1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 changes relative to sbcl-1.0.37:
3 * INCOMPATIBLE CHANGE: Thread names are now restricted to SIMPLE-STRINGs
4 like for any other thread-related datastructure, MUTEX, etc. (lp#547095)
5 * DEPRECATION: the SB-QUEUE contrib was merged into the SB-CONCURRENCY
6 contrib module. New code should depend on SB-CONCURRENCY, not SB-QUEUE.
7 * DEPRECATION: SB-THEAD:GET-MUTEX was deprecated in favor of
9 * new contrib: SB-CONCURRENCY is a new contrib; it's supposed to contain
10 additional data structures and tools for concurrent programming; at the
11 moment it contains a lock-free queue, and a lock-free mailbox
13 * new feature: added SB-THREAD:GRAB-MUTEX; it's like the now deprecated
14 GET-MUTEX but takes &key rather than &optional parameters. Also added
15 :TIMEOUT argument to GRAB-MUTEX on non-sb-lutex platforms like Linux and
17 * new feature: added SB-THREAD:TRY-SEMAPHORE, a non-blocking variant of
18 SB-THREAD:WAIT-ON-SEMAPHORE.
19 * new feature: SB-EXT:ATOMIC-DECF has been added as a companion to
21 * new feature: a CANCEL-DEADLINE is associated with DEADLINE-TIMEOUT
22 conditions to defer the deadline for forever.
23 * enhancement: *STANDARD-OUTPUT*, *STANDARD-INPUT*, and *ERROR-OUTPUT* are
25 * enhancement: errors from NO-APPLICABLE-METHOD and
26 NO-PRIMARY-METHOD now have a RETRY restart available to retry the
27 generic function call.
28 * enhancement: SB-BSD-SOCKET improvements
29 ** sockets and socket streams now have a more informative printed
30 representation based on the corresponding SOCKET-NAME and
32 ** SOCKET-MAKE-STREAM once more supports the :AUTO-CLOSE option.
34 ** SOCKET-CLOSE now accepts :ABORT argument, which is passed on to
35 CL:CLOSE when appropriate, and no longer disassociates the stream
36 from the socket if close failed. (lp#543951)
37 * improvements to the instrumenting profiler
38 ** new feature: report per-function GC overhead. (thanks to John Fremlin)
39 ** optimization: counters no longer use locks for the overflow mode.
40 ** bug fix: whenever a profiling counter wrapped into overflow mode, it
41 incurred an off-by-one miscount.
42 * bug fix: correct restart text for the continuable error in MAKE-PACKAGE.
43 * bug fix: a rare case of startup-time page table corruption.
44 * bug fix: a semaphore with multiple waiters and some of them unwinding due
45 to timeouts could be left in an inconsistent state.
46 * bug fix: fix typo in "Reporting Bugs" section of the manual (lp#520366)
47 * bug fix: misoptimization of multiplication by one in
48 (SB-C::FLOAT-ACCURACY 0) policies.
49 * bug fix: miscounts in SB-PROFILE.
50 * bug fix: Fix lost wakeup bug between SB-THREAD:CONDITION-WAIT and
51 CONDITION-NOTIFY on Linux. See threads "lost wakeup in condition-wait /
52 condition-notify" (Feb 2010) and "Condition-Wait, Deadline handler, waking
53 up itself" (March 2010) for further details.
54 * bug fix: allow forward FIND and POSITION on lists to elide checking :END
55 against length of the list if the element is found before the specified
56 END is reached. (thanks to Alec Berryman, lp#554385)
57 * bug fix: errors signalled during package graph modification no longer
58 block FIND-SYMBOL and FIND-PACKAGE in other threads. (lp#511072)
59 * bug fix: SB-POSIX build was broken when SBCL was compiled without the
60 :SB-DOC feature. (lp#552564)
62 changes in sbcl-1.0.37 relative to sbcl-1.0.36:
63 * enhancement: Backtrace from THROW to uncaught tag on x86oids now shows
64 stack frame thrown from.
