1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
3 * new platform: experimental support for SBCL x86/Darwin, including
4 MacOS X 10.4.5 on Intel.
5 * new feature: Unicode character names are now known to the system
6 (through CHAR-NAME and NAME-CHAR).
7 * minor incompatible change: the contrib modules SB-POSIX and
8 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
9 this change is to make it easier to distribute
10 location-independent binaries.
11 * bug fix: as implied by AMOP, standardized classes no longer have
12 slots named by external symbols of public packages. (reported by
14 * optimization: calling functions via constant symbols -- as in
15 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
16 via the function object as in (FUNCALL #'FOO).
17 * optimization: on x86 and x86-64, the calling convention for the
18 case of dealing with an unknown number of values has been altered
19 to be friendlier to the prediction heuristics implemented,
20 particularly with reference to CALL and RET pairing. (thanks to
22 * enhancement: CONSTANTP is now able to determine constantness of
23 more complex forms, including calls to constant-foldable standardized
24 functions and some special forms beyond QUOTE.
25 * fixed bug: occasional GC crashes on Solaris/x86
26 * optimization: x86-64 supports stack allocation of results of simple
27 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
29 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
30 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
31 be used for bundling the runtime and the core file into one
32 executable binary. This feature is not currently supported on all SBCL
33 platforms. (thanks to James Bielman and NIIMI Satoshi)
34 * new feature: a generational or ephemeral garbage collector is now
35 the default on the PowerPC platform (both Linux and Darwin). The
36 old Cheney (stop and copy) collector is a build-time option.
37 (thanks to Cyrus Harmon, after Raymond Toy)
38 * minor incompatible change: the method by which SBCL finds its
39 contributed modules has changed; it no longer relies on symbolic
40 links from an $SBCL_HOME/systems directory, but searches directly
41 in the subdirectories of $SBCL_HOME.
42 * enhancement: the dynamic heap size on the Linux/PPC platform is
43 markedly larger, even using the older Cheney garbage collector.
44 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
45 certain complicated string types. (reported by Gary King)
46 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
47 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
48 * fixed bug: DOCUMENTATION on structures no longer signals an
49 error if no structure type of the right name exists. (reported by
51 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
52 applicable non-standard primary or :AROUND method on
53 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
54 no longer cause unbound variable errors. (reported by Kevin Reid)
55 * fixed bug: metacircle resolution in cases where methods have slots
56 added before the slots from STANDARD-METHOD. (reported by Jean
58 * fixed bug: the Power PC assembly code for calling into lisp
59 sometimes computed its return address wrongly (depending on the
60 15th bit of the address, fixed at link time). This used to
61 manifest itself in a segmentation violation while building PCL.
62 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
63 many others over the years)
64 * fixed bug: ROOM no longer reports silly numbers for stack usage if
65 the stack pointer (treated as a signed integer) is negative.
66 (thanks to Peter van Eynde)
68 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
69 * new platform: experimental support for the Windows operating
70 system has been added. (thanks to Alastair Bridgewater)
71 * fixed several bugs in and robustified the PPC FFI (including
72 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
73 * bug fix: allow non-simple string symbol names (reported by Paul
75 * bug fix: interrupt handling on NetBSD (thanks to Richard M
77 * bug fix: saving a core corrupted callbacks on x86/x86-64
78 * bug fix: closed a loophole in metacircularity detection and
79 grounding in the PCL implementation of CLOS.
80 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
81 (reported by tomppa on #lisp)
82 * bug fix: clear the direction flag on Lisp -> C transitions, as
83 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
84 (reported by Andrej Grozin and Hendrik Maryns)
85 * optimization: major improvements to GC efficiency on GENCGC platforms
86 * optimization: faster implementation of EQUAL
87 * optimization: emit more efficient opcodes for some common
88 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
89 * optimization: several other minor code-generation improvements on x86-64
90 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
92 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
93 * minor incompatible change: (SETF CLASS-NAME) and (SETF
94 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
95 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
96 by ANSI as a generic function, and for consistency (SETF
97 GENERIC-FUNCTION-NAME) is treated likewise.
98 * fixed bug #233.b: make constraint propagation notice when a variable
99 value is changed after it is referenced but before it is used
100 * fixed bug #296: no more arbitrary behaviour from filenames with
101 odd characters as --load arguments. (reported by Adam Warner)
102 * fixed bug #390: :CHARACTER-SET pathname components now work as
103 expected. (reported by Tim Daly Jr)
104 * fixed bug #391: complicated :TYPE intersections in slot
105 definitions no longer cause an error in PCL internals.
106 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
107 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
108 returning the number of octets which would be written to the
109 file-stream. (thanks to Robert J. Macomber)
110 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
111 arguments to the internal stream functions. (thanks to David
113 * bug fix: stack exhaustion is now handled on the x86/SunOS
114 platform. (thanks to Daisuke Homma)
115 * bug fix: more accurate ROOM results on GENCGC platforms
116 * optimization: improved type inference for arithmetic-for
117 index variables in LOOP
118 * optimization: faster floating-point SQRT on x86-64
119 * fixed some bugs revealed by Paul Dietz' test suite:
120 ** DOCUMENTATION returns NIL instead of "" for method combinations
121 that don't have a docstring
123 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
124 * minor incompatible change: (SETF CLASS-NAME) and (SETF
125 GENERIC-FUNCTION-NAME) are no longer generic functions, and
126 therefore are not customizeable by user code (as seems to be at
127 least permitted and maybe required by AMOP). As a consolation,
128 however, the SBCL implementation of these functions now calls
129 REINITIALIZE-INSTANCE as specified by AMOP.
130 * bug fix: slot-definition documentation is propagated as per ANSI
131 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
132 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
133 argument, as do eslotd initialization methods. (from Pascal
134 Costanza's "Closer" project)
135 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
136 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
138 * bug fix: it is now possible to have more than one subclass of
139 STANDARD-GENERIC-FUNCTION without causing stack overflow.
140 (reported by Bruno Haible, Pascal Costanza and others)
141 * bug fix: the dependent update protocol now works for generic
142 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
144 * bug fix: condition-class instances corresponding to
145 DEFINE-CONDITION forms are now created eagerly. (reported by
146 Kalle Olavi Niemitalo on comp.lang.lisp)
147 * bug fix: floating point printing is more accurate in some
148 circumstances. (thanks to Simon Alexander)
149 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
150 merged with *DEFAULT-PATHNAME-DEFAULTS*.
151 * bug fix: callbacks on OS X now preserve stack-alignment.
152 (thanks to Cyrus Harmon)
153 * enhancement: the x86-64 disassembler is much better at
154 disassembling SSE instructions. (thanks to Lutz Euler)
155 * enhancement: CHECK-TYPE on a variable now allows type-inference
156 (thanks to Vincent Arkesteijn)
157 * optimization: improved performance of EUC-JP external format.
158 (thanks to NIIMI Satoshi)
159 * optimization: performance improvements to IO on file streams of
160 :ELEMENT-TYPE CHARACTER
161 * optimization: much faster memory allocation on x86-64
162 * optimization: faster garbage collections (latency and throughput)
164 * optimization: faster bignum subtraction and division on x86 and x86-64
165 (thanks to Lutz Euler)
166 * optimization: more accurate type-inference for loops that use a
167 floating point index variable or a negative step.
169 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
170 * bug fix: add a workaround to SBCL looping infinitely at startup on
171 Linux kernels with apparently buggy implementations of personality().
172 (thanks to Svein Ove Aas)
173 * bug fix: Unicode symbols are correctly printed in LDB backtraces
174 (thanks to David Lichteblau)
175 * bug fix: local bindings shadow global symbol macros for
176 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
177 * optimization: non-open coded uses of numeric comparison operators
178 (e.g. >) no longer cons when called with more than one parameter
179 on platforms supporting dynamic-extent allocation.
180 * enhancement: saving cores with foreign code loaded is now supported
181 on MIPS/Linux in addition to the previously supported platforms.
182 * bug fix: FIND-RESTART now tests for activity, not applicability when given
183 a restart object as identifier. (reported by Helmut Eller for CMUCL)
184 * bug fix: division by zero in sb-sprof when no samples were collected
185 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
186 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
187 (thanks to Alexander Kjeldaas)
188 * new feature: ignoring the return values of destructive functions
189 like NREVERSE signals a compile-time style-warning.
190 (thanks to Kevin Reid)
191 * new feature: the alignment of alien structure fields can be
192 explicitly specified. (thanks to Cyrus Harmon)
193 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
194 and dump core on SIGQUIT
196 ** incompatible change: threads do not inherit values of specials
197 from their parents (see manual)
198 ** bug fix: threads stacks belonging to dead threads are freed by the
199 next exiting thread, no need to gc to collect thread stacks anymore
200 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
201 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
202 ** bug fix: dynamic variable and thread start related gc lossage
203 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
205 * fixed some bugs revealed by Paul Dietz' test suite:
206 ** SUBTYPEP is slightly more accurate on heinously complicated
207 CONS types where some of the members have uncertain (in the
208 NIL, NIL sense) type relationships to each other.
210 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
211 no longer purified unless :PURIFY T is explicitly specified.
212 ** Non-purified cores are significantly smaller than before
214 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
215 * new feature: timers based on Zach Beane's excellent timer package
216 * added support for the following external formats: koi8-u,
217 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
218 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
219 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
220 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
221 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
222 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
223 * incompatible change: a threaded SBCL will no longer revert to
224 non-threaded mode on non-NPTL systems, but refuse to start entirely.
225 * bug fix: interrupts are disabled until startup is complete; no
226 more sigsegvs when receiving a signal to soon
227 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
229 * bug fix: add a workaround for the memory randomization features in
230 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
231 workaround will only be in effect on systems with the proc filesystem
233 * bug fix: printing objects of type HASH-TABLE signals a
234 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
236 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
237 have been running for over 50 days. (reported by Gilbert Baumann)
238 * bug fix: the logic for getting names of functions gets less
239 confused when confronded with alternate-metaclass
240 funcallable-instances. (reported by Cyrus Harmon)
241 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
242 consistent, even on internal alternate-metaclass objects.
243 * bug fix: flush closure information collected by physical
244 environment analysis prepass before the main pass. (bug reported
246 * bug fix: compiler pack phase does not modify a hashtable, which is
247 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
248 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
249 may be expanded. (reported by Alexander <malishev> on c.l.l)
250 * bug fix: no more style-warnings when compiling code using LAST
251 under high-SPEED low-SPACE settings. (reported by David Wallin
253 * fixed some incompatibilities between SBCL's MOP and the MOP
255 ** the METAOBJECT class is now implemented;
256 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
257 STANDARD-OBJECT, as required;
258 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
259 now compatible (as required by VALIDATE-SUPERCLASS); there
260 remains a consistency requirement over the presence or absence
261 of the FUNCTION class in the superclass list of finalized
262 classes; see the manual for more details;
263 ** the :AROUND method for COMPUTE-SLOTS on
264 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
265 requested slot ordering.
267 ** bug fix: parent thread now can be gc'ed even with a live
269 ** bug fix: binding a special with PROGV to no value is not
270 the same as that symbol not having been bound (thanks to
272 ** bug fix: binding specials is thread safe (thanks to
274 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
275 and SB-SPROF do not die with 'no handler for signal XX in
276 interrupt_handle_now(..)' anymore
277 ** bug fix: WITH-TIMEOUT works with multiple running threads
278 * fixed some bugs revealed by Paul Dietz' test suite:
279 ** ENSURE-GENERIC-FUNCTION should take a method class object for
280 the :method-class keyword argument.
282 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
283 * new port: the Solaris operating system on x86 processors is now
284 mostly supported, though some rough edges in the environment
285 remain. (thanks to Daisuke Homma)
286 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
287 space for its heap. (thanks to Thiemo Seufer)
288 * optimization: SBCL on MIPS platforms now supports dynamic-extent
289 closures. (thanks to Thiemo Seufer)
290 * minor incompatible change: eof selects abort in the debugger.
291 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
292 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
294 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
295 deprecated, and will go away in a future revision of SBCL.
296 * minor incompatible change: GC-ON and GC-OFF are no longer
297 implemented with a counter, it does not matter how many times gc
298 is switched on or off
299 * bug fix: discriminating functions for generic function classes
300 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
301 make invalid assumptions about method precedence order. (reported
303 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
304 platforms. (thanks to Thiemo Seufer)
305 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
306 * bug fix: correct pathname printing with printer escaping is on.
307 (thanks to Kevin Reid)
308 * bug fix: complex VOP definitions in "user-space" no longer trigger
309 package locks. (reported by Zach Beane)
310 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
311 for funcallable-instance functions, and is no different from
313 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
314 characters in the current readtable. (reported by Nicholas Neuss)
315 * bug fix: SUBTYPEP on various CONS types returns more a more
316 accurate acknowledgment of its certainty.
317 * optimizations: REMOVE-DUPLICATES now runs in linear time on
318 lists in some cases. This partially fixes bug 384.
319 * flush all standard streams before prompting in the REPL and the
321 * bug fix: signal handling and triggering gc do not conflict
322 directly anymore, in particular a high frequency sb-sprof does
323 not prevent gc from running
324 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
325 approximation for timezone and DST information between the
326 universal time epoch and the smallest negative 32-bit time_t.
327 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
328 year 1899 so long as the encoded time is non-negative.
329 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
330 the correct answer for non-constant rotate amounts when performing
331 an inline 32-bit rotation.
333 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
334 there is only one thread in the session
335 ** bug fix: memory leak for streams created in one thread and
336 written to in another
337 ** bug fix: lockup when compiled with gcc4
338 ** bug fix: race that allows the gc to be triggered when gc is
340 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
341 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
343 ** bug fix: finished threads are gc'ed properly
344 * fixed some bugs revealed by Paul Dietz' test suite:
345 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
346 has arguments with a different set of applicable methods from
347 the orignal arguments.
348 ** The type error thrown by MAP now has a correct expected-type
350 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
351 name a compiled function.
352 ** Three MISC tests where a large bignum was improperly coerced to
353 a float that couldn't represent that bignum during type
354 derivation were fixed.
355 ** SUBTYPEP can now handle types involving the negation of a
356 list-form FUNCTION type.
357 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
358 as (COMPLEX (AND RATIO (NOT FIXNUM))).
359 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
361 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
362 * New feature: Experimental support for bivalent streams: streams
363 opened with :element-type :default now allow character and binary
364 (unsigned-byte 8) I/O
365 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
366 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
368 * bug fix: no more highly sporadic "couldn't check whether ~S is
369 readable" when reading a stream and an interrupt hits in the middle
370 of a select system call
371 * compiler better recognizes complex arrays (reported by Eduardo
373 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
374 for variables on SBCL built with linkage-tables. (reported by Luis
376 * various error reporting improvements.
377 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
378 (Thanks to Thiemo Seufer)
379 * bug fix: nested reader invokations work correctly
380 * bug fix: it is possible to have simultaneous references to foreign
381 code and foreign data with the same name.
383 ** added x86-64 support
384 ** incompatible change: the threading api now works with thread
385 objects instead of thread ids
386 ** bug fix: threads are protected from signals and interruption when
387 starting up or going down
388 ** bug fix: a race where an exiting thread could lose its stack to gc
389 ** bug fix: don't halt on infinite error in threads if possible
390 ** fixed numerous gc deadlocks introduced in the pthread merge
391 ** bug fix: fixed thread safety issues in read and print
392 ** bug fix: debugger doesn't hang on session lock if interrupted at
393 an inappropriate moment
394 ** bug fix: run-program is now thread safe(r)
395 ** bug fix: inner with-recursive-lock no longer releases the mutex
396 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
397 * fixed some bugs revealed by Paul Dietz' test suite:
398 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
399 EXPECTED-TYPE slots filled.
400 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
402 ** MAKE-CONDITION accepts classes as type-designators.
403 ** COMPILE may never return NIL.
404 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
405 range before calling Unix time functions
407 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
408 * numerous signal handling fixes to increase stability
409 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
410 * minor incompatible change: we now correctly canonize default
411 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
412 as per the MOP, rather than the historical (INITARG INITFUNCTION
413 INITFORM). (reported by Bruno Haible)
414 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
415 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
416 for more information.
417 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
418 pathname is a directory pathname.
419 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
420 * dynamic space size on PPC has been increased to 768Mb. (thanks to
422 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
423 :METACLASS argument in addition to a class name. (reported by
424 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
425 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
426 (reported by Bruno Haible; patch from Gerd Moellmann)
427 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
429 * bug fix: more cleanups to the floating point exception handling on
430 x86-64 (thanks to James Knight)
431 * bug fix: the compiler does not try to constant fold calls of
432 COERCE and BIT-* functions when they return freshly constructed
433 objects. (reported by Nikodemus Siivola and Paul Dietz)
434 * optimization: improved the MIPS versions of generic (in the
435 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
436 * optimization: direct conversion of (unsigned-byte 32) to floats on
437 the PowerPC platform.
438 * optimization: structure instances with raw slots now use less
439 memory, and probably show better memory locality. (thanks to
441 * optimization: DYNAMIC-EXTENT declarations for lists and closures
442 are treated as requests for stack allocation on the x86-64,
443 Alpha-32, PPC and SPARC platforms.
444 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
445 the wrong file descriptor; implementation of SOCKET-OPEN-P.
446 (thanks to Tony Martinez)
448 ** gcing a dead thread can no longer lead to lockups
449 ** threads block signals until they are set up properly
450 ** errno is no longer shared by threads
451 ** interrupt-thread restores the eflags register on x86
452 ** fixed some lockups due to gc/thread interaction
453 * fixed some bugs revealed by Paul Dietz' test suite:
454 ** invalid dotted lists no longer raise a read error when
456 ** signal an error if a symbol that names a declaration is used
457 as the name of a type, or vice versa
458 ** allow using the (declare (typespec var*)) abbreviation for
459 (declare (type typespec var*)) with all type specifiers
460 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
461 ** always return NIL from PPRINT-POP when OBJECT is NIL
462 ** don't signal errors when pretty-printing malformed LABELS,
463 FLET or MACROLET forms
464 ** declarations in a DOLIST body are also in scope for the
466 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
468 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
471 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
472 * fixed cross-compiler leakages that prevented building a 32-bit
473 target with a 64-bit host compiler.
