 the report form was printed.)
3489 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3490 it defines, so that the compiler no longer issues undefined function
3491 warnings for compiled-but-not-yet-loaded generic functions.
3492 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3493 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3494 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3495 section of the ANSI spec are supported.
3496 * The "" nickname for the "KEYWORD" package has been removed.
3497 The reader still handles symbol tokens which begin with a package marker
3498 as keywords, but it doesn't expose its mechanism for doing so in the
3499 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3500 * The system now issues STYLE-WARNINGs for contradictory TYPE
3501 proclamations. (Warnings for contradictory FTYPE proclamations would
3502 be nice too, but those can't be done usefully unless the type system
3503 is made smarter about FUNCTION types.)
3504 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3505 systematized, so that "*target-*.lisp is supposed to exist only on the
3506 target and imply that there's a related file which exists on the
3507 host, and *host-*.lisp is supposed to exist only on the host and imply
3508 that there's a related file which exists on the target. This involves a
3509 lot of renaming. Hopefully the acute confusion caused by the renaming
3510 will be justified by the reduction in chronic confusion..
3511 ** runtime-type.lisp -> early-target-type.lisp
3512 ** target-type.lisp -> late-target-type.lisp
3513 ** early-host-format.lisp -> early-format.lisp
3514 ** late-host-format.lisp -> late-format.lisp
3515 ** host-error.lisp -> misc-error.lisp
3516 ** early-error.lisp -> early-target-error.lisp
3517 ** late-error.lisp -> late-target-error.lisp
3518 ** host-defboot.lisp -> early-defboot.lisp
3519 ** code/misc.lisp -> code/target-misc.lisp
3520 ** code/host-misc.lisp -> code/misc.lisp
3521 ** code/numbers.lisp -> code/target-numbers.lisp
3522 ** code/early-numbers.lisp -> numbers.lisp
3523 ** early-host-type.lisp -> early-type.lisp
3524 ** late-host-type.lisp -> late-type.lisp
3525 ** host-typep.lisp -> typep.lisp
3526 ** load.lisp -> target-load.lisp
3527 ** host-load.lisp -> load.lisp
3528 ** host-disassem.lisp -> disassem.lisp
3529 ** host-insts.lisp -> insts.lisp
3530 ** byte-comp.lisp -> target-byte-comp.lisp
3531 ** host-byte-comp.lisp -> byte-comp.lisp
3532 ** host-signal.lisp -> signal.lisp
3533 ** host-defstruct.lisp -> defstruct.lisp
3534 ** late-target-type.lisp -> deftypes-for-target.lisp
3535 Furthermore, several other previously target-only files foo.lisp (e.g.
3536 hash-table.lisp and random.lisp) have been split into a target-and-host
3537 foo.lisp file and a target-only target-foo.lisp file, with their key type
3538 definitions in the target-and-host part, so that the cross-compiler will
3539 know more about target types.
3540 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3541 gone away. In their place are various *BACKEND-FOO* variables
3542 corresponding to the slots of the old structure.
3543 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3544 into the target SBCL has been fixed.
3545 * The chill.lisp system for loading cold code into a running SBCL
3547 * Support for the CMU CL "scavenger hook" extension has been removed.
3548 (It was undocumented and unused in the CMU CL sources that SBCL was
3549 derived from, and stale in sbcl-0.6.1.)
3550 * Various errors in the cross-compiler type system were detected
3551 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3552 (enabling various consistency checks). Many of them were fixed,
3553 but some hard problems remain, so the compiler is back to
3554 running without *TYPE-SYSTEM-INITIALIZED* for now.
3555 * As part of the cross-compiler type system cleanup, I implemented
3556 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3557 * I have started adding UNCROSS calls throughout the type system
3558 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3559 on cross-compiler input files will be able to go away, and various
3561 * CONSTANTP now returns true for quoted forms (as explicitly required
3564 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3566 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3567 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3568 (MAKE-LIST -1). In the longer term, it ought to become true
3569 that declarations are assertions even at SAFETY 1. For now, it's not
3570 quite true even at SAFETY 3, but it's at least more nearly true..
3571 (Note that this change seems to increases the size of the system by
3572 O(5%) and to decrease the speed of the compiler by 20% or more.)
3573 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3574 for the problem of printing dozens of lines of distracting information
3575 about low-level system machinery as part of the top stack frame
3576 on entry to the debugger when an undefined function was called.
3577 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3578 is not reset to COMMON-LISP-USER.
3579 * Compilation of stuff related to dyncount.lisp has been made conditional
3580 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3581 smaller. The various dyncount-related symbols have been moved into
3582 a new "SB-DYNCOUNT" package.