65 * enhancement: WITH-COMPILATION-UNIT :POLICY allows restricting changes to
66 compiler optimization qualities inside dynamic extent of its body.
67 * enhancement: LOAD-LOGICAL-PATHNAME-TRANSLATIONS can be used to load
68 translations from SYS:SITE;<HOST>.TRANSLATIONS.NEWEST (thanks to Michael
70 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) take advantage of
71 constraint propgation, allowing better compilation eg. when used to
72 access structures with WITH-SLOTS. (lp#520366)
73 * optimization: the compiler is now more aware of the type of the underlying
74 storage vector for multidimensional simple arrays resulting in better code
75 for accessing such arrays.
76 * optimization: passing NIL as the environment argument to TYPEP no longer
77 inhibits optimizing it. (lp#309788)
78 * optimization: more efficient register usage when handling single-float
79 arguments on x86-64. (thanks to Lutz Euler)
80 * optimization: ADJUST-ARRAY and STABLE-SORT on vectors no longer use
81 pre-allocated temporary vectors. (lp#496249)
82 * bug fix: Fix compiler error involving MAKE-ARRAY and IF forms
83 in :INITIAL-CONTENTS. (lp#523612)
84 * bug fix: FUNCTION-LAMBDA-EXPRESSION lost declarations from interpreted
85 functions. (lp#524707)
86 * bug fix: bogus style warnings from certain (SETF SLOT-VALUE) and
87 WITH-SLOTS usages during compilation.
88 * bug fix: SB-C::CLASS-INFO now prints correctly. (lp#514762)
89 * enhancement: Can now build with ud2 instead of int3 as trap instruction on
90 all x86oid platforms with :UD2-BREAKPOINTS target feature.
91 * bug fix: Breakpoints now work when using ud2 instead of int3 as trap
92 instruction (tested on x86oid linux with ud2-breakpoints).
93 * bug fix: slam.sh now works on win32.
94 * bug fix: better differences of numeric types. (lp#309124)
95 * bug fix: arrays declared intersection and union types can have their
96 upgraded element type derived. (lp#316078)
97 * bug fix: SB-SPROF allocation profiling for all threads failed to profile
98 threads started during profiling. (lp#472499)
99 * bug fix: SB-INTROSPECT test failure when building without SB-EVAL feature.
101 * bug fix: SB-CLTL2:DECLARATION-INFORMATION did not take
102 SB-EXT:RESTRICT-COMPILER-POLICY into account. (lp#313337)
103 * bug fix: Comma inside a backquoted array or structure resulted in nonsense
104 values instead of signaling an error. (lp#309093)
105 * bug fix: Spurious unused variable warning in a DEFSTRUCT edge case.
107 * bug fix: More consistent warnings and notes for ignored DYNAMIC-EXTENT
108 declarations (lp#497321)
109 * bug fix: FIND and POSITION on lists did not check sequence bounds properly
110 and failed to detect circular lists (lp#452008)
111 * bug fix: leakage from ~/.asdf-install into the ASDf-INSTALL contrib build
113 * bug fix: LOOP OF-TYPE VECTOR compile-time error. (lp#540186)
114 * bug fix: SIGNAL SB-SYS:INTERACTIVE-INTERRUPT before entering the debugger
115 due to it, so that handlers can run.
116 * bug fix: reparsing undefined types if they have become defined since
118 * bug fix: missing &REST type in a proclamation for a function with both
119 &REST and &KEY in lambda-list caused miscompilation (lp#458354)
120 * bug fix: WHO-CALLS information for source-transformed and compiler-macro
121 expanded calls (lp#542174)
122 * bug fix: more accurate WHO-MACROEXPANDS information; point into rather
123 than just at toplevel form.
125 changes in sbcl-1.0.36 relative to sbcl-1.0.35:
126 * new feature: SB-EXT:TYPEXPAND-1, SB-EXT:TYPEXPAND, and
127 SB-EXT:TYPEXPAND-ALL behave exactly like their MACROEXPAND counterparts
128 but work on type specifiers.