474 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
475 opened with :IF-EXISTS :OVERWRITE.
476 * fixed bug 281: error for an invalid qualifier in a short-form method
477 combination method is not signalled until the faulty method is called.
478 * bug fix: iteration variable type inferrer failed to deal with open
479 intervals. (reported by Alan Shields)
480 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
481 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
482 or more runtime options were provided to the sbcl binary.
483 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
484 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
485 * optimization: there's now a fast-path for fixnum arguments in the
486 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
487 * optimization: the code generated on x86-64 is more compact thanks
488 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
489 * bug fix: floating-point exception handling now partly works on
490 x86-64. (Thanks to James Knight)
491 * improvement to the MIPS backend from Thiemo Seufer: C-style
492 64-bit long long arguments and return values to alien functions
494 * fixed some bugs revealed by Paul Dietz' test suite:
495 ** the type-error signalled from WARN has a filled-in DATUM slot.
496 ** the type-error required when a stream is not associated with
497 a file has the stream as its datum.
498 ** type-errors on single-floats on x86-64 no longer have
499 :INVALID-OBJECT as the datum
500 ** the type-errors signalled for invalid function names now have
501 a correct expected type
502 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
503 for structures defined with a :TYPE. Documentation strings for
504 typed structures are no longer immediately discarded
505 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
506 broadcast streams and synonym streams. FILE-LENGTH now also works
507 on broadcast streams.
509 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
510 * incompatible change: the --noprogrammer option, deprecated since
511 version 0.7.5, has been removed. Please use the equivalent
512 --disable-debugger option instead.
513 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
515 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
516 inoperational for a while now) has been completely removed.
517 * null lexical environments are now printed as #<NULL-LEXENV>,
518 significantly reducing the amount of clutter in typical backtraces.
519 * documentation on weak pointers, finalization, and after GC hooks
520 has been added to the manual.
521 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
522 as other specialized array types, is much faster. SUBSEQ and
523 COPY-SEQ on such arrays have also been sped up.
524 * optimization: EQL is now more efficient when at least other argument
525 is known to be of type (OR FIXNUM (NOT NUMBER)).
526 * fixed bug: the runtime is now less vulnerable to changes in the
527 size of the SBCL object on OS X, and virtual memory is reserved for
528 all spaces that need to be at a fixed address.
529 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
530 * fixed bug: finalizers and after GC hooks that cause consing are now
532 * fixed bug: compiler error messages and summaries are now printed to
533 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
534 * fixed inference of the upper bound of an iteration variable.
535 (reported by Rajat Datta).
536 * fixed bug 211e: calling local functions with duplicated constant
537 keyword argument no longer causes a bogus style warning about an
539 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
540 ftype declaration to be disregarded. (reported by Dave Roberts)
541 * fixed bug 373: caused by erronous compilation of references to alien
542 variables in the runtime on ppc/darwin.
543 * fixed bug 376: CONJUGATE type deriver.
544 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
545 (reported by Baughn on #lisp)
546 * a cross-compiler bug on non-x86 platforms has been identified and
547 fixed. (thanks to Bruno Haible)
548 * improvements to the MIPS runtime code for increased stability.
549 (thanks to Thiemo Seufer)
550 * increased the maximimum compact environment size to allow
551 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
552 * partial workaround for bug 135: don't name the function we're
553 calling for hairy cases of EVAL, so as not to accumulate one environment
554 entry per call to EVAL. (reported by Kevin Reid)
555 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
556 * optimization: type testing for non-vector arrays should be faster.
557 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
559 * bug fix: setting 31st element of a bit vector to zero did not work
561 * bug fix: redefining a class definition which failed due to a
562 previous accessor / function clash now works (but see BUGS entry
563 #380 for more problems in this area). (thanks to Zach Beane)
564 * the long form of DEFINE-METHOD-COMBINATION disables method group
565 checking when given a single method group with pattern *. (thanks
567 * on x86 compiler supports stack allocation of results of simple
568 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
569 * support for building and running on Mac OS X 10.4 "Tiger" has been added
570 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
571 * fixed some bugs related to Unicode integration:
572 ** the restarts for recovering from input and output encoding
573 errors only appear when there is in fact such an error to
575 * fixed some bugs revealed by Paul Dietz' test suite:
576 ** MISC.549 and similar: late transformation of unsafe type
577 assertions into derived types caused unexpected code
579 ** SCALE-FLOAT type deriver is less wrong.
580 ** type derivers for EXP, LOG and similar functions compute result
581 types for complex arguments better.
582 ** (MISC.563) CONJUGATE type deriver works for very restricted
584 ** out-of-line type testers for character strings are available.
585 ** EQUAL compiler transform understands specialness of objects
587 ** accessing double-floats stored on the stack now works on x86-64.
588 ** debugger internals could sometimes create invalid lispobjs,
589 resulting in GC crashes.
590 ** MISC.548: type check weakening can convert required type into
592 ** initialization forms for bindings are not in scope of free special
595 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
596 * incompatible change: thread support for non-NPTL systems has
597 been removed - locking is buggy and unreliable. A threaded
598 SBCL build will now warn at startup and refuse to create
599 new threads, unless futex support is detected in the kernel
600 * incompatible change: the top level REPL now has only an ABORT
601 restart associated with it, not TOPLEVEL and ABORT as it used to.
602 TOP and TOPLEVEL are now available as debugger commands for
603 returning to the top level.
604 * incompatible change: forms evaluated in the REPL now use the
605 global optimization policy.
606 * incompatible change: user- and system-initialization files are
607 no longer processed with LOAD, but by READ and EVAL; hence the
608 global optimization policy, startup package, readtable, etc,
610 * The .fasl file format number has been incremented because of
611 various incompatible changes.
612 * internal entry point details and argument counts no longer appear
613 in backtraces unless explicitly requested by setting
614 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
615 * built-in and standard functions no longer have names like "top
616 level local call to FOO".
617 * fixed bug 32: functions defined in non-null lexical environments
618 now have more legible printed representation
619 * fixed bug 33: functions defined in non-null lexical environemnts
620 are now more amenable to inspection by INSPECT.
621 * workaround for bug 354: XEPs no longer appear in backtraces unless
622 explicitly requested.
623 * fixed bug: receiving the signal which results from attempting to
624 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
625 SIGBUS on other *BSDs) on architectures where the C stack is also
626 the Lisp stack (x86 and x86-64) from foreign code no longer leads
627 to debugger confusion or wild instability. (reported by Cheuksan
629 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
630 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
631 (reported by Lutz Euler)
632 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
633 compiled code. (reported by Paul Dietz)
634 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
635 specialized parameter to a method to something that is not TYPEP
636 the specializer is now possible.
637 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
638 face of package deletion.
639 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
640 pathname. (reported by David Tolpin on c.l.l)
641 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
642 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
643 STRING-STREAM and FILE-STREAM, respectively.
644 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
645 than previously; in addition, its test suite is now run on build.
646 (thanks to Luke Gorrie)
647 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
648 * optimization: added a immediate representation for single-floats
650 * fixed some bugs revealed by Paul Dietz' test suite:
651 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
652 ** attempting to create a package with a colliding nickname causes
653 correctable errors to be signalled.
654 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
655 ** several x86-64 backend bugs related to sign-extension of immediate
658 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
659 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
660 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
662 * fixed loading of multiply forward-referenced layouts.
663 (thanks to Cheuksan Wang)
664 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
665 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
666 (see COMPILE-FILE documentation for details of :PRINT option)
667 * fixed bugs 19 and 317: fixed-format floating point printing is
668 more accurate. This also fixes a bug reported by Adam Warner
669 related to the ~@F format directive.
670 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
672 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
673 dispatch table if the from-char is a dispatch macro character.
674 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
675 by the wordsize no longer ignore the last word. (reported by Lutz
677 * fixed bug in type checking of dynamic-extent variables. (reported
679 * optimization: sequence traversal functions use their freedom to
680 coerce function designators to functions.
681 * optimization: code with many calls to CLOS methods specialized on
682 CLOS classes has had redundant type checks removed.
683 * optimization: added declarations to speed up operations that access
684 the internal character database (for example STRING-UPCASE)
685 * optimization: comparison operations between floats and sufficiently small
686 fixnums no longer create extra rationals
687 * fixed some bugs related to Unicode integration:
688 ** portions of multibyte characters at the end of buffers for
689 character-based file input are correctly transferred to the
690 start of the buffer at the next read.
691 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
692 passing it through to OPEN.
693 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
694 argument given to internal calls to OPEN.
695 ** the built-in comment readers (introduced by character sequences
696 ";" and "#|") are more forgiving to encoding errors; they will
697 STYLE-WARN and then attempt to resync the stream at a character
698 boundary. (thanks to Teemu Kalvas)
699 * fixed some bugs in the x86-64 port:
700 ** Negative short int return values from c-calls are sign-extended
702 ** The stack is aligned to 16-bytes for c-calls, as required by
703 the ABI. (thanks to Cheuksan Wang)
704 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
705 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
706 for enums. (thanks to Vincent Arkesteijn)
707 ** Multiple small optimizations and bugfixes for floating point
709 * fixed some bugs revealed by Paul Dietz' test suite:
710 ** Space, Tab, Linefeed, Return and Page have the invalid
711 secondary constituent character trait.
712 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
714 ** WITH-INPUT-FROM-STRING should only update the index place on
716 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
718 ** Bit-array operations (BIT-AND and similar) worked incorrectly
719 with one-dimensional arrays with fill pointers.
720 ** TYPE-OF failed on a complex with an integer realpart and a
722 ** compiler failure during type inference for the code of form
723 (IF (EQL X (THE ...)) ...) (MISC.535).
725 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
726 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
727 platform. The port supports 61-bit fixnums, large memory spaces
728 and reloading shared object files.
729 * enhancement: saving cores with foreign code loaded is now
730 supported on ppc/Darwin in addition to the previously supported
732 * enhancement: the statistical profiler now walks deeper into the
733 call stack for more meaningful call-graphs and accrued time
734 reports (x86/x86-64 only). It also now reports time spent in
736 * enhancement: it is now possible to trace most individual methods
737 of a generic function in addition to tracing the generic function
739 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
741 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
742 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
743 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
744 produces an error. (thanks to Vincent Arkesteijn)
745 * bug fix: NAMESTRING on pathnames with :WILD components in their
746 directories works correctly. (thanks to Artem V. Andreev)
747 * fixed bug 125: compiler preserves identity of closures. (reported
749 * bug fixed: functions with &REST arguments sometimes failed with
750 "Undefined function" when compiled with (DEBUG 3). (reported by
752 * bug fix: overflow during compiling of setting element of a bit
753 vector with constant index and value. (reported by Timmy Douglas)
754 * build fix: fixed the dependence on *LOAD-PATHNAME* and
755 *COMPILE-FILE-PATHNAME* being absolute pathnames.
756 * on x86 compiler partially supports stack allocation of dynamic-extent
758 * GO and RETURN-FROM do not check the extent of their exit points
759 when compiled with SAFETY 0.
760 * fixed some bugs related to Unicode integration:
761 ** encoding and decoding errors are now much more robustly
762 handled; it should now be possible to recover even from invalid
763 input or output to the terminal. (thanks to Teemu Kalvas)
764 ** provided a first cut at implementing STRING-TO-OCTETS and
765 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
766 ** altered the SB-MD5 contributed module slightly, changing the
767 interface just enough for it to be supportable for builds where
768 lisp characters are not eight bits.
769 * fixed some bugs revealed by Paul Dietz' test suite:
770 ** the FORMATTER-generated functions for ~V[ conditionals require
771 the correct number of arguments.
772 ** READ-FROM-STRING returns the mandated second value when applied
773 to displaced strings.
774 ** the #\Rubout and #\Backspace characters are treated as invalid
775 constituent characters by the tokenizer.
777 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
778 * new feature: reloading changed shared object files with
779 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
780 * new feature: references to foreign variables and functions
781 can now be compiled and loaded before the corresponding shared
782 object file is loaded, as long as the foreign definitions are
783 available at runtime.
784 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
785 supported. (thanks to Dan Debertin)
786 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
787 just keywords. (thanks to Vincent Arkesteijn)
788 * fixed bug #331: structure-class instances corresponding to
789 DEFSTRUCT forms are now created eagerly.
790 * fixed bug #345: backtraces from calls to undefined functions work
791 on x86 as well. Related bug #61 is now also partially fixed on x86
792 (backtraces from throws to unknown catch tags.)
793 * bug fix: lambda-list parsing is now stricter vrt. order and number
794 of lambda-list keywords.
795 * bug fix: as specified by AMOP, an error is signalled if a
796 class-option appears multiple times in a DEFCLASS form. (reported
798 * bug fix: RANDOM can be compiled when the compiler derives the type
799 of its numeric argument as a disjoint set of small integers.
800 (reported by Paul Dietz)
801 * bug fix: starting a core saved with shared objects loaded when
802 those objects are not available no longer causes threaded SBCL to
803 hang. (reported by Sean Ross)
804 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
806 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
807 argument, which must always be NIL. (reported by Kalle Niemitalo)
808 * bug fix: printing 1.0d+23 no longer results in an error.
809 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
810 * bug fix: structure slot setters preserve evaluation order. (thanks
812 * fixed some bugs related to Unicode integration:
813 ** RUN-PROGRAM can allow its child to take input from a Lisp
814 stream. (reported by Stefan Scholl)
815 ** ASDF-INSTALL successfully downloads .tar.gz files in all
816 locales. (reported by Ken Causey)
817 * fixed some bugs revealed by Paul Dietz' test suite:
818 ** INCF, DECF and REMF evaluate their place form as specified in
820 ** FORMATTER expands ~{ ~} iteration directives with V or #
821 parameters correctly.
822 ** FORMATTER deals with the ~@[ ~] conditional directive where the
823 consequent uses no arguments correctly.
824 ** the system has a partial understanding of the (COMPLEX RATIO)
826 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
829 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
830 * new feature: a build-time option (controlled by the :SB-UNICODE
831 keyword feature, enabled by default) for building the system with
832 support for the entire 21-bit character space defined by the
834 * new feature: the system now has rudimentary external-format
835 support; the primary user-visible change at this time is that
836 characters with the high bit set (such as lower-case-e-acute) will
837 print correctly to a terminal in a UTF-8 environment.
838 * minor incompatible change: BASE-CHAR no longer names a class;
839 however, CHARACTER continues to do so, as required by ANSI.
840 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
841 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
842 has been moved to the SB-EXT package (temporarily re-exported from
844 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
845 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
847 * fixed bug #351: better error-handlind and reporting for malformed
849 * fixed bug #350: bignum-printing is now more memory-efficient,
850 allowing printing of very large bignums, eg. (expt 2 10000000).
851 (reported by Bruno Haible)
852 * fixed bug #302: better primitive-type selection for intersection
854 * fixed bug #308: non-graphic characters now all have names, as
855 required. (reported by Bruno Haible)
856 * bug fix: redefining a class with different superclasses now correctly
857 removes it from the direct-subclasses of its previous superclasses.
858 (reported by David Morse)
859 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
860 new value now works. (reported by Bruno Haible)
861 * bug fix: correct canonicalization of multiple non-standard slot
862 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
863 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
864 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
865 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
866 now exists, an signals an error.
867 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
868 during macroexpansion for non-iteration variables as well. (reported
869 by Bruno Haible for CMUCL)
870 * bug fix: Cyclic structures and unprintable objects in compiler
871 messages no longer cause errors. (reported by Bruno Haible)
872 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
873 and READ-FROM-STRING all now return a primary value of NIL if
874 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
875 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
876 (reported by Bruno Haible for CMUCL)
877 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
878 do not designate a condition. (reported by Bruno Haible for
880 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
881 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
882 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
883 permitting resolution in favour of any of the conflicting symbols.
884 (reported by Bruno Haible for CMUCL)
885 * bug fix: EQUAL compiler optimizations is less aggressive on
886 strings which can potentially compare true despite having distinct
887 specialized array element types.
888 * bug fix: unit enumerations can be defined without dividing by
889 zero. (thanks to Vincent Arkesteijn)
890 * FORMAT compile-time argument count checking has been enhanced.
891 (report from Bruno Haible for CMUCL)
892 * a partial workaround for the bug 262: the compiler does not try to
893 inline-expand a local function doing RETURN-FROM from a deleted
894 BLOCK. (thanks to Peter Denno for the bug report and to David
895 Wragg for the simple test case)
896 * fixed some bugs revealed by Paul Dietz' test suite:
897 ** RENAME-PACKAGE allows all package designators as new package
899 ** constraint propagation and lambda variable substitution are
900 more cautious in dealing with partially deleted code.
901 ** compiler sometimes lost reoptimization passes.
902 ** CERROR, given a condition as condition designator, passes its
903 remaining arguments to the continue format control without
905 ** Case-altering FORMAT directives work correctly on non-ASCII
907 ** The REAL type specifier handles bounds outside the floating
908 point ranges without signalling FLOATING-POINT-OVERFLOW.
909 ** Functions with IR1-transformations can create intercomponent
910 references to global functions.
911 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
913 ** FORMAT ~R treats a nil value for its first parameter correctly.
915 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
916 * enhancement: saving cores with foreign code loaded is now
917 supported on x86/NetBSD and sparc/Linux in addition to the previously
919 * bug fix: on some platforms repeated installations caused multiple
920 copies of HTML documentation to be installed -- should not happen
921 any more. (reported by Stefan Scholl)
922 * bug fix: parsing self-recursive alien record types multiple times
923 no longer causes infinite recursion. (reported by Thomas F. Burdick,
924 original patch by Helmut Eller for CMUCL)
925 * bug fix: stack-exhaustion detection works now on NetBSD as well.