3583 * tty-inspect.lisp has been renamed to inspect.lisp.
3584 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3585 feature has gone away. (When we eventually port to other flavors of
3586 libc and/or Unix, we'll try to make the differences between flavors
3587 invisible at the user level.)
3588 * Various other *FEATURES* tags, and/or their associated conditionals,
3589 have been removed if obsolescent, or given better documentation, or
3590 sometimes given more-mnemonic names.
3592 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3594 * tidied up "make.sh" script
3595 * tidied up system directory structure
3596 * better "clean.sh" behavior
3597 * added doc/FOR-CMUCL-DEVELOPERS
3598 * many many small tweaks to output format, e.g. removing possibly-confusing
3599 trailing #\. character in DESCRIBE-INSTANCE
3600 * (EQUALP #\A 'A) no longer signals an error.
3601 * new hashing code, including EQUALP hashing
3602 * tidied up Lisp initialization and toplevel
3603 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3604 * command line argument processing
3605 * added POSIX-GETENV function to deal with Unix-ish environment variables
3606 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3607 terminating SBCL on EOF
3608 * non-verbose GC by default
3609 * There is no more "sbcl" shell script; the sbcl file is now the C
3610 runtime executable (just like CMU CL).
3611 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3613 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3614 symbol and package information as Lisp native symbol and package objects
3615 instead of strings naming symbols and strings naming packages. This way,
3616 whenever packages are renamed (as in warm init), debug information is
3617 transformed along with everything else.
3618 * tweaked the optimization policy declarations which control the building
3619 of SBCL itself. Now, among other things, the system no longer saves
3620 source location debugging information. (This helps two problems at once
3621 by reducing SBCL size and by keeping SBCL from trying to look for its
3622 sources -- which may not exist -- when reporting errors.)
3623 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3624 debugging and testing purposes
3625 * cleaned up printing, making the printer call PRINT-OBJECT for
3626 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3627 methods, giving nearly-ANSI behavior
3628 * converted almost all special variables to use *FOO* naming convention
3629 * deleted PARSE-TIME functionality, since it can be done portably
3630 * moved some files out of cold init into warm init
3631 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3633 * regularized formatting of source files
3634 * added an install.sh script
3635 * fixed ridiculous memory usage of cross-compiler by making
3636 compiler/alloc.lisp not try to do pooling unless it can hook
3637 itself into the GC of the cross-compilation host. Now the system
3638 builds nicely on my old laptop.
3639 * added :SB-ALLOC in target-features.lisp-expr
3640 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3641 was not implemented)
3642 * re-did condition handling and note reporting in the compiler. Notes
3643 are no longer handled by signalling conditions. Style warnings
3644 and warnings are handled more correctly and reported in such a way
3645 that it's easy to find one or the other in your output (so that you
3646 can e.g. figure out which of many problems caused COMPILE-FILE to
3648 * changed the severity of several compiler warnings from full WARNING
3649 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3650 compiler note reporting so that it doesn't use the condition system
3651 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3652 COMPILE-FILE command)
3653 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3654 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3655 no longer supported, since their implementation was deeply intertwingled
3656 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3657 * removed bogus "support" for compiler macros named (SETF FOO), and
3658 removed the compiler macro for SETF INFO (but only after making a fool
3659 of myself on the cmucl-imp mailing list by posting a bogus patch for
3660 DEFINE-COMPILER-MACRO..)
3661 * Compiled files containing forms which have side effects on the Lisp
3662 reader (such as DEFPACKAGE forms) are now handled more correctly.
3663 (Compiler queuing of top level lambdas has been suppressed by setting
3664 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3665 * deleted various currently-unused source files, e.g. gengc.lisp. They
3666 may be added back at some point e.g. when porting to other architectures,
3667 but until they are it's distracting to distribute them and to try to
3669 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3670 there were so many of them they're just distractions, and UNCROSS is
3671 known to be able to handle the current sources
3672 * moved PROFILE functionality into TRACE, so that it will be clear
3673 how the wrapping and unwrapping of functions when you profile them
3674 interacts with the wrapping and unwrapping of functions when you
3675 trace them. (Actually, the functionality isn't there yet, but at least
3676 the interface specification is there. Hopefully, the functionality will
3677 arrive with some future maintenance release.)
3678 * removed host-oops.lisp
3679 * changed signature of QUIT function to allow UNIX-CODE argument
3680 * fixed READ-SEQUENCE bug
3681 * tweaked verbose GC output so that it looks more like the progress
3682 output that ANSI specifies for functions like LOAD
3683 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3684 * added <http://sbcl.sourceforge.com> to the banner information printed by