129 * new feature: SB-EXT:DEFINED-TYPE-NAME-P returns whether a symbol is known
130 to name a type specifier.
131 * new feature: SB-EXT:VALID-TYPE-SPECIFIER-P returns whether a given type
132 specifier is valid where "valid" basically means "would be accepted as
133 second argument of TYPEP".
134 * new feature: SB-INTROSPECT:FUNCTION-TYPE takes a function-designator and
135 returns the function's declared, or derived FTYPE.
136 * new feature: SB-POSIX now supports accessing the d_ino member of
137 dirent structures. (Thanks to Philipp Marek and Pierre THEIRRY)
138 * new feature: The function SB-EXT:SEED-RANDOM-STATE has been added to
139 provide for seeding a RANDOM-STATE object with user-provided data or
140 from the operating system's PRNG. Also, (MAKE-RANDOM-STATE T) will
141 attempt to initialize the returned state from the operating system's
142 PRNG where possible. (Thanks to Fare Rideau; launchpad bug lp#310116)
143 * bug fix: Fix SB-SIMPLE-STREAMS:READ-VECTOR to correctly set the
144 FILE-POSITION of the stream being read from. (launchpad bug lp#491087)
145 * bug fix: Fix grammar and style issues for the docstrings of
146 printer-related variables and functions. (Thanks to mon_key; launchpad
148 * bug fix: Fix compilation on chenygc platforms. Thanks to Larry Valkama and
150 * bug fix: SB-THREAD:CONDITION-WAIT sometimes signaled a deadline twice
151 in a row even though a handler defered the deadline long into the
153 * bug fix: A deadline handler was run without interrupts enabled for a
154 deadline signaled within SB-THREAD:CONDITION-WAIT. That could result
155 in infinitely spinning, non-killable threads.
156 * bug fix: Backtrace from internal-errors on x86-64 os x was truncated
157 before reaching the erring stack frame.
158 * bug fix: Fix type derivation for EXPT when raising a fixnum to a
159 real power. (launchpad bug lp#525949)
160 * bug fix: Fix SB-EXT:GENERATION-* accessors for generations > 0 on
161 GENCGC platforms. (launchpad bug lp#529014)
162 * bug fix: More robust checks for invalid DEFMETHOD argument specializers.
163 (launchpad bug lp#525916)
164 * bug fix: Fix building on Darwin when sysctl is not in the user's PATH.
165 (Thanks to Robert Goldman)
167 changes in sbcl-1.0.35 relative to sbcl-1.0.34:
168 * optimization: ROUND with a single single-float or double-float argument
169 is properly inlined when possible.
170 * optimization: Slightly better code is generated for integer<->float
171 conversions and for single-float<->double-float conversions on x86-64.
172 * optimization: SB-ROTATE-BYTE:ROTATE-BYTE now generates more efficient
173 code for 32-bit and 64-bit rotations on x86-64.
174 * bug fix: The install script changes the ownership of directories as well
175 as files for contrib modules using asdf. (thanks to Eugene Ossintsev;
176 launchpad bug lp#508485)
177 * bug fix: TRUNCATE with a single single-float or double-float argument is
178 properly inlined when possible. (launchpad bug lp#489388)
179 * bug fix: Passing a rotation count of zero to SB-ROTATE-BYTE:ROTATE-BYTE
180 no longer causes a compiler error on x86 and ppc.
181 * bug fix: GET-MACRO-CHARACTER bogusly computed its second return value
182 always relative to *READTABLE* rather than the passed argument.
184 changes in sbcl-1.0.34 relative to sbcl-1.0.33:
185 * minor incompatible change: threading support is now enabled by default
187 * enhancement: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST now also works on most
189 * enhancement: Errors during compile-time-too processing (i.e. EVAL-WHEN)
190 are now caught and reported just like errors during macroexpansion.
191 * enhancement: SB-POSIX now provides access to tcdrain(), tcflow(),
192 tcflush(), tcgetsid(), and tcsendbreak(). (thanks to Jerry James)
193 * enhancement: ASDF systems can now depends on SB-QUEUE.