926 (thanks to Richard Kreuter)
927 * bug fix: defining classes whose accessors are methods on existing
928 generic functions in other (locked) packages no longer signals
929 bogus package lock violations. (reported by François-René Rideau)
930 * bug fix: special variables as DEFMETHOD parameters no longer have
931 associated bogus type declarations. (reported by David Wragg and
933 * bug fix: read-write consistency on streams of element-type
934 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
935 * bug fix: redefiniton of the only method of a generic function with
936 no DEFGENERIC no longer emits a full WARNING. In addition,
937 redefinition of generic functions with no DEFGENERIC to an
938 incompatible lambda list now signals an error. (thanks to Zach
940 * bug fix: DEFGENERIC now works even when there's a function of the
941 same name in an enclosing lexical environment. (thanks to Zach
943 * fixed compiler failure, caused by instrumenting code during
944 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
945 * optimization: added loop analysis and improved register allocation
946 * fixed some bugs revealed by Paul Dietz' test suite:
947 ** POSITION on displaced vectors with non-zero displacement
948 returns the right answer.
949 ** (SIMPLE-STRING) is a valid type specifier for sequence
951 ** *PRINT-LEVEL* handling for slotless structures is pedantically
953 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
954 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
956 ** The justification version of the FORMAT ~< directive treats
957 non-zero minpad parameter correctly.
959 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
960 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
961 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
962 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
963 the supported interface.
964 * new feature: Single-stepping of code compiled with DEBUG 2 or
965 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
966 * new feature: saving cores with foreign code loaded is now
967 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
968 Timothy Moore's work for CMUCL)
969 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
970 parameters to * instead of NIL if no initform is supplied.
971 (reported by Johan Bockgaard)
972 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
973 correctly when FUNCALL appears as the car of the form. Note:
974 despite this FUNCALL forms are not currently subject to
975 compiler-macro expansion. (port of Raymond Toy's fix for the
976 same from CMUCL, reported by Johan Bockgaard)
977 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
978 (thanks for Teemu Kalvas)
979 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
980 case instead of terminating the iteration (thanks for Julian
981 Squires, Sean Champ and Raymond Toy)
982 * bug fix: incorrect expansion of defgeneric that caused a style
983 warning. (thanks for Zach Beane)
984 * bug fix: slot accessor effective method computation works properly
985 for classes with multiple non-standard applicable methods on
986 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
987 * on x86 compiler supports stack allocation of results of LIST and
988 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
989 CMUCL implementation by Gerd Moellmann)
990 * fixed some bugs revealed by Paul Dietz' test suite:
991 ** FORMAT strings with both the ~<~:;~> form of the justification
992 directive and pretty-printing directives cause an error.
994 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
995 * incompatible change: the internal functions
996 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
997 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
998 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
999 instead of the old functions.
1000 * new feature: on platforms where "dladdr" is available foreign
1001 function names now appear in backtraces. (based on Helmut Eller's
1003 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1004 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1006 * documentation: ASDF manual is now distributed with SBCL.
1007 * bug fix: SBCL can now load its contributed modules with REQUIRE
1008 even if the system-provided entries have been removed from
1009 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1011 * bug fix: backtraces involving undefined functions or assembly
1012 routines are more informative. (thanks to Brian Downing)
1013 * bug fix: mutually referent alien structures now work correctly.
1014 (reported by Rick Taube)
1015 * bug fix: structures defined by WITH-ALIEN can be referred to
1016 within other definitions in the same WITH-ALIEN.
1017 * bug fix: division operators (MOD, TRUNCATE and the like) with
1018 constant zero divisors and integer dividends no longer generate
1020 * bug fix: provide default methods for INPUT-STREAM-P and
1021 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1022 * bug fix: improve the detection and resolution of MOP metacycles
1023 (where effective method computation depends on the generic function
1024 having its effective method computed). (reported by Bruno Haible)
1025 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1026 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1027 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1028 * bug fix: pathnames with relative directory components can be
1029 represented relative to default pathnames.
1030 * optimization: in taking the GCD of bignums, reduce the two bignums
1031 to approximately the same size (using Euclid's algorithm) before
1032 applying the more sophisticated binary GCD. (thanks to Juho
1034 * optimization: COUNT on bitvectors now operates word-at-a-time.
1035 * optimization: ASH with a positive, but not necessarily constant,
1036 (leftwards) shift, when in a modular context, is compiled to a
1038 * fixed some bugs revealed by Paul Dietz' test suite:
1039 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1040 if the corresponding argument is NIL.
1041 ** FORMAT directives accepting a minpad parameter treat negative
1043 ** literal commas as character parameters to FORMAT directives are
1045 ** literal spaces directly after ~<Newline> directives within a
1046 format-logical-block (~:< ~@:>) do not induce :FILL-style
1047 conditional newlines.
1048 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1049 as specified: it no longer includes conditional newlines).
1050 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1052 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1053 :PER-LINE-PREFIX argument does not evaluate to a string.
1055 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1056 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1057 the manual for details; for now, package locks can be disabled by
1058 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1059 if no major problems are found then it is likely that they will be
1060 compiled in unconditionally.
1061 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1062 unsupported operators on all platforms. To load a shared library
1063 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1064 object file, link it into a shared library outside of SBCL and
1065 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1067 * bug workaround: ROOM T and the GC were not working together
1068 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1069 reporting facility aren't preserved. That reporting has been
1070 disabled, so now until and unless someone figures out how to make it
1071 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1072 * minor incompatible change: as threatened around sbcl-0.8.0, the
1073 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1074 The symbols are also exported from SB-PCL for backwards
1075 compatibility, but more so than before SB-PCL should be treated as
1076 an implementation-internal package.
1077 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1079 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1080 It is known as of this release not to work on the Alpha, however.
1081 * fixed bug #167: errors signalled due to illegal syntax in method
1082 bodies are now more legible.
1083 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1084 designators and can hence be used with TYPEP.
1085 * fixed bug #333: CHECK-TYPE now ensures that the type error
1086 signalled, if any, has the right object to be accessed by
1087 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1088 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1089 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1091 * fixed bug #334: programmatic addition of slots using specialized
1092 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1093 and :ALLOCATION :CLASS slots.
1094 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1095 float exponents. (rereported by Peter Seibel)
1096 * fixed a bug: #\Space (and other whitespace characters) are no
1097 longer considered to be macro characters in standard syntax by
1098 GET-MACRO-CHARACTER.
1099 * fixed bug: initialization of condition class metaobjects no longer
1100 causes an instance of the condition to be created. (reported by
1102 * fixed bug: it is now possible to have slots such that
1103 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1104 description is neither :INSTANCE nor :CLASS.
1105 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1106 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1107 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1108 non-standard SLOT-DEFINITION-ALLOCATION values.
1109 * fixed bugs: various race conditions handling exiting threads.
1110 CL-PPCRE's thread torture test now passes.
1111 * fixed arguably-a-bug: GC time proportional to number of threads in
1112 system even when most of them are idle
1113 * optimization: improved performance of BIT and SBIT on bit-vectors.
1114 * .fasl file incompatibility: The fasl file version number has
1115 been incremented because of changes associated with package locks.
1117 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1118 * minor incompatible change: the system no longer provides
1119 optimization hints (conditions which are TYPEP
1120 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1121 mode; these hints are still emitted when higher SPEED optimization
1123 * new contrib module: a sampling profiler (profiling by statistical
1124 sampling, rather than by instrumenting functions) is available as
1125 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1126 * the behaviour of the standard function ED is now customizeable by
1127 third parties through a hook variable: see ED's documentation
1128 string for information on the protocol.
1129 * the compiler no longer emits efficiency notes for (FUNCALL X)
1130 when the type of X is uncertain under default optimization
1132 * fixed bug 276: mutating a binding of a specialized parameter to a
1133 method to something that is not TYPEP the specializer is now
1135 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1136 exhibit the required behaviour on the x86 platform. (thanks to
1137 Peter van Eynde, Eric Marsden and Bruno Haible)
1138 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1139 even for difficult arguments. (reported by Peter Graves)
1140 * fixed bug 141a: the backquote printer now descends quoted
1142 * fixed another bug in backquote printing: no more destructive
1143 modification of the form's list structure. (reported by Brian
1145 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1146 move between its address being taken and the call to
1147 interrupt_thread, fixing a crashing race condition.
1148 * the SB-POSIX contrib implementation has been adjusted so that it
1149 no longer exhibits ridiculously poor performance when constructing
1150 instances corresponding to C structs.
1152 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1153 * minor incompatible change: the sb-grovel contrib now treats C
1154 structures as alien (in the sense of SB-ALIEN) objects rather than
1155 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1156 has implications for memory management of client code
1157 (sb-grovel-returned objects must now be manually managed) and for
1158 type safety (alien objects now have full types).
1159 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1160 used to control emission of compiler diagnostics, rather than the
1161 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1162 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1163 quality should be considered deprecated.
1164 * install.sh now installs the user manual as well
1165 * (not quite a new documentable feature, but worth considering in
1166 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1167 beginnings of a semantically meaningful condition hierarchy is
1168 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1170 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1171 designator as the defaults argument.
1172 * fixed bug: Displaced arrays whose displaced-to array has become
1173 too small now cause ARRAY-DIMENSION to signal an error, providing
1174 for safer bounds-checking. (reported by Bruno Haible)
1175 * fixed bug: DEFCLASS slot definitions with identical :READER and
1176 :WRITER names now signal a reasonable error. (reported by Thomas
1178 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1180 * fixed bug: Invalid slot specification errors now print correctly.
1181 (thanks to Zach Beane)
1182 * fixed bug 320: Shared to local slot value transfers in class
1183 redefinitions now happen corrently from superclasses as
1184 well. (reported by Bruno Haible)
1185 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1187 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1188 improper lists correctly. (reported by Bruno Haible)
1189 * fixed bug 313: source-transform for <fun-name> was erroneously
1190 applied to a call of a value of a variable with name <fun-name>.
1191 (reported by Antonio Menezes Leitao)
1192 * fixed bug 307: The obsolete instance protocol ensures that
1193 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1194 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1195 remove dead unknown-values globs from the middle of the stack.
1196 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1197 easier to resume long computations after using *BREAK-ON-SIGNALS*
1198 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1199 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1200 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1201 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1202 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1203 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1205 * fixed a (fairly theoretical) bug in string printing: if
1206 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1207 does not have array-element-type equal to the most general string
1209 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1210 function name. (thanks to Nikodemus Siivola)
1211 * fixed bug: compile-time format string checker failed on
1212 non-closed ~{. (reported by Thomas F Burdick)
1213 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1214 don't warn on legal loop code involving more than one aggregate
1215 boolean. (thanks to Nikodemus Siivola)
1216 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1217 in keyword position to MAKE-PACKAGE was not being reported as
1219 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1220 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1222 * fixed bugs 280 and 312: the checking for multiple definitions in a
1223 file is less likely to become confused by uses of inline
1225 * fixed bug: the #S reader macro performs the keyword coercion
1226 specified for slot names. (reported by Kalle Niemitalo)
1227 * fixed bug: lambda lists may contain symbols whose names start with
1228 & but are not lambda-list-keywords; their occurrence triggers a
1230 * fixed bug 321: define-method-combination argument lambda lists do
1231 not cause invalid code to be generated when &optional or &aux
1232 variables have default values. (reported by Bruno Haible)
1233 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1234 have CLOS classes; this appears also to have fixed reported
1235 instability in stack exhaustion detection.
1236 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1237 MUFFLE-WARNING when no associated restart is present is now
1239 * optimization: rearranged the expansion of various defining macros
1240 so that each expands into only one top-level form in a
1241 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1243 * optimization: used a previously little-used slot in symbols to
1244 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1246 * fixed some bugs revealed by Paul Dietz' test suite:
1247 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1249 ** Fixed an optimization bug related to inheritance of initforms
1250 from local to shared slots.
1251 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1252 ** CAST optimizer forgot to flush argument derived type.
1253 ** print/read consistency on floats is now orders of magnitude
1254 more likely. (thanks also to Bruno Haible for a similar report
1256 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1257 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1258 for objects of type REAL. Make it so.
1259 ** SXHASH is specified (infelicitously) to respect similarity,
1260 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1261 it so. (thanks to Markus Ziegler)
1262 ** on the Alpha, the compiler succeeds in compiling functions
1263 returning a known number of arguments greater than 63.
1264 ** fixed handling of invalid NIL arguments in keyword position
1266 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1268 ** provided a readably-printable representation for RANDOM-STATE
1270 ** ensured that pathnames, where they have a namestring, always
1271 print using #P"..." syntax.
1273 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1274 * Support for the forthcoming 2.0 version of the NetBSD kernel
1275 running on x86 hardware has been added. (thanks to Perry
1276 E. Metzger most immediately, and others for their past work)
1277 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1278 current release version; SBCL's OpenBSD support had been broken
1279 since about the time of OpenBSD's switch to ELF binary format.)
1280 * [placeholder for DX summary]
1281 ** user code with &REST lists declared dynamic-extent, under high
1282 speed or space and low safety and debug optimization policy.
1283 * The manual has been converted to Texinfo format and the debugger
1284 chapter from the cmucl manual has been added.
1285 * A facility has been added to extract documentation strings from
1286 sbcl and store them as Texinfo-formatted snippets for inclusion in
1287 the manual (via Texinfo's @include directive)
1288 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1289 the test case to Dave Roberts)
1290 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1291 pointers. (reported by Sean Ross)
1292 * bug fix: PROFILE output is printed nicely even for large numerical
1293 values. (thanks to Zach Beane)
1294 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1295 greater than 32 handle EOF correctly.
1296 * bug fix: on X86 an immediate argument of the IMUL instruction is
1297 correctly printed in disassembly. (thanks to Lutz Euler)
1298 * bug fix: class slots in redefined classes preserve their old
1299 values. (thanks to Bruno Haible and Nikodemus Siivola)
1300 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1301 succeeds. (reported by Marco Baringer)
1302 * fixed some bugs revealed by Paul Dietz' test suite:
1303 ** READ-SEQUENCE now works on ECHO-STREAMs.
1304 ** RATIONALIZE works more according to its specification. (thanks
1307 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1308 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1309 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1310 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1311 should matter to you only if you rebind the printer control
1312 variables and then find you want different bindings in the
1313 debugger than in the ordinary execution of your program.)
1314 * The runtime build system has been tweaked to support building
1315 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1316 assembler and linker. (thanks to Nikodemus Siivola)
1317 * Unbound, undefined, undeclared variables now trigger full
1318 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1319 more often programmer error than deliberate exploitation of undefined
1321 * optimization: the hash algorithm for strings has changed to one
1322 that is less vulnerable to spurious collisions. (thanks to Juho
1324 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1325 needless bounds checking. (thanks to Juho Snellman)
1326 * optimization: implemented multiplication as a modular
1327 (UNSIGNED-BYTE 32) operation on the PPC backend.
1328 * fixed some bugs revealed by Paul Dietz' test suite:
1329 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1331 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1332 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1333 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1335 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1337 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1339 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1340 their output stream on EOF from read.
1341 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1342 have been read to end-of-file.
1343 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1345 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1346 description of determination of which consecutive characters
1348 ** Printing the "Space" character with escaping on now yields "#\\ ",
1349 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1350 ** Reading floating-point numbers with *READ-BASE* set to a number
1351 less than 10 works correctly.
1352 ** Reading floating-point numbers with *READ-BASE* set to a number
1353 more than 10 works correctly.
1354 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1355 the readtable currently in effect.
1357 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1358 * minor incompatible change: parsing of namestrings on a physical
1359 (Unix) host has changed; numbers after the final #\. in a
1360 namestring are no longer interpreted as a version field. This is
1361 intented to be largely invisible to the user, except that the
1362 meaning of the namestring "*.*.*" has changed: it now refers to a
1363 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1364 should usually be replaced by
1365 (make-pathname :name :wild :type :wild :version :wild)
1366 with the added benefit that this is more likely to be portable.
1367 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1368 option to OPEN now signals an error if the file being opened
1369 exists; this may have an impact on existing code.
1370 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1371 signals received, on the PowerPC platforms (both Linux and
1372 Darwin). (thanks to Pierre Mai for pointing out the location of
1374 * several fixes on OS X: The system now builds and runs cleanly on
1375 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1376 (thanks to Brian Mastenbrook)
1377 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1378 second-resolution: integer multiples of 1/3600 between -24 and 24.
1379 (thanks to Vincent Arkesteijn)
1380 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1381 type when called with 1 argument; PEEK-CHAR checked type of
1382 PEEK-TYPE only after having read first character from a
1383 stream. (reported by Peter Graves)
1384 * bug fix: the garbage collector now has much better locality
1385 behaviour, and in particular no longer treats all memory as being
1386 exhausted when many small objects point to each other in a deeply
1388 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1389 recognized as being TYPEP their class.
1390 * bug fix: the PUSHNEW documentation string has been corrected.
1391 (thanks to Vincent Arkesteijn)
1392 * bug fix: defaulting of the value for the last of an atypically
1393 large number of multiple values being bound was not being
1394 performed correctly on the Alpha or PPC platforms
1395 * optimization: implemented multiplication as a modular
1396 (UNSIGNED-BYTE 32) operation on the x86 backend.
1397 * optimization: SEARCH on simple-base-strings can now be open-coded.
1398 (see also contrib/compiler-extras.lisp for inspiration for
1399 teaching the compiler about the Boyer-Moore algorithm).
1400 * value, returned by MAX (and MIN) called with several EQUALP, but
1401 not EQL, arguments now does not depend on compiler settings.
1402 * fixed some bugs revealed by Paul Dietz' test suite:
1403 ** in stack analysis liveness information is propagated from
1404 non-local entry points.
1405 ** pathwise CAST removing failed when the CAST node did not start
1407 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1408 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1410 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1411 host is already defined.
1412 ** RENAME-FILE works on streams instead of signalling an internal
1414 ** PEEK-CHAR uses the current readtable when determining whether
1415 or not a character is whitespace.
1416 ** MERGE-PATHNAMES handles the case when the pathname does not
1417 specify a name while the default-pathname specifies a version.
1418 ** Pathnames now stand a better chance of respecting print/read
1420 ** Attempting to use standardized file system operators with a
1421 pathname with invalid :DIRECTORY components signals a
1423 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1425 ** DIRECTORY on logical pathnames is more correct.
1426 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1427 signal a TYPE-ERROR if their argument is not a stream.
1428 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1429 designator argument does not designate a stream.
1430 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1431 examining the synonym.
1432 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1434 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1435 element-type larger than ([UN]SIGNED-BYTE 32).
1437 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1438 * When built with the :SB-FUTEX feature, threaded builds now take
1439 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1440 for faster/more reliable mutex and condition variable support.
1441 * Incompatible change (but one you probably shouldn't have been using
1442 anyway): the interface and code for arbitrating between multiple
1443 threads in the same user session has been redesigned.
1444 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1445 error when called without an explicit environment argument.
1446 (thanks to Nikodemus Siivola)
1447 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1448 confused about their position. (thanks to Adam Warner and Gerd
1450 * bug fix: Pretty printing backquoted forms with unquotations in the
1451 argument list position of various code constructs such as LAMBDA
1452 now works correctly. (reported by Paul Dietz)
1453 * bug fix: Pretty printing unquotations no longer loses all
1454 stream position information.
1455 * optimization: performance of string output streams is now less
1456 poor for multiple small sequence writes.
1457 * optimization: performance of CSUBTYPEP in the presence of complex
1458 expressions involving CONS and NOT many times has been improved.
1459 (reported by Paul Dietz)
1460 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1462 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1463 streams for the REPL, for the debugger, and so on. (thanks to
1465 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1466 * fixed some bugs revealed by Paul Dietz' test suite:
1467 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1468 is ignored if it is too small, rather than propagating through
1469 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1470 ** extremely complex negations of CONS types were not being
1471 sufficiently canonicalized, leading to inconsistencies in
1473 ** VALUES tranformer lost derived type.
1475 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1476 * fixed a bootstrapping bug: the build process no longer assumes
1477 that the various BOOLE-related constants have the same value in
1478 host and target lisps. (noted by Paul Dietz' test suite on an
1479 SBCL binary built from CLISP)
1480 * The system can now be dynamically linked on the MIPS platform,
1481 which enables dynamic loading of foreign code from Lisp. (thanks
1482 to Ralf Baechle for discussions on the MIPS ABI)
1483 * The system now records debugging information for its own source
1484 files in a filesystem-position-independent manner, relative to
1485 the "SYS" logical host.
1486 * fixed a compiler bug: MV-LET convertion did not check references
1487 to the "max args" entry point. (reported by Brian Downing)
1488 * tweaked disassembly notes to be less confident about proclaiming
1489 some instruction as an LRA. (thanks to Brian Downing)
1490 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1491 now each have their own history, command character, and other
1492 characteristics. (thanks to David Lichteblau)
1493 * fixed some bugs revealed by Paul Dietz' test suite:
1494 ** compiler failure in compiling LOGAND expressions including a
1496 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1497 shift greater than 32.
1498 ** FLUSH-DEST did not mark blocks for type check regeneration.
1499 ** HANDLER-CASE failed to accept declarations in handler clauses
1500 in some circumstances.
1502 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1503 * New code in contrib/sb-introspect (still probably not entirely
1504 stable yet) provides some support for smart Lisp development
1505 environments like SLIME.
1506 * The conditions signalled for errors occurring when loading .fasl
1507 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1508 in a way which should help ASDF recover gracefully.
1509 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1510 stuff has been cleaned up. If you code contrib/ stuff, this might
1511 affect you, and you can look at contrib/README, contrib/STANDARDS,
1512 and/or the 0.8.4.27 diff to check.
1513 * In full calls the compiler now does not generate checks for declared
1514 argument types for all arguments.
1515 * various threading fixes
1516 ** and some experimental patches which didn't make it into
1517 the main tree for this release, but which are shipped in
1518 contrib/experimental-thread.patch as a possible fix for some
1519 failures (deadlock, spinning...) in GC-intensive multithreaded
1521 * fixed PPC build problem (source code incompatibility of different
1522 library versions): added offsetof() hackery which attempts to divine
1523 where glibc maintainers put uc_mcontext today
1524 * fixed bug 282: compiler does not trust type assertions while passing
1525 arguments to a full call.
1526 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1527 &OPTIONAL VALUES type parameter.
1528 * fix bug 214: algorithm for noting rejected templates is now more
1529 similar to that of template seletion. (also reported by rydis on
1531 * fixed bug 141b: printing backquoted information readably and prettily
1532 inserts a space where necessary.
1533 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1534 (CHARACTER 10) are now reported as errors, rather than propagated
1535 as unknown types. (reported by piso on #lisp)
1536 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1537 with values NIL and :ERROR. (thanks to Milan Zamazal)
1538 * fixed bug 191c: CLOS now does proper keyword argument checking as
1539 described in CLHS 7.6.5 and 7.6.5.1.
1540 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1541 longer raise an error; further, using a list as a for-as-arithmetic
1542 counter now raises a meaningful error.
1543 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1544 now understood by sequence creation functions such as MAKE-SEQUENCE
1546 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1547 a STRING-INPUT-STREAM.
1548 * compiler enhancement: SIGNUM is now better able to derive the type
1550 * type declarations inside WITH-SLOTS are checked. (reported by
1552 * fixed some bugs revealed by Paul Dietz' test suite:
1553 ** incorrect optimization of TRUNCATE for a positive first
1554 argument and negative second.
1555 ** compiler failure in let-convertion during flushing dead code.
1556 ** compiler failure while deriving type of TRUNCATE on an
1557 interval, containing 0.
1558 ** ASH of a negative bignum by a negative bignum count now returns
1560 ** intersection of CONS types now canonicalizes properly, fixing
1561 inconsistencies in SUBTYPEP.
1563 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1564 * incompatible change: The --disable-debugger command line
1565 option now clobbers the debugger at a more fundamental
1566 level, by redefining #'INVOKE-DEBUGGER instead of by
1567 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1568 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1569 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1570 enter the debugger REPL and then suffer recursive errors
1571 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1572 while under the new system, BREAK in --disable-debugger mode
1573 terminates the system just as an unhandled error would.
1574 * fixed compiler performance when processing loops with a step >1;
1575 * bug fix: DOCUMENTATION now retrieves generic function
1576 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1577 support has been systematized, and now supports the methods
1578 specified by ANSI, along with a default method and a method for
1579 slot documentation. (reported by Nathan Froyd)
1580 * bug fix: effective methods associated with a generic function are
1581 no longer cached over a change of that generic function's method
1582 combination. (reported by Andreas Fuchs)
1583 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1584 names a non-existent directory. (thanks to Andreas Fuchs)
1585 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1586 the PPC platform, lead to stack corruption; this has been fixed.
1587 (reported by Rainer Joswig)
1588 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1589 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1590 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1591 arithmetic in general, is now much more reliable on the PPC
1593 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1594 platform now returns the right answer.
1595 * optimization: restored some effective method precomputation in
1596 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1597 precomputation is now tunable.
1598 * optimization: compiler-internal data structure use has been
1599 reviewed, and changes have been made that should improve the
1600 performance of the compiler by about 20%.
1601 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1602 simple-base-strings and simple-bit-vectors is improved.
1603 * optimization: the optimization of 32-bit logical and arithmetic
1604 functions introduced in version 0.8.3 on the x86 has been
1605 implemented on the mips, ppc and sparc platforms; an
1606 implementation of the same facility, but for 64-bit arithmetic,
1607 has been added for the alpha.
1608 * microoptimization: the compiler is better able to make use of the
1609 x86 LEA instruction for multiplication by constants.
1610 * bug fix: in some situations compiler did not report usage of
1611 generic arithmetic in (SPEED 3) policy.
1612 * bug 145b fix: compiler used wrong type specifier while converting
1613 MEMBER-types to numeric.
1614 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1616 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1617 index argument now works properly on the Alpha platform.
1618 * bug fix: floating point exception treatment on the Alpha platform
1620 * bug fix: FILE-POSITION works much better on string input and
1621 output streams. (thanks to Nikodemus Siivola)
1622 * bug fix: many threading/garbage collection symptoms sorted.
1623 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1624 might be pseudo-atomic.
1625 * internal change: Stopping for GC is now done with signals not
1626 ptrace. GC is now done in whichever thread wanted it, instead of
1628 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1630 * bug fix: result form in DO is not contained in the implicit
1632 * incompatible change: ICR structure is changed; the value part of
1633 CONTINUATION is now called LVAR; corresponding functions are
1634 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1635 * added type deriver for ISQRT (thanks to Robert E. Brown).
1636 * bug fix: better support for loading from the command line when an
1637 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1639 * fixed some bugs revealed by Paul Dietz' test suite:
1640 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1641 ** ROUND and FROUND now give the right answer when given very
1642 small float arguments.
1643 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1645 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1646 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1647 ** LCM should return a non-negative integer.
1648 ** PARSE-INTEGER returned the index of a terminator instead of the
1649 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1650 ** PARSE-INTEGER returned an incorrect index being applied to a
1652 ** LCM with two arguments of 0 returns 0 rather than signalling
1654 ** unsigned addition of a 32-bit constant with the high bit set no
1655 longer causes an internal compiler error on the x86.
1656 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1657 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1658 with negative last argument.
1659 ** byte specifiers with zero size and position no longer cause
1660 an error during type derivation.
1661 ** bignum multiplication on the Alpha platform now returns the
1663 * porting: The system now builds on SuSE AMD64, although it still
1664 generates a 32-bit binary.
1665 * .fasl file incompatibility: The fasl file version number has
1666 been incremented (because of the changes to internal compiler
1667 data structures referred to above).
1669 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1670 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1671 more accurately, on the Darwin kernel running on PowerPC hardware.
1672 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1673 * Compiler code deletion notes now signal a condition of type
1674 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1675 an associated MUFFLE-WARNING restart.
1676 * The compiler now performs limited argument count validation of
1677 constant format strings in FORMAT, and where appropriate in ERROR,
1678 CERROR and WARN. (thanks to Gerd Moellmann)
1679 * New ASDF-INSTALL contrib can be used for automatic download and
1680 installation of third-party Lisp code from CCLAN or other sites
1682 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1683 which forces another thread to execute a function supplied by the
1685 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1686 now accept and act upon their :ELEMENT-TYPE keyword argument.
1687 (reported by Martin Atzmueller, Edi Weitz)
1688 * bug fix: FILE-POSITION now accepts position designators up to
1689 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1690 is the greater. (thanks to Patrik Nordebo)
1691 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1692 Kalle Olavi Niemitalo)
1693 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1694 printable. (reported by Eric Marsden)
1695 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1696 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1697 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1698 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1699 (thanks to Stig Sandoe)
1700 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1701 UNREAD-CHAR now correctly includes the unread character in the
1702 target sequence. (thanks to Gerd Moellmann)
1703 * bug fix in threaded builds: the system can now be suspended and
1704 resumed by shell job control with minimal disruption.
1705 * bug fixes in times and timezones >2038AD
1706 * better handling of "where is GNU make?" problem in build scripts
1707 (thanks to Nikodemus Siivola)
1708 * new optimization: inside a named function any reference to a
1709 function with the same name is considered to be a self-reference;
1710 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1711 optimization quality.
1712 * new optimization on x86: logical functions and + now have
1713 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1714 used when the result is truncated to 32 bits.
1715 * VALUES declaration is partially enabled.
1716 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1717 * fixed some bugs revealed by Paul Dietz' test suite:
1718 ** The system now obeys the constraint imposed by
1719 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1720 types form a lattice under type intersection.
1721 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1722 ** ASSOC now ignores NIL elements in an alist.
1723 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1724 and (1+ MOST-POSITIVE-FIXNUM) answers.
1725 ** The addition of a method with invalid qualifiers to a generic
1726 function does not cause an error to be signalled immediately;
1727 a warning is signalled, and the error is generated only on
1728 calling the generic function.
1729 * changed .fasl file version number, in part to add type codes for
1730 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1731 obscure ANSI requirements
1733 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1734 * fixed bug 148: failure to inline-expand a local function left
1735 garbage, confusing the compiler.
1736 * fixed bugs 3cd: structure slot readers perform type check if the
1737 slot can have an invalid value (i.e. it is either not initialized
1738 or can be written with a less specific slot writer).
1739 * bug fix: the compiler now traps array references to elements off
1740 the end of an array; previously, the bounds checking in some
1741 circumstances could go off-by-one.
1742 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1744 * type declarations for array element types now obey the description
1745 on the CLHS page "Declaration TYPE", as per discussions on
1746 sbcl-help around 2003-05-08. This means that a declaration
1747 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1748 declaration, all references to BAR will be asserted or assumed
1749 (with THE, so dependent on compiler policy) to involve objects of
1750 type FOO. Note that no such declaration is implied in
1751 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1752 * declared types of functions from the "Conditions"
1753 chapter. (reported by Paul Dietz)
1754 * bug fix: CERROR accepts a function as its first argument.
1755 * bug fix: NTH an NTHCDR accept a bignum as index
1756 arguments. (reported by Adam Warner)
1757 * optimization: character compare routines now optimize comparing
1758 against a constant character. (reported by Gilbert Baumann)
1759 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1760 argument now works properly on the MIPS platform.
1761 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1762 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1763 declared to have a functional type, does not cause an error
1765 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1766 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1767 * fixed bug: if last continuation of a deleted block has a
1768 destination, this destination should be deleted too. (reported by
1770 * fixed a bug in the bootstrap process: the host compiler's values
1771 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1772 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1773 test case from Patrik Nordebo)
1774 * improved the ability of the disassembler on the PPC platform to
1775 provide helpful disassembly notes.
1776 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1777 the class in more cases than previously.
1778 * bug fix: FILE-POSITION now understands :START and :END for
1779 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1780 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1781 8) data. (thanks to David Lichteblau)
1782 * bug fix: it is possible to add a method to a generic function
1783 without lambda list.
1784 * bug fix: reader failed to signal END-OF-FILE inside an
1785 object representation. (reported by Nikodemus Siivola)
1786 * fixed some bugs revealed by Paul Dietz' test suite:
1787 ** LAST and [N]BUTLAST should accept a bignum.
1788 ** condition slot accessors are methods.
1789 ** (VECTOR NIL) is a subtype of STRING.
1791 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1792 * minor incompatible change: some nonsensical specialized lambda
1793 lists (used in DEFMETHOD) which were previously ignored now signal
1795 * minor incompatible change: the system is now aware of the types of
1796 variables in the COMMON-LISP package, and will signal errors for
1797 most violations of these type constraints (where previously they
1798 were silently accepted).
1799 * minor incompatible change: COMPILE-FILE now uses the freedom
1800 afforded (ANSI 3.2.2.3) to use derived function types for
1801 functions defined in the same file. This also permits the system
1802 to warn on static type mismatches and function
1803 redefinition. (Currently it does not work with high DEBUG level.)
1804 * minor incompatible change: VALUES declaration is disabled.
1805 * When issuing notes, the compiler now signals a condition of type
1806 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1807 restart for use in user handlers. It is expected that the
1808 COMPILER-NOTE condition will eventually become a condition
1809 supertype to a hierarchy of note types, which will then be
1810 handleable in a similar fashion. However, at the moment, no such
1811 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1812 but it's an implementation detail, not a classification for the
1814 * Changes in type checking closed the following bugs:
1815 ** type checking of unused values (192b, 194d, 203);
1816 ** template selection based on unsafe type assertions (192c, 236);
1817 ** type checking in branches (194bc).
1818 * A short form of VALUES type specifier has ANSI meaning (it has
1819 increased the number of situations when SBCL cannot perform type
1821 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1822 keywords or constants is permissible.
1823 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1824 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1825 classes. (thanks to Antonio Martinez)
1826 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1827 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1828 argument to a pretty-print stream. Instead, it leaves any such
1829 operations to DESCRIBE-OBJECT methods.
1830 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1832 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1833 lambda lists. (thanks to David Lichteblau)
1834 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1835 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1836 (thanks to Gerd Moellmann)
1837 * fixed bug 246: increased compilation speed of long
1838 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1840 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1842 * DEFCONSTANT now throws a condition of type
1843 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1844 constant to a non-EQL value; CONTINUE and ABORT restarts
1845 respectively change and preserve the value.
1846 * fixed bug 63: The code walker, part of the implementation of CLOS,
1847 is now better at handling symbol macros.
1848 * bug fix: There is no longer an internal implementation type named
1849 CL:LENGTH. (reported by Raymond Toy)
1850 * bug fix: In macro-like defining macros/special operators the
1851 implicit block does not enclose the lambda list.
1852 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1853 enclosed in parentheses) are not suitable as type specifiers, and
1854 their use properly signals an error now.
1855 * bug fix: An argument count mismatch for a type specifier in code
1856 being compiled no longer causes an unhandled error at compile
1857 time, but signals a compile-time warning.
1858 * fixed simple vector readable printing
1859 * bug fix: DESCRIBE takes more care over whether the class
1860 precedence list slot of a class is bound before accessing it.
1861 (reported by Markus Krummenacker)
1862 * bug fix: FORMATTER can successfully compile pretty-printer format
1863 strings which use variants of the ~* directive inside.
1864 * bug fix: SEARCH now applies its TEST predicate to the elements of
1865 the arguments in the correct order. (thanks to Wolfhard Buss)
1866 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1867 in type inference. (thanks to Robert E. Brown)
1868 * bug fix: Reading in symbols with an explicit package name of ""
1869 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1870 * fixed some bugs revealed by Paul Dietz' test suite:
1871 ** NIL is now allowed as a structure slot name.
1872 ** Arbitrary numbers, not just REALs, are allowed in certain
1873 circumstances in LOOP for-as-arithmetic clauses.
1874 ** Multiple class redefinitions before slot access no longer
1875 causes a type error.
1876 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1877 association between the name and a class.