194 * fixes and improvements related to Unicode and external formats:
195 ** bug fix: error handling and restart usage in the ucs-2 external format
197 ** there is now an implementation of the ucs-4 external format.
198 ** the utf-16 and utf-32 external formats are supported.
199 * bug fix: SB-POSIX wrapper for putenv no longer tries to put lisp strings
200 in the environment. setenv() and unsetenv() are also provided. (reported by
201 Fare Rideau; launchpad bug lp#460455)
202 * bug fix: LOAD of both .fasl and .FASL type files now forces fasl-style
203 loading. This ensures sensible errors for .FASL files from other
204 implementations on case-insensitive filesystems. (reported by Willem
205 Broekema; launchpad bug lp#489417)
206 * bug fix: #p"\\\\" can now be read without error on Win32. (reported by
207 Willem Broekema; launchpad bug lp#489698).
208 * bug fix: some minor code rearrangements to reenable warning-free building
209 from CMUCL (reported by xme@gmx.net; launchpad bug lp#491104)
210 * bug fix: PRINT-OBJECT for clos instances respects the right margin when
212 * bug fix: FIND-PACKAGE & DEFPACKAGE were not thread safe. (reported by
215 changes in sbcl-1.0.33 relative to sbcl-1.0.32:
216 * new port: support added for x86-64 NetBSD. (thanks to Aymeric Vincent)
217 * improvement: support O_LARGEFILE access to files larger than 2GB on
218 x86-64/linux. (thanks to Daniel Janus; launchpad bug lp#453080)
219 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-DIRECTLY to get a list of
220 definitions for methods specializing on the passed class itself.
221 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-GENERALLY to get a list of
222 definitions for methods specializing on the passed class itself, or on
224 * new build flag: :sb-xref-for-internals; SBCL will collect xref information
225 about itself during the build (e.g. for M-? in Slime), if this flag is
226 enabled in customize-target-features.lisp. This will increase the core
227 size by about 5-6mb, though, so it's mostly interesting to SBCL
229 * new feature: various GENCGC tuning parameters have been experimentally
230 documented and exported from SB-EXT. See documentation for details.
231 * fixes and improvements related to Unicode and external formats:
232 ** the Unicode character database has been upgraded to the
233 Unicode 5.2 standard, giving names and properties to a number of new
234 characters, and providing a few extra characters with case
236 ** improvement: restarts for providing replacement input/output on coding
237 errors for fd-stream external formats.
238 ** improvement: where :<encoding> is a keyword corresponding to an
239 external format the system supports, it is now possible to specify
240 (:<encoding> :replacement <character>) as an external format which will
241 automatically substitute <character> on encoding or decoding errors for
242 streams and for STRING-TO-OCTETS and its inverse. (launchpad bug
244 ** improvement: the file streams underlying the standard streams (such as
245 *STANDARD-INPUT*, *TERMINAL-IO*) are opened with an external format
246 which uses the replacement mechanism to handle encoding errors,
247 preventing various infinite error chains and unrecoverable I/O
249 ** minor incompatible change: the utf-8 external format now correctly
250 refuses to encode Lisp characters in the surrogate range (char-codes
251 between #xd800 and #xdfff).
252 ** fix a typo preventing conversion of strings into octet vectors
253 in the latin-2 encoding. (reported by Attila Lendvai; launchpad bug
255 ** fix a bug in the octet multibyte handling of decoding errors and the
256 USE-VALUE restart. (launchpad bug lp#314939)
257 ** fix the bug underlying the expected failure in the FORCE-END-OF-FILE
258 restart on fd-stream decoding errors.
259 ** fix a bug in the ATTEMPT-RESYNC fd-stream decoding restart when the
260 error is near the end of file.
261 ** fix a double-error case in unibyte octet conversions, when the first
262 use of USE-VALUE is ignored.
263 ** fix bugs in handling of undefined code points in unibyte encodings.
264 ** fix LISTEN (and consequent hangs in READ-CHAR-NO-HANG) on bivalent
265 streams after an UNREAD-CHAR.
266 * enhancement: SB-INTROSPECT:ALLOCATION-INFORMATION also reports if the
267 object is allocated in a boxed region of dynamic space.
268 * enhancement: SB-POSIX:FORK now signals an error if an attempt to
269 fork with multiple Lisp threads running is made, instead of going
270 ahead with unpredictable consequences. (reported by Leslie Polzer)
271 * bug fix: uses of slot accessors on specialized method parameters within
272 the bodies of SLOT-VALUE-USING-CLASS methods no longer triggers a type
273 error while finalizing the class. This fix may cause classes with slot
274 accessors to be finalized later than previously. (reported by Lars Rune
275 Nøstdal; launchpad bug lp#473699)
276 * bug fix: restore buildability on the MIPS platform. (regression from
277 1.0.30.38, reported by Samium Gromoff)
278 * bug fix: inspecting closures is less likely to fail with a type error.
279 * bug fix: no timer starvation when setting the system clock back.
280 (launchpad bug lp#460283)
281 * bug fix: WITH-STANDARD-IO-SYNTAX now binds *PRINT-PPRINT-DISPATCH* to the
282 standard pprint dispatch table as specified by CLHS.
283 * bug fix: give CLISP a hint about a type declaration to enable it to build
284 the cross-compiler without warnings. (thanks to Josh Elasser; launchpad
286 * bug fix: correctly dump literal objects in defaulting forms of arglists.
287 (reported by Attila Lendvai; launchpad bug lp#310132)
288 * bug fix: distinguish in type specifiers between arrays that might be
289 complex and arrays that are definitely complex. (launchpad bug lp#309129)
290 * bug fix: SUBTYPEP knows that the SYMBOL type is not SUBTYPEP the KEYWORD
291 type. (reported by Levente Mészáros; launchpad bug lp#485972)
292 * bug fix: setting the value of a symbol-macro within a method in the
293 presence of type declarations works properly again. (reported by Iban
294 Hatchondo; launchpad bug lp#485019)
296 changes in sbcl-1.0.32 relative to sbcl-1.0.31:
297 * optimization: faster FIND and POSITION on strings of unknown element type
298 in high SPEED policies. (thanks to Karol Swietlicki)
299 * optimization: faster CONCATENATE 'STRING in low SPEED policies (reported
301 * improvement: better error signalling for bogus parameter specializer names
302 in DEFMETHOD forms (reported by Pluijzer)
303 * bug fix: DELETE-FILE once again works on logical pathnames (regression
305 * bug fix: LOGICAL-PATHNAME signals a TYPE-ERROR if pathspec is specified
307 * bug fix: redefinition of a class via DEFCLASS without :DEFAULT-INITARGS
308 removes previous default initargs (reported by Lars Rune Nøstdal and
310 * bug fix: correct WHO-CALLS information for inlined lambdas with complex
311 lambda-lists. (reported by Peter Seibel)
312 * bug fix: SAVE-LISP-AND-DIE option :SAVE-RUNTIME-OPTIONS did not work
313 correctly when starting from an executable core without saved runtime
314 options (reported by Faré Rideau, thanks to Zach Beane)
315 * bug fix: (SETF SLOT-VALUE) signalled a warning which should have been
316 an optimization note instead. (reported by Martin Cracauer)
317 * bug fix: WITH-SLOTS did not work with THE forms. (thanks to David Tolpin)
318 * bug fix: Have RUN-PROGRAM with :INPUT T only run the subprocess in a
319 new process group if it doesn't need to share stdin with the sbcl
320 process. (thanks to Leslie Polzer)
321 * bug fix: SATISFIES could be misoptimized to refer to a local function.
322 (reported by Stanislaw Halik)
324 changes in sbcl-1.0.31 relative to sbcl-1.0.30:
325 * improvement: stack allocation is should now be possible in all nested
326 inlining cases: failure to stack allocate when equivalent code is manually
327 open coded is now considered a bug.