1878 ** Generic functions with non-standard method-combination and over
1879 five methods all of which return constants no longer return NIL
1880 after the first few invocations. (thanks to Gerd Moellmann)
1881 ** CALL-NEXT-METHOD with no arguments now passes the original
1882 values of the arguments, even in the presence of assignment.
1883 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1885 ** DISASSEMBLE works with closures and funcallable instances.
1886 ** ADD-METHOD now returns the generic function, not the new method.
1887 ** FIND-METHOD signals an error if the lengths of the specializers
1888 is incompatible with the generic function, even if the ERRORP
1890 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1891 which its argument is a member.
1892 ** DEFCLASS only redefines the class named by its class-name
1893 argument if that name is the proper name of the class;
1894 otherwise, it creates a new class.
1895 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1896 of the UNBOUND-SLOT condition to the name of the slot.
1897 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1898 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1899 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1900 SLOT-MAKUNBOUND in the specified fashion.
1902 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1903 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1904 cross-compilation host. As a consequence, we can now bootstrap our
1905 way up to SBCL starting with a bare gcc toolchain and human-readable
1906 source code (first the source to CLISP, then the source to SBCL).
1907 * A contributed module containing a partial implementation of the
1908 simple-streams interface has been included. (thanks to Rudi
1910 * A contributed module implementing the RFC1321 Message Digest
1911 Algorithm, known as MD5, has been included.
1912 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1913 no longer has any effect, as the code controlled by this feature
1914 has been deleted. (As far as we know, no-one has ever built using
1915 this feature, and its semantics were confused in any case).
1916 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1917 work on conditions (as required by the ANSI specification),
1918 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1919 expected behaviour on conditions. Users should note, however,
1920 that such behaviour is not required by the ANSI specification,
1921 and so use of this behaviour may render their code unportable.
1922 * fixed some bugs revealed by Paul Dietz' test suite:
1923 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1925 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1926 ** SXHASH on condition objects no longer returns NIL.
1927 ** :ALLOCATION :CLASS slots are better treated; their values are
1928 updated on class redefinition, and initforms inherited from
1929 superclasses are applied.
1930 ** REMOVE-METHOD returns its generic function argument even when
1931 no method was removed.
1932 ** SHARED-INITIALIZE now initializes the values of the requested
1933 slots, including those with :ALLOCATION :CLASS.
1934 ** ALLOCATE-INSTANCE now works on structure classes defined via
1935 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1937 ** SLOT-EXISTS-P now works on conditions, as well as structures
1939 ** MAKE-LOAD-FORM now has the required methods on
1940 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1941 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1942 keyword, and now implements the SLOT-NAMES argument.
1943 ** methods with &OPTIONAL arguments no longer allow too many
1944 arguments to be passed in the call without error.
1945 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1946 option is consistent with the required arguments of the generic
1947 function lambda list.
1948 * bug fix: REQUIRE accepts a string designator. (Thanks to
1950 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1951 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1952 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1953 * bug fix: The compiler now checks for duplicated variables in macro
1955 * bug fix: SETQ on globals returns the correct value.
1956 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1957 not a COMPILER-ERROR (followed by some other strange error on
1958 choosing the CONTINUE restart).
1959 * bug fix: make.sh and friends are now more consistent in the way that
1960 they look for GNU "make".
1962 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1963 * experimental native threads support (on x86 Linux >=2.4 only).
1964 This is not compiled in by default: you need to add :SB-THREAD to
1965 the target features. See the "Beyond ANSI" chapter of the manual
1967 * fix for longstanding nonANSIism: The old distinction between
1968 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1969 The return value from CL:FIND-CLASS is now a CLOS class, and
1970 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1971 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1972 * An interface to the MetaObject Protocol, as described in Kiczales,
1973 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1974 MIT Press, 1991, is available from the SB-MOP package.
1975 * incompatible change: the SB-PCL package should now be considered
1976 a private implementation detail, and no longer a semi-private MOP
1978 * minor incompatible change: due to rearrangement for threads, the
1979 control stack and binding stack are now allocated at arbitrary
1980 addresses instead of being hardcoded per-port. Users affected by
1981 this probably have to be doing advanced things with shared
1982 libraries, and will know who they are.
1983 * minor incompatible change: Previously, all --eval forms used were
1984 processed with READ before any of them were processed with EVAL.
1985 Now each --eval form is processed with both READ and EVAL before
1986 the next --eval form is processed. (Thus package operations like
1987 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1988 work as the user might reasonably expect.)
1989 * minor incompatible change: *STANDARD-INPUT* is now only an
1990 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1992 * minor incompatible change: Y-OR-N-P is now character-oriented,
1993 not line oriented. Also, YES-OR-NO-P now works without errors.
1994 (thanks to Antonio Martinez)
1995 * sb-aclrepl module improvements: an integrated inspector, added
1996 repl features, and a bug fix to :trace command.
1997 * Known functions, which cannot be open coded by the backend, are
1998 considered to be able to check types of their arguments. (fixing
1999 a bug report by Nathan J. Froyd)
2000 * fixed a bug in computing method discriminating functions: It is
2001 now possible to define methods specialized on classes which have
2002 forward-referenced superclasses. (thanks to Gerd Moellmann)
2003 * fixed evaluation order in optional entries (reported by Gilbert
2005 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2006 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2007 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2008 required-by-AMOP NAME argument, as well as CLASS and
2009 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2010 * fixed bug 20: DEFMETHOD can define methods using names that are
2011 not the proper names of classes to designate class specializers.
2012 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2013 Unix file descriptors, instead of blowing up. (thanks to Antonio
2015 * Garbage collection refactoring: user-visible change is that a
2016 call to the GC function during WITHOUT-GCING will not do garbage
2017 collection until the end of the WITHOUT-GCING. If you were doing
2018 this you were probably losing anyway.
2019 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2020 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2021 TYPEP the latter but not the former.
2022 * The compiler issues a full WARNING for calls to undefined functions
2023 with names from the CL package.
2024 * MAP-INTO for a vector destination is open coded. (reported by
2025 Brian Downing on c.l.l)
2026 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2027 documentation string.
2028 * fixed some bugs revealed by Paul Dietz' test suite:
2029 ** COPY-ALIST now signals an error if its argument is a dotted
2031 ** Condition slots are now accessed more correctly in the presence
2032 of multiple initargs for a given slot.
2033 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2034 exclude restarts of the same name associated with a different
2036 ** DEFCLASS of forward-referenced classes with another
2037 forward-referenced class in the superclasses list no longer
2039 ** Condition slots are now initialized once each, not multiple
2040 times. (thanks to Gerd Moellmann)
2041 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2043 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2044 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2045 arguments contain duplicated elements.
2046 ** RESTART-CASE understands local macros.
2047 ** RESTART-CASE associates exactly its own restarts with a condition.
2048 ** ENDP in safe mode checks its argument to be of type LIST.
2049 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2050 in question is unbound.
2051 ** Optimized MAKE-INSTANCE functions no longer cause internal
2052 assertion failures in the presence of duplicate initargs.
2053 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2054 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2055 propagated even in the ctor optimized implementation of
2057 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2059 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2060 * a better implementation of SXHASH on (simple) bit vectors,
2061 measured both in execution speed and in distribution of results
2062 over the positive fixnums, has been installed. Likewise, a better
2063 implementation of EQUAL for simple bit vectors is now available.
2064 * fixed CEILING optimization for a divisor of form 2^k.
2065 * fixed bug 240 (emitting extra style warnings "using the lexical
2066 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2067 by Antonio Martinez)
2068 * fixed SXHASH, giving different results for NIL depending on type
2069 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2070 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2071 variable. (found by Rolf Wester)
2072 * SBCL does not ignore type declarations for special
2073 variables. (reported by rif on c.l.l 2003-03-05)
2074 * some bug fixes in contrib/sb-aclrepl/
2075 * fixed some bugs revealed by Paul Dietz' test suite:
2076 ** a bug in the CONS type specifier, whereby the CAR and CDR
2077 types got intertwined, has been fixed;
2078 ** the type system is now able to reason about the interaction
2079 between INTEGER and RATIO types more completely;
2080 ** APPEND, [N]REVERSE and NRECONC check that those their
2081 arguments, which must be proper lists, are really so;
2082 ** An array specialized to be unable to hold elements has been
2083 implemented, as required -- yes, really -- by ANSI;
2084 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2085 on malformed property lists;
2087 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2088 * incompatible packaging change: in line with Unix convention,
2089 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2090 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2091 * REQUIRE and PROVIDE are now optionally capable of doing something
2092 useful. See the documentation string for REQUIRE.
2093 * infrastructure for a managed SBCL contrib system: contributed
2094 modules in this release include:
2095 ** the ASDF system definition facility;
2096 ** an interface to the BSD Sockets API;
2097 ** an ACL-like convenience interface to the repl;
2098 (thanks to Kevin Rosenberg)
2099 ** an implementation of ROTATE-BYTE, with efficient implementation
2101 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2102 gives a closer approximation to the right answer than previously.
2103 (thanks to Raymond Toy)
2104 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2105 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2106 optional environment arguments, as required by ANSI.
2107 * fixed bugs in other functions taking environment objects, allowing
2108 calls with an explicit NIL environment argument to be compiled
2110 * fixed bug 228: primary return values from
2111 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2112 COMPILE or FUNCTION.
2113 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2114 :TYPE will no longer signal errors on innocuous objects.
2115 * fixed bug 231b: SETQ is better at respecting type declarations in
2116 the lexical environment.
2117 * fixed a bug in DEFCLASS: classes named by symbols with no or
2118 unprintable packages can now be defined.
2119 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2120 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2121 * fixed compiler failure related to checking types of functions.
2122 (reported by Robert E. Brown)
2123 * the compiler is now much more consistent in its error-checking
2124 treatment of bounding index arguments to sequence functions: in
2125 (SAFETY 3) code, errors will be signalled in almost all cases if
2126 invalid sequence bounding indices are passed to functions defined
2127 by ANSI to operate on sequences.
2128 * fixed a bug in the build procedure: documentation of SBCL-specific
2129 packages is now preserved and available in the final Lisp image.
2130 * lifted FDEFINITION lookup out of loops in the implementation of
2131 many list operations. (thanks to Robert E. Brown)
2132 * fixed a bug in the reader: the #n# reader macro now works for
2133 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2134 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2135 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2136 E. Brown and Tony Martinez respectively)
2137 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2138 SLOT-BOUNDP: in particular, it is now able to optimize them much
2139 better, and is now not vulnerable to having packages renamed.
2140 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2141 always called when a slot is not present in an instance. (thanks
2143 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2144 longer signal an error on unreading a character following EOF on
2145 the previous constituent stream. (thanks to Tony Martinez)
2146 * fixed some bugs revealed by Paul Dietz' test suite:
2147 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2148 not just nonnegative fixnums;
2149 ** the logical bit-array operators such as BIT-AND now accept an
2150 explicit NIL for their "opt-arg" argument (to indicate a
2151 freshly-consed result bit-array);
2152 ** ELT now signals an error on an invalid sequence index in safe
2154 ** the type system is now cleverer about negations of numeric
2155 types, and consequently understands the BIGNUM and RATIO types
2157 ** the type system is now cleverer about the interaction between
2158 INTEGER and RATIO types: while bugs still remain, many more
2159 cases are accurately computed;
2160 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2161 if it is in the last clause;
2162 ** CONSTANTLY now correctly returns a side-effect-free function in
2164 ** DECLARE is no longer treated as a special-operator; in
2165 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2166 * incremented fasl file version number due to the change in the
2167 DEFSTRUCT-SLOT-DESCRIPTION structure.
2169 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2170 * minor incompatible change: code processed by the "interpreter" or
2171 EVAL now has a compilation optimization policy of (DEBUG 2)
2172 (changed from (DEBUG 1)) to improve debuggability of interactive
2173 development, and to allow the use of the debug RETURN command in
2175 * an experimental implementation of the RETURN command for the
2176 debugger has been included. (thanks to Frederik Kuivinen)
2177 * fixed bug 62: constraints were not propagated into a loop.
2178 * fixed bug in embedded calls of SORT (reported and investigated by
2180 * fixed some bugs revealed by Paul Dietz' test suite:
2181 ** printing and reading of arrays with some dimensions having
2182 length 0 (thanks to Gerd Moellmann);
2183 ** BOA constructor with &AUX argument without a default value does
2184 not cause a type error;
2185 ** CONSTANTP now returns true for all self-evaluating objects.
2187 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2188 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2189 accessors that are related by inheritance, as specified in the
2190 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2191 Valtteri Vuorikoski)
2192 * The compiler is now able to inline functions that were defined in
2193 a complex lexical environment (e.g. inside a MACROLET).
2194 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2196 * Support for the upcoming FreeBSD-5.0 release has been included.
2197 (thanks to Dag-Erling Smorgrav)
2198 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2199 effect when it is not in a toplevel context.
2200 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2201 stand a better chance of being correct. (thanks to Gerd
2203 * fixed bug in COERCE, which now signals an error on coercing a
2204 rational to a bounded real type which excludes the expected
2206 * The compiler is now able to derive types more accurately from the
2207 COERCE and COMPILE functions.
2208 * fixed bug 223: functional binding is considered to be constant
2209 only for symbols in the CL package.
2210 * fixed bug 231: SETQ did not check the type of a variable being set
2211 (reported by Robert E. Brown)
2212 * A new optimization for MAKE-INSTANCE has been included, fixing
2213 various bugs (including relating to :ALLOCATION :CLASS slots and
2214 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2216 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2217 ** As required by ANSI, LOOP now disallows anonymous collection
2218 clauses such as COLLECT I in conjunction with aggregate boolean
2219 clauses such as THEREIS (= I 1);
2220 ** LOOP now signals an error when any variable is reused in the
2221 same loop (including the potentially useful construct analogous
2222 to WITH A = 1 WITH A = (1+ A);
2223 ** IT is only a special loop symbol within the first clause of a
2224 conditional loop clause;
2225 ** LOOP with a typed iteration variable over a hashtable now
2226 signals a type error iff it should.
2227 * fixed some other bugs revealed by Paul Dietz' test suite:
2228 ** FILE-STREAM now names the class previously known as FD-STREAM;
2229 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2230 argument) no longer signals an error;
2231 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2232 of the slot symbol, rather than using the current package
2233 ((:CONC-NAME "") continues to intern the slot's name in the
2235 * incremented fasl file version number, because of the incompatible
2236 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2237 of the new implementation of DEFINE-COMPILER-MACRO.
2239 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2240 * Support for building SBCL for MIPS platforms running in
2241 little-endian mode has now been checked in, and basic
2242 functionality on said platforms verified.
2243 * minor incompatible change: PCL now records the pathname of a file
2244 in which methods and the like are defined, rather than its
2246 * minor incompatible change: TRUENAME now considers the truename of
2247 a file naming a directory to be the pathname with :DIRECTORY
2248 component indicating that directory.
2249 * minor incompatible change: a NAMED clause in the extended form of
2250 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2251 reason for the previous behaviour is unclear.
2252 * more systematization and improvement of CLOS and MOP conformance
2253 in PCL (thanks to Gerd Moellman and Pierre Mai):
2254 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2256 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2257 primary methods with no specializers;
2258 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2260 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2261 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2262 FORWARD-REFERENCED-CLASSes; error reporting on
2263 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2265 ** SXHASH on CLOS instances now uses a slot internal to the
2266 instance to return different numbers on distinct instances,
2267 while preserving the same return value through invocations of
2269 ** DEFMETHOD signals errors when methods with longer incongruent
2270 lambda lists are added to generic functions;
2271 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2272 CLASS, as specified in AMOP;
2273 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2274 on the order returned by the primary method for classes of
2275 class STANDARD-CLASS;
2276 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2277 * fixed some bugs shown by Paul Dietz' test suite:
2278 ** DOLIST puts its body in TAGBODY;
2279 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2281 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2282 value producing form;
2283 ** if more variables are given to PROGV than values, extra
2284 variables are bound and made to have no value;
2285 ** NSUBSTITUTE on list arguments gets the right answer with
2287 ** ELT signals an error of type TYPE-ERROR when the index argument
2288 is not a valid sequence index;
2289 ** LOOP signals (at macroexpansion time) an error of type
2290 PROGRAM-ERROR when duplicate variable names are found;
2291 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2292 ** FUNCALL of special-operators now cause an error of type
2294 ** PSETQ now works as required in the presence of side-effecting
2295 symbol-macro places;
2296 ** NCONC accepts any object as its last argument;
2297 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2299 ** loop-for-as-package does not require a package to be explicitely
2301 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2303 * fixed bug 166: compiler preserves "there is a way to go"
2304 invariant when deleting code.
2305 * fixed bug 172: macro lambda lists with required arguments after
2306 &REST arguments now cause an error to be signalled. (thanks to
2308 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2309 bug. (thanks to Matthew Danish)
2310 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2312 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2313 arguments are lexically rebound. (thanks to Gerd Moellmann and
2315 * fixed bug 194: error messages are now more informative when there
2316 is no primary method applicable in a call to a generic
2317 function. (thanks to Gerd Moellmann)
2318 * fixed bug in command line argument checking (thanks to Julian
2320 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2321 (thanks to Matthew Danish)
2322 * incremented fasl file version number, because of the
2323 SXHASH-related changes in the layout of CLOS data structures
2325 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2326 * minor incompatible change: The runtime (the Unix executable named
2327 "sbcl") is now much pickier about the .core files it will load.
2328 Essentially it now requires .core files to descend from the same
2329 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2330 as the runtime does. (The intent is to prevent the crashes which
2331 can occur, and which can even be reported as mysterious failures,
2332 when people patch the sources or change the build parameters
2333 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2334 sbcl and .core files.)
2335 * fixed bug: VALUES-LIST is no longer optimized away.