328 * improvements related to Unicode:
329 ** the Unicode character database has been upgraded to the
330 Unicode 5.1 standard, giving names and properties to a number of new
331 characters, and providing a few extra characters with case
333 ** the system now recognizes and produces names for Unicode Hangul
335 ** the EBCDIC-US external-format is now supported for octet operations
336 (as well as for stream operations).
337 * new feature: experimental :EMIT-CFASL parameter to COMPILE-FILE can
338 be used to output toplevel compile-time effects into a separate .CFASL
340 * optimization: COERCE to VECTOR, STRING, SIMPLE-STRING and recognizable
341 one-dimenstional subtypes of ARRAY is upto 70% faster when the coercion is
343 * optimization: TRUNCATE on known single- and double-floats is upto 25%
345 * optimization: division of floating point numbers by constants uses
346 multiplication by reciprocal when an exact reciprocal exists.
347 * optimization: multiplication of single- and double-floats floats by
348 constant two has been optimized.
349 * optimization: ARRAY-IN-BOUNDS-P is resolved at compile-time when
350 sufficient type information is available. (thanks to Leslie Polzer)
351 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) with constant slot names on
352 known structure objects are as efficient as defstruct generated accessors.
353 * optimization: unused vector creation can now be optimized away.
354 * improvement: ASDF systems can now depends on SB-INTROSPECT.
355 * improvement: a STYLE-WARNING is signalled when a generic function
356 clobbers an earlier FTYPE proclamation.
357 * improvement: the compiler is able to track the effective type of
358 generic function across method addition and removal even in the
359 absence of an explicit DEFGENERIC.
360 * improvement: DESCRIBE now reports on symbols naming undefined
361 but assumed or declared function as well.
362 * improvement: recompilation of systems using SB-GROVEL now works
363 (thanks to Leslie Polzer)
364 * improvements to SB-CLTL2 (thanks to Larry D'Anna):
365 ** functions DECLARATION-INFORMATION, PARSE-MACRO, and ENCLOSE have been
367 ** AUGMENT-ENVIRONMENT and DEFINE-DECLARATION have been implemented.
368 ** DECLARATION-INFORMATION now supports declaration name DECLARATION as
369 well as user defined declaration names.
370 ** VARIABLE-INFORMATION is now aware of alien variables.
371 * improvement: improved address space layout on OpenBSD (thanks to Josh
373 * improvement: pretty-printing of various Lisp forms has been improved
374 (thanks to Tobias Rittweiler)
375 * bug fix: calls to DECODE-FLOAT and INTEGER-DECODE-FLOAT whose value was
376 unused were deleted in safe code. (reported by John Fremlin)
377 * bug fix: a failing AVER compiling certain MAKE-ARRAY forms. (reported
379 * bug fix: some out-of-line array predicates were missing (reported by
381 * bug fix: a failing AVER in CONVERT-MV-CALL has been fixed. (thanks to
383 * bug fix: a failing AVER in %ALLOCATE-CLOSURES conversion has been fixed
384 (thanks to Larry D'Anna)
385 * bug fix: SLEEP supports times over 100 million seconds on long on OpenBSD
386 as well. (reported by Josh Elsasser)
387 * bug fix: DELETE-FILE on streams no longer closes the stream with :ABORT T,
388 leading to possible attempts to delete the same file twice. See docstring
389 on DELETE-FILE for details. (reported by John Fremlin)
390 * bug fix: DELETE-FILE once again deletes the file named by the pathname
391 designator argument, rather than its truename. (reported by Luis
393 * bug fix: the low-level debugger had 32-bit assumptions and was missing
394 information about some array types. (thanks to Luis Oliveira)
395 * bug fix: moderately complex combinations of inline expansions could
396 be miscompiled if the result was declared to be dynamic extent.
397 * bug fix: on x86, SAP-REF of sizes greater than 8 bits with offsets of the
398 form (+ <variable> <integer>) were miscompiled under certain
400 * bug fix: in some cases no compiler note about failure to stack allocate
401 was emitted, even if the objects were in fact heap allocated.
402 * bug fix: minor violation of "otherwise inaccessible" rule for stack
403 allocation could cause objects users might reasonably expect to
404 be heap allocated to be stack allocated.
405 * bug fix: DESCRIBE signalled an error for generic functions under
406 certain circumstances. (thanks to Leslie Polzer)
407 * bug fix: Fixed spelling of an error message.
409 changes in sbcl-1.0.30 relative to sbcl-1.0.29:
410 * minor incompatible change: SB-THREAD:JOIN-THREAD-ERROR-THREAD and
411 SB-THREAD:INTERRUPT-THREAD-ERROR-THREAD have been deprecated in favor
412 of SB-THREAD:THREAD-ERROR-THREAD.
413 * new contrib module: SB-QUEUE provides thread-safe lockless FIFO queues.
414 * new feature: docstrings for local and anonymous functions are no longer
415 discarded. (thanks to Leslie Polzer)
416 * new feature: SB-THREAD:SYMBOL-VALUE-IN-THREAD provides access to symbol
417 values in other threads.
418 * new feature: SB-INTROSPECT:ALLOCATION-INFORMATION provides information
419 about object allocation.
420 * optimization: division of a real float by a complex float is implemented
421 with a specialised code sequence.
422 * optimization: MAKE-INSTANCE with non-constant class-argument but constant
423 keywords is an order of magnitude faster.
424 * optimization: MAKE-INSTANCE with constant keyword arguments is x2-4 faster
425 in the presence of :AROUND or non-standard primary INITIALIZE-INSTANCE
426 methods, and similarly for non-standard metaclass classes as long as there
427 are no methods additional on MAKE-INSTANCE.
428 * optimization: more efficient type-checks for FIXNUMs when the value
429 is known to be a signed word on x86 and x86-64.
430 * optimization: compiler now optimizes (EXPT -1 INTEGER), (EXPT -1.0 INTEGER),
431 and (EXPT -1.0d0 INTEGER) into an ODDP test. (thanks to Stas Boukarev and
433 * optimization: compiler is smarter about delegating argument type checks to
435 * optimization: several character functions are now compiled somewhat more
436 efficiently. (reported by Lynn Quam)
437 * optimization: the compiler now derives simple types for LOAD-VALUE-FORMs.
438 * improvement: less unsafe constant folding in floating point arithmetic,
439 especially for mixed complex/real -float operations.
440 * optimization: constant double and single floats are stored in native
441 unboxed format on x86[-64].
442 * optimization: smarter code for arithmetic operations with constant floats,
443 complex floats, or integers on x86[-64].
444 * optimization: smarter code for conjugate/multiplication of float complexes
445 and abs/negate of floats on x86-64.
446 * optimization: more efficient complex float and real float operations on
448 * improvement: complex float division is slightly more stable.
449 * improvement: DESCRIBE output has been reworked to be easier to read and
450 contains more pertinent information.
451 * improvement: failure to provide requested stack allocation compiler notes
452 provided in all cases (requested stack allocation not happening without a
453 note being issued is now considered a bug.)
454 * bug fix: SB-POSIX exports the documented types and functions
455 FILE-DESCRIPTOR and FILENAME, and also the corresponding -DESCRIPTOR
456 types. (reported by "abhi")
457 * bug fix: on 64 bit platforms FILL worked incorrectly on arrays with
458 upgraded element type (COMPLEX SINGLE-FLOAT), regression from 1.0.28.55.
459 (thanks to Paul Khuong)
460 * bug fix: looping around HANDLER-CASE could silently consume stack space
461 on each iteration. (reported by "foobar")
462 * bug fix: better error signalling when calls to functions seeking elements
463 from lists (eg. ADJOIN) are compiled with both :TEST and :TEST-NOT.
464 (reported by Tobias Rittweiler)
465 * bug fix: regressions in DIRECTORY from 1.0.28.61: pattern matching of
466 directory components now works as it used to. (various prolems reported by
467 Michael Becker, Gabriel Dos Reis, Cyrus Harmon, and Harald Hanche-Olsen)
468 * bug fix: :PTY option in RUN-PROGRAM was broken with stream arguments.