2336 * fixed bug 142: The FFI conversion of C string values to Lisp
2337 string values no longer conses excessively. (thanks to Nathan
2338 Froyd porting Raymond Toy's fix to CMU CL)
2339 * began to systematize and improve MOP conformance in PCL (thanks to
2340 Nathan Froyd, Gerd Moellman and Pierre Mai):
2341 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2343 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2344 ** FINALIZE-INHERITANCE is now called on class finalization;
2345 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2346 argument precedence order.
2347 * fixed bug 202: The compiler no longer fails on functions whose
2348 derived types contradict their declared type.
2349 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2350 so it can be non-toplevel.
2351 * The fasl file version number has changed (because of the new
2352 implementation of DEFMACRO).
2353 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2354 safe code, that any length requirement by their type-specifier
2355 argument is valid. The exceptions to this are described in bug
2357 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2358 functions now check the types of their inputs as required by ANSI.
2359 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2360 introduced symbol is DECLAREd to be SPECIAL.
2361 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2362 * fixed bug: :COUNT argument to sequence functions may be negative
2363 * fixed bug: body of DO-SYMBOLS may contain declarations
2364 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2365 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2366 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2367 symbol macro only once
2368 * fixed printing of call frame when argument list is unavailable
2369 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2370 * compiler no longer signals WARNING on unknown keyword
2373 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2374 * A beta-quality port to the mips architecture running Linux,
2375 based on the old CMUCL backend, has been made. It has been tested
2376 on a big-endian kernel, and works sufficiently well to be able to
2377 rebuild itself; it has not been tested in little-endian mode.
2378 * fixed an inconsistency between gencgc.c and purify.c which made
2379 dumping/loading .core files unreliable
2380 * fixed bug 120a: The compiler now deals correctly with IFs where
2381 the consequent is the same as the alternative, instead of
2382 misderiving the return type. (thanks to Alexey Dejneka)
2383 * fixed bug 113: Logical pathnames are now dumpable (the logical
2384 host is resolved at load-time, throwing an error if it is not
2386 * fixed bug 174: FORMAT's error message is slightly clearer when a
2387 non-printing character is used in a format directive.
2388 * fixed several bugs in compiler checking of type declarations, i.e.
2389 violations of the Python "declarations are assertions" principle
2390 (thanks to Alexey Dejneka)
2391 * fixed several bugs in PCL's error checking (thanks to Gerd
2393 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2395 * fixed bug in compilation of functions as first class values
2396 (thanks to Antonio Martinez-Shotton)
2397 * The compiler's handling TYPE-ERRORs which it can prove will
2398 inevitably happen at runtime has been cleaned up and corrected
2399 in several ways. (thanks to Alexey Dejneka)
2400 * improved argument type checking for various basic arithmetic
2401 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2402 lavished on them in the past that they can be compiled in many
2403 ways in different special cases
2404 * fixed bug 181: compiler checks validity of user supplied type
2406 * cleaned up code flushing in optimization: Function calls which
2407 should signal errors for safety purposes (e.g. which ANSI says
2408 should signal errors when their arguments are of incorrect type)
2409 are no longer optimized away.
2410 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2411 * incremented fasl file version number, because changes in the
2412 implementation of sequence functions like COERCE caused
2413 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2414 in old inline expansions) to become undefined. (Actually these
2415 changes were later undone, so we might very well be binary
2416 compatible with 0.7.7 after all, but leaving the version number
2417 incremented seemed like the simplest and most conservative
2420 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2421 * An alpha-quality port to the parisc architecture running Linux,
2422 based on the old CMUCL backend, has been made. This, even more so
2423 than the other backends, should be considered still a work in
2424 progress; known problems include that the Linux kernel in 64-bit
2425 mode does not propagate the correct sigcontext structure to
2426 userspace, and consequently SBCL on a parisc64 kernel will not
2428 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2429 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2430 suppressed.) Also now that the compiler is looking at declarations
2431 in the environment, it checks optimization declarations as well,
2432 and suppresses inlining when (> DEBUG SPEED).
2433 * More fixes have been made to treatment of floating point exception
2434 treatment and other Unix signals. In particular, floating point
2435 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2436 * The detection and handling of control stack exhaustion (infinite
2437 or very deeply nested recursion) has changed. Stack exhaustion
2438 detection is now done by write-protecting pages at the OS level
2439 and applies at all optimization settings; when found, a
2440 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2441 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2442 be caught using IGNORE-ERRORS.
2443 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2444 correctly when transforming recursive LABELS functions to LETs.
2445 (thanks to Alexey Dejneka)
2446 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2447 that are names of constants or global variables.
2448 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2449 alien routines with docstrings.
2450 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2451 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2453 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2454 (thanks to Eric Marsden)
2455 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2456 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2457 * Bug fix: The SPARC backend can now compile functions involving
2458 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2459 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2460 to a functions expecting a PATHNAME-DESIGNATOR.
2461 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2462 lambda lists. (thanks to Alexey Dejneka)
2463 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2464 functions so that EXPORTing the name of the class doesn't cause
2465 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2466 to Antonio Martinez for reporting this)
2467 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2468 constant-folded without failing an assertion. (thanks to Einar
2469 Floystad Dorum for reporting this)
2470 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2471 types are now dealt with more correctly by the compiler.
2472 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2473 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2474 behaviour analogous to RENAME-FILE. This puts its behaviour more
2475 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2477 * The fasl file version number has changed again. (because of the
2478 bug fix involving the names of PCL MAKE-INSTANCE functions)
2480 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2481 * bug fix: Floating point exceptions are treated much more
2482 consistently on the x86/Linux and PPC/Linux platforms.
2483 * Array initialization with :INITIAL-ELEMENT is now much faster for
2484 cases when the compiler cannot open code the array creation, but
2485 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2486 array accesses have also seen a speed increase.
2487 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2488 specified. (This was at the root of some bad interactions between
2489 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2491 * bug fix: Internal error arguments for undefined functions are now
2492 computed correctly on the PPC/Linux platform.
2493 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2494 Raymond Toy's patch for CMU CL)
2495 * Support for the Solaris 9 operating environment has been included
2496 (thanks to Daniel Merritt)
2497 * A very ugly but hopefully complete draft of the missing FFI chapter
2498 of the manual has been created by reformatting the corresponding
2499 CMU CL manual chapter into (currently very ugly and incoherent)
2500 DocBook and bringing it up to date for SBCL behavior. Thus, the
2501 manual is now essentially complete, at least by my extreme
2502 once-and-only-once standards, whereby it's acceptable to refer to
2503 the doc strings of SB-EXT functions as the primary documentation.
2504 * The fasl file version number has changed again, due to cleanup of
2505 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2506 is no longer a static symbol.)
2508 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2509 * SBCL now builds with OpenMCL (version 0.12) as the
2510 cross-compilation host; also, more progress has been made toward
2511 bootstrapping under CLISP.
2512 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2514 * bug 158 fixed: The compiler can now deal with integer loop
2515 increments different from 1; fixing this turned out also to fix
2517 * bug 169 fixed: no more bogus warnings about using lexical bindings
2518 despite the presence of perfectly good SPECIAL declarations (thanks
2519 to David Lichteblau)
2520 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2521 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2522 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2524 * bug 180 fixed: Method combination specifications no longer ignore
2525 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2526 * bug fix: Structure type predicate functions now check their argument
2527 count as they should.
2528 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2529 correctly. (thanks to Pierre Mai)
2530 * minor incompatible change: The --noprogrammer option is deprecated
2531 in favor of the new --disable-debugger option, which is very similar.
2532 (The major difference is that it takes effect at a slightly different
2533 time at startup, causing handling of errors in --sysinit and
2534 --userinit files will be affected differently.) The
2535 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2536 been added to allow this functionality to be controlled from ordinary
2537 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2538 maintainers, who might want to run non-interactive scripts to
2539 build SBCL cores which will later be used interactively.)
2540 * minor incompatible change: The LOAD function no longer, when given
2541 a wild pathname to load, loads all files matching that pathname.
2542 Instead, an error of type FILE-ERROR is signalled.
2544 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2545 * bug 147 fixed: The compiler preserves its block link/count
2546 invariants more correctly now so that it doesn't crash. (thanks
2548 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2550 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2551 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2552 no-value-for-this-character as NIL (as specified by ANSI).
2553 * HOST-NAMESTRING on physical pathnames now returns a string that is
2554 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2555 (thanks to Christophe Rhodes)
2556 * The Alpha port handles icache flushing more correctly. (thanks to
2558 * More progress has been made toward bootstrapping under CLISP. (thanks
2559 to Christophe Rhodes)
2560 * The fasl file format has changed again, because dynamic loading
2561 on OpenBSD (which has non-ELF object files) motivated some cleanups
2562 in the way that foreign symbols are transformed and passed around.
2563 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2564 is no longer treated as whitespace by the reader, but instead as
2565 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2566 |A
\7fB|, instead of A as it used to.
2568 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2569 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2570 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2571 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2572 this as of 0.7.1.45, but was left out of the previous news section
2573 (thanks to Dan Barlow)
2574 * SBCL now runs on the Solaris operating system on SPARC architectures
2575 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2576 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2577 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2578 * SPARC backend cleanups, allowing builds of cores optimized for V8
2579 and V9 SPARCS, and also emission of code targeted to a particular
2580 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2582 * SBCL is closer to bootstrapping under CLISP, thanks to various
2583 fixes by Christophe Rhodes.
2584 * The fasl file format has changed again, to allow the compiler's
2585 INFO database to support symbol macros.
2586 * The user manual (in doc/) is formatted into HTML more nicely.
2587 (thanks to coreythomas)
2588 * The system is smarter about SUBTYPEP relationships, especially
2589 those involving NOT types (including types such as ATOM which are
2590 represented internally using NOT types). Thus SUBTYPEP is less
2591 likely to return (VALUES NIL NIL) in general, and in particular
2592 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2594 * The fasl file format has changed again, because the internal
2595 representation of types now includes a new slot to support the new
2596 SUBTYPEP-of-NOT-types logic.
2597 * (not a change in the main branch of SBCL, but a related prototype
2598 which can hopefully be merged into the main branch of SBCL in the
2599 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2600 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2601 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2602 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2603 dispatch macro character combinations. (thanks to Alexey Dejneka)
2604 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2605 that superficially look like logical namestrings correctly.
2606 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2608 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2609 default supplied-p) for &optional and &key arguments. (thanks to
2612 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2613 * incompatible change: The compiler is now less aggressive about
2614 tail call optimization, doing it only when (> SPACE DEBUG) or
2615 (> SPEED DEBUG). (This is an incompatible change because there are
2616 programs which relied on the old CMU-CL-style behavior to optimize
2617 away their unbounded recursion which will now die of stack overflow.)
2618 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2619 for non-GENCGC systems has been increased to 20M (since that
2620 seems much closer to the likely performance optimum for modern
2621 systems than the old 4M value was)
2622 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2623 and *DEBUG-PRINT-LEVEL*
2624 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2625 of CMU CL's support for SPARC, and various endianness and other
2626 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2627 * new syntactic sugar for the Unix command line: --load foo.bar is now
2628 an alternate notation for --eval '(load "foo.bar")'.
2630 ** The system now detects stack overflow and handles it gracefully,
2631 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2632 optimization settings. (This is a good thing in general, and
2633 its introduction in this version should be particularly timely
2634 for anyone whose code fails because of suppression of tail
2636 ** The system now hunts for the C variable "environ" in a more
2637 devious way, to avoid segfaults when the C library version differs
2638 between compile time and run time. (thanks to Christophe Rhodes)
2639 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2640 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2641 ** The compiler no longer issues bogus style warnings for undefined
2642 classes in the same source file as the DEFCLASSes which defined
2643 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2645 ** fixes in CONDITION class precedence list for undefined function
2646 errors (thanks to Alexei Dejneka)
2647 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2648 correctly. (thanks to Dan Barlow)
2649 ** portability fixes aiming at bootstrapping under CLISP (thanks
2650 to Dave McDonald and Christophe Rhodes)
2651 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2652 ** fixes in type translation and and type inference (thanks to
2654 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2655 ** various fixes in the new ports (thanks to Dan Barlow)
2656 * several changes related to debugging:
2657 ** suppression of tail recursion, as noted above
2658 ** stack overflow detection, as noted above
2659 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2660 is now the default. (For some time encapsulation has been more
2661 reliable than the breakpoint-based :ENCAPSULATE NIL
2662 implementation, at least on X86 systems; and I just noticed that
2663 encapsulation also seems closer to the spirit of the ANSI
2666 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2668 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2669 up properly again. (There was a packaging bug in 0.7.0 which
2670 left their definitions in SB-SYS::LOAD-FOREIGN and
2671 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2672 vital for most things which interface to C-level interfaces,
2673 like extensions working with sockets or databases or
2674 Perl-compatible regexes or whatever, and the need to fix
2675 this bug is the main reason that 0.7.1 was released so
2677 ** DEFGENERIC is now choosier about the methods it redefines, so that
2678 reLOADing a previously-LOADed file containing DEFGENERICs does
2679 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2680 cycle now works as it should. (thanks to Alexey Dejneka)
2681 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2682 fixed by Christophe Rhodes. (He actually submitted this patch
2683 months ago, and I delayed until after 0.7.0.)
2684 ** Bug 111 (internal compiler confusion about runtime checks on
2685 FUNCTION types) was fixed by Alexey Dejneka.
2686 * Some internal cleanups (getting rid of variables which aren't
2687 needed now that the byte interpreter is gone) caused the fasl
2688 file format number to change again.
2690 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2691 * major incompatible change: The default fasl file extension, i.e. the
2692 default extension for files produced by COMPILE-FILE, has changed
2693 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2695 ** There are many changes in the implementation of the compiler.
2696 SBCL is now essentially a compiler-only implementation of ANSI
2697 Common Lisp. EVAL still "interprets" a few special cases, but
2698 almost all the interesting cases are handled by creating
2699 a LAMBDA expression, calling COMPILE on it, then calling
2700 FUNCALL on the result.
2701 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2702 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2703 Since the code is newer, there might still be some new bugs
2704 (though not as many as before Martin Atzmueller's fixes:-). But
2705 the new code is substantially simpler and clearer, and hopefully
2706 any remaining bugs will be simpler, less fundamental, and more
2707 fixable then the bugs in the old code.
2708 ** The revised compiler is still a little unsteady on its feet.
2710 *** The debugging information it produces (particularly the names
2711 of FUNCTION objects) is sometimes much less useful than what
2712 the old compiler produced.
2713 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2714 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2715 within a MACROLET) has been temporarily weakened.
2716 ** There are new compiler optimizations for various functions:
2717 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2718 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2719 *** the math functions TRUNCATE, FLOOR, and CEILING
2720 *** the function-of-all-trades COERCE
2721 Mostly these should be transparent, but there's one
2722 potentially-annoying problem (bug 117): when the compiler
2723 inline-expands a function and does type analysis on the result,
2724 it can create control paths which have type mismatches, and
2725 when it can't prove that those control paths aren't taken,
2726 it will issue WARNINGs about the type mismatches. This is
2727 a particular problem in practice for the new sequence functions.
2728 It's not clear how this should be fixed, and for now, a
2729 workaround is given in the entry for 117 in the BUGS file.
2730 ** (Because of the interaction between the two previous items --
2731 occasional inlining problems and new inline expansions -- some
2732 of the new sequence function optimizations won't really kick in
2733 completely until debugging information, and then inlining, are
2734 straightened out in some future version.)
2735 * minor incompatible changes:
2736 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2737 DIRECTORY no longer implicitly promotes NIL slots of its
2738 pathname argument to :WILD. In particular, when you ask for the
2739 contents of a directory (which you used to be able to do without
2740 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2741 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2742 ** changes in behavior that ANSI explicitly defines to be
2743 implementation dependent:
2744 *** The new compiler-only implementation still conforms with ANSI,
2745 but acts a little different than before. Besides the obvious
2746 changes in performance tradeoffs (that the cost per form passed
2747 to EVAL has gone up, and the cost per form executed by EVAL
2748 has gone down), the behavior of the system changes a little
2749 because there are no longer any interpreted function objects.
2750 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2751 e.g. doing COMPILE on the output of interactive DEFUN is
2753 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2755 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2756 has changed from (:CL) to NIL.
2757 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2758 e.g. #\Newline and #\Tab, don't have names specified in the
2759 ANSI Common Lisp standard, is now based on their ASCII symbolic
2760 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2761 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2762 are no longer used for output.
2763 ** changes in internal implementation constants:
2764 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2765 4 million. (If your application spends a lot of time GCing and
2766 you have a lot of RAM, you might want to experiment with
2767 increasing it even more.)
2768 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2769 However, almost all old code should still continue to work without
2770 immediate update, as SB-C-CALL is now a (deprecated) nickname
2772 ** Old operator names in the style DEF-FOO are now deprecated in
2773 favor of new corresponding names DEFINE-FOO, for consistency with
2774 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2775 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2776 internal symbols, but a few supported extensions like
2777 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2778 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2779 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2780 etc. as you get deeper into recursive calls to the debugger
2781 command loop, instead of the old "5]", "5]]", "5]]]"
2782 sequence. (I was motivated to do this when squabbles between
2783 ILISP and SBCL left me very deeply nested in the debugger. In the
2784 short term, this change will probably provoke more ILISP/SBCL
2785 squabbles, but hopefully it will be an improvement in the long run.)
2786 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2787 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2788 function, since declaiming return types involving aliens is
2789 (1) annoyingly messy to do by hand and (2) vital to efficient
2790 compilation of code which calls such functions.
2791 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2792 longer reexported by the SB-EXT package. They're solely useful
2793 for alien code, so it seems more logical that you should get
2794 them from the SB-ALIEN package, not in SB-EXT.
2795 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2796 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2797 features. Instead, the code that they used to control is always
2798 built into the system.
2799 * many other bug fixes
2800 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2801 advantage of the new EVAL-WHEN stuff and to clean them up in
2802 general, and they are now more ANSI-compliant in a number of
2803 ways. Martin Atzmueller is responsible for a lot of this.
2804 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2806 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2807 *** correct ERROR type for various file operations
2808 *** some fixes for Lisp streams
2809 *** DEFMETHOD syntax checking
2810 *** changing old weird representation of debug information as
2811 strings (which, among their other deficiencies, don't transform
2812 correctly when you rename packages, and don't change their
2813 print representation when you change things like *PACKAGE*
2814 and *PRINT-LENGTH*) to symbols and lists of symbols
2815 He also made several improvements and fixed several bugs in DESCRIBE.
2816 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2818 *** misbehavior of WRITE-STRING/WRITE-LINE
2819 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2820 and several other LOOP problems as well
2821 *** DIRECTORY when similar filenames are present
2822 *** DEFGENERIC with :METHOD options
2823 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2824 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2825 *** argument ordering in FIND with :TEST option
2826 *** mishandled package designator argument in APROPOS-LIST
2827 *** various problems in the backquote readmacro
2828 *** a bug in APROPOS
2829 *** probably some others that I'm not describing very well here,
2830 since the CVS log documents them by reference to sbcl-devel
2831 messages, and the SourceForge archives aren't working well.:-(
2832 ** Dan Barlow improved the Alpha port (and is making progress on the
2833 PPC port, for those of you who think different).
2834 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2835 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2836 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2837 continues to work on the SPARC port (for those of you in a position
2838 to look down upon our little PC-compatible boxes from a great height).
2839 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2840 and uses that copy, rather than the current dynamic value, when
2841 it's trying to decide whether to truncate output. Thus e.g.
2842 (let ((*print-lines* 50))
2843 (pprint-logical-block (stream nil)
2845 (let ((*print-lines* 8))
2846 (print (aref possiblybigthings i) stream)))))
2847 should now truncate the logical block only at 50 lines, instead of
2848 often truncating it at 8 lines, as it did before.
2849 * The doc/cmucl/ directory, containing old CMU CL documentation
2850 from the time of the fork, is no longer part of the base system.
2851 SourceForge has shut down its anonymous FTP service, and with it
2852 my original plan for distributing the old CMU CL documentation
2853 there. For now, if you need these files you can download an old
2854 SBCL source release and extract them from it.
2855 * The fasl file version number changed again, for dozens of reasons,
2856 some of which are apparent above.
2858 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2859 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2860 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2861 patch, so that DISASSEMBLE output is much nicer.
2862 * The code in the SB-PROFILE package now seems reasonably stable.
2863 I still haven't decided what the final interface should look like
2864 (I'd like PROFILE to interact cleanly with TRACE, since both
2865 facilities use function encapsulation) but if you have a need
2866 for profiling now, you can probably use it successfully with
2867 the current CMU-CL-style interface.
2868 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2869 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2870 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2871 link now returns the dangling link itself, and for similar
2872 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2873 link itself. (In these cases the old code signalled an error and
2874 looped endlessly, respectively.) Thus, DIRECTORY now works even
2875 in the presence of dangling and cyclic symbolic links.
2876 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2877 is now a supported extension again, since the consensus on
2878 sbcl-devel was that it can be useful for ordinary development
2879 work, not just for debugging SBCL itself.
2880 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2881 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2882 that currently-defined functions might be redefined later with
2883 different return types.
2884 * Hash tables can be printed readably, as inspired by CMU CL code
2885 of Eric Marsden and SBCL code of Martin Atzmueller.
2886 * better error handling in CLOS method combination, thanks to
2887 Martin Atzmueller porting Pierre Mai's CMU CL patches
2888 * more overflow fixes for >16Mbyte I/O buffers
2889 * A bug in READ has been fixed, so that now a single Ctrl-D
2890 character suffices to cause end-of-file on character streams.
2891 In particular, now you only need one Ctrl-D at the command
2892 line (not two) to exit SBCL.
2893 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2895 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2896 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2897 does the right thing.
2898 * The implementation of some type tests, especially for CONDITION
2899 types, is now tidier and maybe faster, due to CMU CL code
2900 originally by Douglas Crosher, ported by Martin Atzmueller.
2901 * Some math functions have been fixed, and there are new
2902 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2903 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2904 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2905 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2906 likely to be built into sbcl-0.7.0.)
2907 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2908 CL, ported to SBCL by Martin Atzmueller)
2909 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2910 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2911 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2912 * There's a new slam.sh hack to shorten the edit/compile/debug
2913 cycle for low-level changes to SBCL itself, and a new
2914 :SB-AFTER-XC-CORE target feature to control the generation of
2915 the after-xc.core file needed by slam.sh.
2916 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2917 COMPILE-FILE is no longer supported, so that now every function
2918 gets an entry point, so that block compilation looks a little
2919 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2920 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2921 returns the number of bytes consed since the system started,
2922 rather than the number consed since the first time the function
2923 was called. (The new definition parallels ANSI functions like
2924 CL:GET-INTERNAL-RUN-TIME.)
2925 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2926 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2927 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2928 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2929 * Fasl file version numbers are now independent of the target CPU,
2930 since historically most system changes which required version
2931 number changes have affected all CPUs equally. Similarly,
2932 the byte fasl file version is now equal to the ordinary
2935 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2936 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2937 is no longer recognized. I apologize for this, because it was
2938 listed in SB-EXT as a supported extension, but I found that
2939 its existing behavior was poorly specified, as well as incorrectly
2940 specified, and it looked like too much of a mess to straighten it
2941 out. I have enough on my hands trying to get ANSI stuff to work..
2942 * many patches ported from CMU CL by Martin Atzmueller, with
2943 half a dozen bug fixes in pretty-printing and the debugger, and
2944 half a dozen others elsewhere
2945 * fixed bug 13: Floating point infinities are now supported again.
2946 They might still be a little bit flaky, but thanks to bug reports
2947 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2948 as flaky as they were.
2949 * The --noprogrammer command line option is now supported. (Its
2950 behavior is slightly different in detail from what the old man
2951 page claimed it would do, but it's still appropriate under the
2952 same circumstances that the man page talks about.)
2953 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2954 are now supported, and enabled by default. Thus, the compiler can
2955 handle many floating point and complex operations much less
2956 inefficiently. (Thus e.g. you can implement a complex FFT
2958 * The compiler now detects type mismatches between DECLAIM FTYPE
2959 and DEFUN better, and implements CHECK-TYPE more correctly, and
2960 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2961 thanks to patches from Martin Atzmueller.
2962 * various fixes to make the cross-compiler more portable to
2963 ANSI-conforming-but-different cross-compilation hosts (notably
2964 Lispworks for Windows, following bug reports from Arthur Lemmens)
2965 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2966 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2967 fixed thanks to Pierre Mai's CMU CL patches.
2968 * fixes to keep the system from overflowing internal counters when
2969 it tries to use i/o buffers larger than 16M bytes
2970 * fixed bug 45a: Various internal functions required to support
2971 complex special functions have been merged from CMU CL sources.
2972 (When I was first setting up SBCL, I misunderstood a compile-time
2973 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2974 * improved support for type intersection and union, fixing bug 12
2975 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2976 more obscure bugs as well
2977 * some steps toward byte-compiling non-performance-critical
2978 parts of the system, courtesy of patches from Martin Atzmueller
2979 * Christophe Rhodes has made some debian packages of sbcl at
2980 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2981 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2982 stable, but are nonetheless usable. When he's ready, I'd be happy
2983 to add them to the SourceForge "File Releases" section. (And if
2984 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2985 * new fasl file format version number (because of changes in
2986 internal representation of (OR ..) types to accommodate the new
2987 support for (AND ..) types, among other things)
2989 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2990 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2992 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2994 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2995 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2996 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2997 * DESCRIBE now works on structure objects again.
2998 * Most function call argument type mismatches are now handled as
2999 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3000 know whether the function will be redefined before the call is
3001 executed. (The compiler could flag local calls with full WARNINGs,
3002 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3003 it doesn't keep track of enough information to know whether calls
3004 are local in this sense.)
3005 * Compiler output is now more verbose, with messages truncated
3006 later than before. (There should be some supported way for users
3007 to override the default verbosity, but I haven't decided how to
3008 provide it yet, so this behavior is still controlled by the internal
3009 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3010 src/compiler/ir1util.lisp.)
3011 * Fasl file format version numbers have increased again, because
3012 support for the Gray streams extension changes the layout of the
3013 system's STREAM objects.
3014 * The Gray subclassable streams extension now works, thanks to a
3015 patch from Martin Atzmueller.
3016 * The full LOAD-FOREIGN extension (not just the primitive
3017 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3018 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3019 but like most other programs, it defaults to copying the Unix
3020 environment from the original process instead of starting the
3021 new process in an empty environment.
3022 * Extensions which manipulate the Unix environment now support
3023 an :ENVIRONMENT keyword option which doesn't smash case or
3024 do other bad things. The CMU-CL-style :ENV option is retained
3025 for porting convenience.
3026 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3027 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3029 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3031 * A patch from Martin Atzmueller seems to have solved the SIGINT
3032 problem, and as far as we know, signal-handling now works cleanly.
3033 (If you find any new bugs, please report them!)
3034 * The system no longer defaults Lisp source file names to types
3035 ".l", ".cl", or ".lsp", but only to ".lisp".
3036 * The compiler no longer uses special default file extensions for
3037 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3038 seems to expect a single default extension for all compiled code,
3039 and there's no compelling reason to try to stretch the standard
3040 to allow two different extensions.) Instead, byte-compiled files
3041 default to the same extension as native-compiled files.
3042 * Fasl file format version numbers have increased again, because
3043 a rearrangement of internal implementation packages made some
3044 dumped symbols in old fasl files unreadable in new cores.
3045 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3046 many fewer weird special cases.
3047 * Bug #17 (differing COMPILE-FILE behavior between logical and
3048 physical pathnames) has been fixed, and some related misbehavior too,
3049 thanks to a patch from Martin Atzmueller.
3050 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3051 by Tim Moore, ported to SBCL by Martin Atzmueller.
3052 * Martin Atzmueller fixed several filesystem-related problems,
3053 including bug #36, in part by porting CMU CL patches, which were
3054 written in part by Paul Werkowski.
3055 * More compiler warnings in src/runtime/ are gone, thanks to
3056 more patches from Martin Atzmueller.
3057 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3060 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3062 * DESCRIBE now works on CONDITION objects.
3063 * The debugger now handles errors which arise when trying to print
3064 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3066 * The build system now uses an additional file, customize-target-features.lisp,
3067 to allow local modifications to the target *FEATURES* list. (The point of
3068 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3069 debugging features enabled, without having to worry about propagating it
3070 into everyone's system when I do a "cvs update".) When no
3071 customize-target-features.lisp file exists, the target *FEATURES* list
3072 should be constructed the same way as before.
3073 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3074 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3075 warn about behavior which is undefined under the ANSI spec. Note: This
3076 is specified by ANSI, but it's not very popular with programmers.
3077 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3078 macro in the SBCL sources for an example of a workaround which you
3079 might use to make portable ANSI-standard code which does what you want.
3080 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3081 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3082 when it's used as a non-toplevel form. (This is required in order
3083 to implement the DEFCONSTANT-EQX macro.)
3084 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3085 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3086 funny things, due to bugs in the implementation of EVAL-WHEN
3087 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3088 non-toplevel DEFCONSTANTs are uncommon.)
3089 * The core file version number and fasl file version number have been
3090 incremented, because the old noncompliant DEFCONSTANT behavior involved
3091 calling functions which no longer exist, and because I also took the
3092 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3093 * fixed bug 1 (error handling before read-eval-print loop starts), and
3094 redid debugger restarts and related debugger commands somewhat while
3096 ** The QUIT debugger command is gone, since it did something
3097 rather different than the SB-EXT:QUIT command, and since it never
3098 worked properly outside the main toplevel read/eval/print loop.
3099 Invoking the new TOPLEVEL restart provides the same functionality.
3100 ** The GO debugger command is also gone, since you can just invoke
3101 the CONTINUE restart directly instead.
3102 ** The TOP debugger command is also gone, since it's redundant with the
3103 FRAME 0 command, and since it interfered with abbreviations for the
3105 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3107 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3108 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3109 some porting work by Martin Atzmueller)
3110 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3111 with a boatload of patches which he ported to SBCL. Now that those
3113 ** The system tries to make sure that its low-priority messages
3114 are prefixed by semicolons, to help people who like to use
3115 syntax highlighting in their ILISP buffer. (This patch
3116 was originally due to Raymond Toy.)
3117 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3118 patches originally written by Raymond Toy.
3119 ** The compiler understands coercion between single-value and
3120 multiple-VALUES type expressions better, getting rid of some very
3121 weird behavior, thanks to patches originally by Robert MacLachlan
3122 and Douglas Crosher.
3123 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3124 lambda lists, thanks to a patch originally by Pierre Mai.
3125 ** The system no longer bogusly warns about "abbreviated type
3127 ** The compiler gets less confused by inlining and RETURN-FROM,
3128 thanks to some patches originally by Tim Moore.
3129 ** The system no longer hangs when dumping circular lists to fasl
3130 files, thanks to a patch originally from Douglas Crosher.
3131 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3132 undefined function error.
3133 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3134 in structure slot initforms) for now, documented workaround instead:-|
3135 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3136 * fixed bug 5: added stubs for various Gray stream functions called
3137 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3138 installed, at least appropriate type errors are generated
3139 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3140 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3141 consistently in DEFMETHOD forms.
3142 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3143 it doesn't seem to affect SBCL after all
3144 * The C runtime system now builds with better optimization and many
3145 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3147 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3149 * The system is now under CVS at SourceForge (instead of the
3150 CVS repository on my home machine).
3151 * The new signal handling code has been tweaked to treat register
3152 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3153 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3154 downstream, e.g. in debug-int.lisp, has implicit dependencies
3155 on the unsignedness of integer representation of machine words,
3156 and that caused the system to bomb out with infinite regress
3157 when trying to recover from type errors involving signed values,
3158 e.g. (BUTLAST '(1 2 3) -1).)
3159 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3160 (This was one of the bugs Peter Van Eynde reported back in July.)
3161 * The system now uses code inspired by Colin Walters' O(N)
3162 implementation of MAP (from the cmucl-imp@cons.org mailing
3163 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3164 inline the MAP operation, and there is more than one
3165 sequence argument to the MAP call (so that it can't just
3166 do ETYPECASE once and for all based on the type of the
3167 single sequence argument). (The old non-inline implementation
3168 of the general M-argument sequence-of-length-N case required
3169 O(M*N*N) time when any of the sequence arguments were LISTs.)
3170 * The QUIT :UNIX-CODE keyword argument has been renamed to
3171 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3172 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3174 * Raymond Wiker's patches to port dynamic loading from Linux to
3175 FreeBSD have been added.
3176 * The BUGS file is now more nearly up to date, thanks in large part
3177 to Martin Atzmueller's review of it.
3178 * The debugger now flushes standard output streams before it begins
3179 its output ("debugger invoked" and so forth).
3180 * The core version number and fasl file version number have both
3181 been incremented, because of incompatible changes in the layout
3183 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3184 used to not be called for a saved Lisp image.)
3185 * Martin Atzmueller's version of a patch to fix a compiler crash,
3186 as posted on sbcl-devel 13 September 2000, has been installed.
3187 * Instead of installing Martin Atzmueller's patch for the
3188 compiler transform for SUBSEQ, I deleted the compiler transform,
3189 and transforms for some similar consing operations.
3190 * A bug in signal handling which kept TRACE from working on OpenBSD
3192 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3193 away by constant folding
3194 * The system now defines its address space constants in one place
3195 (in the Lisp sources), and propagates them automatically elsewhere
3196 (through GENESIS and the sbcl.h file). Therefore, patching the
3197 address map is less unnecessarily tedious and error-prone. The
3198 Lisp names of address space constants have also been systematized.
3199 * CVS tags like dollar-Header-dollar have been removed from
3200 the sources, because they have never saved me trouble and
3201 they've been source of trouble working with patches and other
3202 diff-related operations.
3203 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3204 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3206 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3208 * The system has been ported to OpenBSD.
3209 * The system now compiles with a simple "sh make.sh" on the systems
3210 that it's supported on. I.e., now you no longer need to tweak
3211 text in the target-features.lisp-expr and symlinks in src/runtime/
3212 by hand, the make.sh takes care of it for you.
3213 * The system is no longer so grossly inefficient when compiling code
3214 involving vectors implemented as general (not simple) vectors (VECTOR T),
3215 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3216 which dares to use the various sequence functions on non-simple
3217 vectors, takes less of a performance hit.
3218 * There is now a primitive type predicate VECTOR-T-P
3219 to test for the (VECTOR T) type, so that e.g.
3220 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3221 can now be compiled with some semblance of efficiency. (The old code
3222 turned the type declaration into a full call to %TYPEP at runtime!)
3223 * AREF on (VECTOR T) is still not fast, since it's still compiled
3224 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3225 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3226 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3227 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3228 without making full calls to SUBTYPEP at runtime.
3229 (Some analogous efficiency issues for non-simple vectors specialized to
3230 element types other than T, or for non-simple multidimensional arrays,
3231 have not been addressed. They could almost certainly be handled the
3232 same way if anyone is motivated to do so.)
3233 * The changes in array handling break binary compatibility, so
3234 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3235 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3236 instead of (VALUES T T).
3237 * By following the instructions that Dan Barlow posted to sbcl-devel
3238 on 2 July 2000, I was able to enable primitive dynamic object
3239 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3240 functionality is not implemented (since it calls ld to resolve
3241 library references automatically, requiring RUN-PROGRAM for its
3242 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3243 try to resolve library references) is now supported.
3244 * The system now flushes the standard output streams when it terminates,
3245 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3246 them at several other probably-convenient times, e.g. in each pass of
3247 the toplevel read-eval-print loop, and after evaluating a form given
3248 as an "--eval" command-line option. (These changes were motivated by a
3249 discussion of stream flushing issues on cmucl-imp in August 2000.)
3250 * The source transform for TYPEP of array types no longer assumes
3251 that an array whose element type is a not-yet-defined type
3252 is implemented as an array of T, but instead punts, so that the
3253 type will be interpreted at runtime.
3254 * There is now some support for cross-compiling in make.sh: each of
3255 the phases of make.sh has its own script. (This should be transparent
3256 to people doing ordinary, non-cross-compile builds.)
3257 * Since my laptop doesn't have hundreds of megabytes of memory like
3258 my desktop machine, I became more motivated to do some items on
3259 my to-do list in order to reduce the size of the system a little:
3260 ** Arrange for various needed-only-at-cold-init things to be
3261 uninterned after cold init. To support this, those things have
3262 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3263 symbols with such names are now uninterned after cold init).
3264 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3265 fasl files for cold load.
3266 ** Remove the old compiler structure pooling code (which used to
3267 be conditional on the target feature :SB-ALLOC) completely.
3268 ** Redo the representation of some data in cold init to be more compact.
3269 (I also looked into supporting byte compiled code at bootstrap time,
3270 which would probably reduce the size of the system a lot, but that
3271 looked too complicated, so I punted for now.)
3272 * The maximum signal nesting depth in the src/runtime/ support code has
3273 been reduced from 4096 to 256. (I don't know any reason for the very
3274 large old value. If the new smaller value turns out to break something,
3275 I'll probably just bump it back up.)
3276 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3278 * Many, many bugs reported by Peter Van Eynde have been added to
3279 the BUGS list; some have even been fixed.
3280 * While enabling dynamic object file loading, I tried to make the
3281 code easier to understand, renaming various functions and variables
3282 with less ambiguous names, and changing some function calling
3283 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3284 * While trying to figure out how to do the OpenBSD port, I tried to
3285 clean up some of the code in src/runtime/. In particular, I dropped
3286 support for non-POSIX signal handling, added various comments,
3287 tweaked the code to reduce the number of compilation warnings, and
3288 renamed some files to increase consistency.
3289 * To support the new automatic configuration functionality in make.sh,
3290 the source file target-features.lisp-expr has been replaced with the
3291 source file base-target-features.lisp-expr and the machine-generated
3292 file local-target-features.lisp-expr.
3293 * fixed a stupid quoting error in make.sh so that using CMU CL
3294 "lisp -batch" as cross-compilation host works again
3296 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3298 * DESCRIBE no longer tries to call itself recursively to describe
3299 bound/fbound values, so that it no longer fails on symbols which are
3300 bound to themselves (like keywords, T, and NIL).
3301 * DESCRIBE now works on generic functions.
3302 * The printer now prints less-screwed-up representations of closures
3303 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3304 * A private symbol is used instead of the :EMPTY keyword previously
3305 used to mark empty slots in hash tables. Thus
3306 (DEFVAR *HT* (MAKE-HASH-TABLE))
3307 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3308 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3309 now does what ANSI says that it should. (You can still get
3310 similar noncompliant behavior if bang on the hash table
3311 implementation with all the symbols you get back from
3312 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3313 This breaks binary compatibility, since tests for equality to
3314 :EMPTY are wired into things like the macroexpansion of
3315 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3317 * There's now a minimal placeholder implementation for CL:STEP,
3318 as required by ANSI.
3319 * An obscure bug in the interaction of the normal compiler, the byte
3320 compiler, inlining, and structure predicates has been patched
3321 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3322 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3324 * Missing ordinary arguments in a macro call are now detected even
3325 when the macro lambda list contains &KEY or &REST.
3326 * The debugger no longer complains about encountering the top of the
3327 stack when you type "FRAME 0" to explicitly instruct it to go to
3328 the top of the stack. And it now prints the frame you request even
3329 if it's the current frame (instead of saying "You are here.").
3330 * As specified by ANSI, the system now always prints keywords
3331 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3333 * The default initial SIZE of HASH-TABLEs is now smaller.
3334 * Type information from CLOS class dispatch is now propagated
3335 into DEFMETHOD bodies, so that e.g.
3336 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3338 is now basically equivalent to
3339 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3340 (DECLARE (TYPE SINGLE-FLOAT X))
3342 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3343 operation, without having to do run-time type dispatch.
3344 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3345 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3346 * Since I don't seem to be making big file reorganizations very often
3347 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3348 is overflowing my ability to conveniently back them up), I've finally
3349 checked the system into CVS. (The CVS repository is on my home system,
3350 not at SourceForge -- putting it on SourceForge might come later.)
3351 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3352 high-level GC-NOTIFY-FOO functions send their output. (There's
3353 still very little control of where low-level verbose GC functions
3354 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3355 less than it used to -- the GC-NOTIFY-FOO functions are now under
3356 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3357 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3358 in only one place in the source code, and propagates it automatically
3359 everywhere that it's needed. Thus e.g. when I bump the version from
3360 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3361 * The C source files now include boilerplate legalese and documentation
3362 at the head of each file (just as the Lisp source files already did).
3363 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3364 to his page will be replaced with a link to his new CLiki service.
3366 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3368 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3369 * The build process now looks for GNU make under the default name "gmake",
3370 instead of "make" as it used to. If GNU make is not available as "gmake"
3371 on your system, you can change this default behavior by setting the
3372 GNUMAKE environment variable.
3373 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3374 can build without error under CMU CL.
3376 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3378 * There is now a partial SBCL user manual (with some new text and some
3379 text cribbed from the CMU CL manual).
3380 * The beginnings of a profiler have been added (starting with the
3381 CMU CL profiler and simplifying and cleaning up). Eventually the
3382 main interface should be through the TRACE macro, but for now,
3383 it's still accessed through vaguely CMU-CL-style functions and macros
3384 exported from the package SB-PROFILE.
3385 * Some problems left over from porting CMU CL to the new
3386 cross-compilation bootstrap process have been cleaned up:
3387 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3388 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3389 part when getting rid of the compiler *BACKEND* stuff.)
3390 ** The value of *NULL-TYPE* was screwed up, because it was
3391 being initialized before the type system knew the final
3392 definition of the 'NULL type. This screwed up several key
3393 optimizations in the compiler, causing inefficiency in all sorts
3394 of places. (I found it because I wanted to understand why
3395 GET-INTERNAL-RUN-TIME was consing.)
3396 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3397 PROCLAIM FTYPE information. Unfortunately this broke binary
3398 compatibility again, since now the forms output by DEFGENERIC
3399 to refer to functions which didn't exist in 0.6.3.
3400 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3401 can use the new (as of 0.6.3) transform for SOME into MAP into
3403 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3404 values to (INTEGER 0 1000000), so that the time code will no longer
3405 occasionally get blown up by Linux returning 1000000 microseconds
3406 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3407 its output conform to the ANSI spec. (Alas, this makes its output
3408 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3409 * A full call to MAP NIL with a single sequence argument no longer conses.
3410 * fixes to problems pointed out by Martin Atzmueller:
3411 * The manual page no longer talks about multiprocessing as though
3412 it were currently supported.
3413 * The ILISP support patches have been removed from the distribution,
3414 because as of version 5.10.1, ILISP now supports SBCL without us
3415 having to maintain patches.
3416 * added a modified version of Raymond Toy's recent CMU CL patch for
3417 EQUALP comparison of HASH-TABLE
3419 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3421 * The system still can't cross-compile itself with
3422 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3423 entails), but at least it can compile more of itself that way
3424 than it used to be able to, and various buglets which were uncovered
3425 by trying to cross-compile itself that way have now been fixed.
3426 * This release breaks binary compatibility again. This time
3427 at least I've incremented the FASL file format version to 2, so that the
3428 problem can be detected reliably instead of just causing weird errors.
3429 * various new style warnings:
3430 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3431 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3432 ** using the lexical binding of a variable named in the *FOO* style
3433 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3434 as specified by ANSI.
3435 * *RANDOM-STATE* is no longer automatically initialized from
3436 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3437 default behavior of the system is to repeat its behavior every time
3438 it's run. If you'd like to change this behavior, you can always
3439 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3440 old convention there was no comparably easy way to get the system to
3441 repeat its behavior every time it was run.
3442 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3443 FTYPE declarations has been removed, in favor of their ANSI
3444 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3445 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3446 the types of their sequence arguments can be determined at compile time.
3447 This is done through a new open code expansion for MAP which eliminates
3448 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3449 argument types can be determined at compile time.
3450 * The optimizer now transforms COERCE into an identity operation when it
3451 can prove that the coerced object is already of the correct type. (This
3452 can be a win for machine generated code, including the output of other
3453 optimization transforms, such as the MAP transform above.)
3454 * Credit information has been moved from source file headers into CREDITS.
3455 * Source file headers have been made more standard.
3456 * The CASE macro now compiles without complaining even when it has
3459 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3461 * (Note that the way that the PCL macroexpansions were rewritten
3462 to accommodate the change in DEFGENERIC below breaks binary
3463 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3464 not run under sbcl-0.6.2. Once we get out of alpha releases,
3465 i.e. hit release 1.0.0, we'll probably try to maintain binary
3466 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3467 and sbcl-1.4.4. Until then, however, it might be fairly common
3468 for maintenance releases to break binary compatibility.)
3469 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3471 * The reporting of unhandled signals has been changed to print some
3472 explanatory text as well as the report form. (Previously only
3473 the report form was printed.)
3474 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3475 it defines, so that the compiler no longer issues undefined function
3476 warnings for compiled-but-not-yet-loaded generic functions.
3477 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3478 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3479 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3480 section of the ANSI spec are supported.
3481 * The "" nickname for the "KEYWORD" package has been removed.
3482 The reader still handles symbol tokens which begin with a package marker
3483 as keywords, but it doesn't expose its mechanism for doing so in the
3484 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3485 * The system now issues STYLE-WARNINGs for contradictory TYPE
3486 proclamations. (Warnings for contradictory FTYPE proclamations would
3487 be nice too, but those can't be done usefully unless the type system
3488 is made smarter about FUNCTION types.)
3489 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3490 systematized, so that "*target-*.lisp is supposed to exist only on the
3491 target and imply that there's a related file which exists on the
3492 host, and *host-*.lisp is supposed to exist only on the host and imply
3493 that there's a related file which exists on the target. This involves a
3494 lot of renaming. Hopefully the acute confusion caused by the renaming
3495 will be justified by the reduction in chronic confusion..
3496 ** runtime-type.lisp -> early-target-type.lisp
3497 ** target-type.lisp -> late-target-type.lisp
3498 ** early-host-format.lisp -> early-format.lisp
3499 ** late-host-format.lisp -> late-format.lisp
3500 ** host-error.lisp -> misc-error.lisp
3501 ** early-error.lisp -> early-target-error.lisp
3502 ** late-error.lisp -> late-target-error.lisp
3503 ** host-defboot.lisp -> early-defboot.lisp
3504 ** code/misc.lisp -> code/target-misc.lisp
3505 ** code/host-misc.lisp -> code/misc.lisp
3506 ** code/numbers.lisp -> code/target-numbers.lisp
3507 ** code/early-numbers.lisp -> numbers.lisp
3508 ** early-host-type.lisp -> early-type.lisp
3509 ** late-host-type.lisp -> late-type.lisp
3510 ** host-typep.lisp -> typep.lisp
3511 ** load.lisp -> target-load.lisp
3512 ** host-load.lisp -> load.lisp
3513 ** host-disassem.lisp -> disassem.lisp
3514 ** host-insts.lisp -> insts.lisp
3515 ** byte-comp.lisp -> target-byte-comp.lisp
3516 ** host-byte-comp.lisp -> byte-comp.lisp
3517 ** host-signal.lisp -> signal.lisp
3518 ** host-defstruct.lisp -> defstruct.lisp
3519 ** late-target-type.lisp -> deftypes-for-target.lisp
3520 Furthermore, several other previously target-only files foo.lisp (e.g.
3521 hash-table.lisp and random.lisp) have been split into a target-and-host
3522 foo.lisp file and a target-only target-foo.lisp file, with their key type
3523 definitions in the target-and-host part, so that the cross-compiler will
3524 know more about target types.
3525 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3526 gone away. In their place are various *BACKEND-FOO* variables
3527 corresponding to the slots of the old structure.
3528 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3529 into the target SBCL has been fixed.
3530 * The chill.lisp system for loading cold code into a running SBCL
3532 * Support for the CMU CL "scavenger hook" extension has been removed.
3533 (It was undocumented and unused in the CMU CL sources that SBCL was
3534 derived from, and stale in sbcl-0.6.1.)
3535 * Various errors in the cross-compiler type system were detected
3536 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3537 (enabling various consistency checks). Many of them were fixed,
3538 but some hard problems remain, so the compiler is back to
3539 running without *TYPE-SYSTEM-INITIALIZED* for now.
3540 * As part of the cross-compiler type system cleanup, I implemented
3541 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3542 * I have started adding UNCROSS calls throughout the type system
3543 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3544 on cross-compiler input files will be able to go away, and various
3546 * CONSTANTP now returns true for quoted forms (as explicitly required
3549 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3551 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3552 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3553 (MAKE-LIST -1). In the longer term, it ought to become true
3554 that declarations are assertions even at SAFETY 1. For now, it's not
3555 quite true even at SAFETY 3, but it's at least more nearly true..
3556 (Note that this change seems to increases the size of the system by
3557 O(5%) and to decrease the speed of the compiler by 20% or more.)
3558 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3559 for the problem of printing dozens of lines of distracting information
3560 about low-level system machinery as part of the top stack frame
3561 on entry to the debugger when an undefined function was called.
3562 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3563 is not reset to COMMON-LISP-USER.
3564 * Compilation of stuff related to dyncount.lisp has been made conditional
3565 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3566 smaller. The various dyncount-related symbols have been moved into
3567 a new "SB-DYNCOUNT" package.
3568 * tty-inspect.lisp has been renamed to inspect.lisp.
3569 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3570 feature has gone away. (When we eventually port to other flavors of
3571 libc and/or Unix, we'll try to make the differences between flavors
3572 invisible at the user level.)
3573 * Various other *FEATURES* tags, and/or their associated conditionals,
3574 have been removed if obsolescent, or given better documentation, or
3575 sometimes given more-mnemonic names.
3577 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3579 * tidied up "make.sh" script
3580 * tidied up system directory structure
3581 * better "clean.sh" behavior
3582 * added doc/FOR-CMUCL-DEVELOPERS
3583 * many many small tweaks to output format, e.g. removing possibly-confusing
3584 trailing #\. character in DESCRIBE-INSTANCE
3585 * (EQUALP #\A 'A) no longer signals an error.
3586 * new hashing code, including EQUALP hashing
3587 * tidied up Lisp initialization and toplevel
3588 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3589 * command line argument processing
3590 * added POSIX-GETENV function to deal with Unix-ish environment variables
3591 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3592 terminating SBCL on EOF
3593 * non-verbose GC by default
3594 * There is no more "sbcl" shell script; the sbcl file is now the C
3595 runtime executable (just like CMU CL).
3596 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3598 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3599 symbol and package information as Lisp native symbol and package objects
3600 instead of strings naming symbols and strings naming packages. This way,
3601 whenever packages are renamed (as in warm init), debug information is
3602 transformed along with everything else.
3603 * tweaked the optimization policy declarations which control the building
3604 of SBCL itself. Now, among other things, the system no longer saves
3605 source location debugging information. (This helps two problems at once
3606 by reducing SBCL size and by keeping SBCL from trying to look for its
3607 sources -- which may not exist -- when reporting errors.)
3608 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3609 debugging and testing purposes
3610 * cleaned up printing, making the printer call PRINT-OBJECT for
3611 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3612 methods, giving nearly-ANSI behavior
3613 * converted almost all special variables to use *FOO* naming convention
3614 * deleted PARSE-TIME functionality, since it can be done portably
3615 * moved some files out of cold init into warm init
3616 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3618 * regularized formatting of source files
3619 * added an install.sh script
3620 * fixed ridiculous memory usage of cross-compiler by making
3621 compiler/alloc.lisp not try to do pooling unless it can hook
3622 itself into the GC of the cross-compilation host. Now the system
3623 builds nicely on my old laptop.
3624 * added :SB-ALLOC in target-features.lisp-expr
3625 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3626 was not implemented)
3627 * re-did condition handling and note reporting in the compiler. Notes
3628 are no longer handled by signalling conditions. Style warnings
3629 and warnings are handled more correctly and reported in such a way
3630 that it's easy to find one or the other in your output (so that you
3631 can e.g. figure out which of many problems caused COMPILE-FILE to
3633 * changed the severity of several compiler warnings from full WARNING
3634 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3635 compiler note reporting so that it doesn't use the condition system
3636 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3637 COMPILE-FILE command)
3638 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3639 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3640 no longer supported, since their implementation was deeply intertwingled
3641 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3642 * removed bogus "support" for compiler macros named (SETF FOO), and
3643 removed the compiler macro for SETF INFO (but only after making a fool
3644 of myself on the cmucl-imp mailing list by posting a bogus patch for
3645 DEFINE-COMPILER-MACRO..)
3646 * Compiled files containing forms which have side effects on the Lisp
3647 reader (such as DEFPACKAGE forms) are now handled more correctly.
3648 (Compiler queuing of top level lambdas has been suppressed by setting
3649 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3650 * deleted various currently-unused source files, e.g. gengc.lisp. They
3651 may be added back at some point e.g. when porting to other architectures,
3652 but until they are it's distracting to distribute them and to try to
3654 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3655 there were so many of them they're just distractions, and UNCROSS is
3656 known to be able to handle the current sources
3657 * moved PROFILE functionality into TRACE, so that it will be clear
3658 how the wrapping and unwrapping of functions when you profile them
3659 interacts with the wrapping and unwrapping of functions when you
3660 trace them. (Actually, the functionality isn't there yet, but at least
3661 the interface specification is there. Hopefully, the functionality will
3662 arrive with some future maintenance release.)
3663 * removed host-oops.lisp
3664 * changed signature of QUIT function to allow UNIX-CODE argument
3665 * fixed READ-SEQUENCE bug
3666 * tweaked verbose GC output so that it looks more like the progress
3667 output that ANSI specifies for functions like LOAD
3668 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3669 * added <http://sbcl.sourceforge.com> to the banner information printed by