469 (reported by Elliot Slaughter, thanks to Stas Boukarev)
470 * bug fix: bogus undefined variable warnings from fopcompiled references to
471 global variables. (thanks to Lars Rune Nøstdal)
472 * bug fix: foreign function names should now appear in backtraces on
473 FC6 as well. (reported by Tomasz Skutnik and Tobias Rautenkranz)
474 * bug fix: SETF compiler macro documentation strings are not discarded
476 * bug fix: GENTEMP is now unaffected by pretty printer dispatch table.
477 (thanks to Alex Plotnick)
478 * bug fix: SLEEP accepts large integer arguments, truncating them to
479 SIGNED-WORD on the assumption that sleeping for 68 years is sufficient
480 for anyone. (reported by Leslie Polzer, thanks to Stas Boukarev)
481 * bug fix: compiler notes for expensive slot type checks could be emitted
482 at runtime MAKE-INSTANCE calls. (reported by Samium Gromoff)
484 changes in sbcl-1.0.29 relative to 1.0.28:
485 * IMPORTANT: bug database has moved from the BUGS file to Launchpad
486 https://bugs.launchpad.net/sbcl
487 Bugs can be reported directly there, or by sending email to
488 sbcl-bugs@lists.sourceforge.net
489 (no subscription required.)
490 * minor incompatible change: under weak type checking policy integer
491 types are weakened less aggressively.
492 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
493 allowed to return, which causes SBCL to quit with exit status 0. Previously
494 if the function returned with a small integer return value, that value
495 was accidentally reused as the exit status.
496 * new feature: SB-EXT:DEFINE-HASH-TABLE-TEST allows defining new arguments
497 to MAKE-HASH-TABLE :TEST, and MAKE-HASH-TABLE has been extended with
498 :HASH-FUNCTION argument. Refer to user manual for details.
499 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
501 * new feature: SB-EXT:GET-TIME-OF-DAY provides access to seconds and
502 microseconds since the Unix epoch on all platforms.
503 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
504 allows the compiler to safely elide boundedness checks for special
506 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
507 the symbol, prohibits both lexical and dynamic binding. This is mainly an
508 efficiency measure for threaded platforms, but also valueable in
510 * new feature: UNC pathnames are now understood by the system on Windows.
511 * optimization: the compiler uses a specialized version of FILL when the
512 element type is know in more cases, making eg. (UNSIGNED-BYTE 8) case
514 * optimization: accesses to potentially non-simple arrays where element type
515 is known are 50% faster.
516 * optimization: compiler now generates faster array typechecking code.
517 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
519 * optimization: multidimensional array accesses in the absence of type
520 information regarding array rank are approximately 10% faster due to
521 open coding of ARRAY-RANK.
522 * optimization: result of (FILL (MAKE-ARRAY ...) ...) and (REPLACE
523 (MAKE-ARRAY ...) ...) can be stack allocated if the result of MAKE-ARRAY
525 * optimization: result of call to VECTOR can now be stack allocated.
526 * optimization: MAKE-ARRAY with :INITIAL-CONTENTS is now vastly faster
527 as long as the resulting array is one-dimensional and has a known
528 element type. In particular, :INITIAL-CONTENTS (LIST ...) where the
529 length of the list matches the known length of the vector does not
530 allocate the list as an intermediate step. Ditto for VECTOR and simple
532 * optimization: MAKE-ARRAY can now stack allocate in the presence of
533 :INITIAL-CONTENTS and :INITIAL-ELEMENT as long as the result has a
534 known element type, and is known to be simple and one dimensional.
535 * improvement: SBCL now emits a compiler note where stack allocation was
536 requested but could not be provided (not in all cases, unfortunately)
537 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
538 * improvement: pretty-printing loop has been implemented properly. (thanks
539 to Tobias Rittweiler)
540 * documentation: CLOS slot typechecing policy has been documented.
541 * bug fix: FILE-AUTHOR no longer signals an error on Windows.
542 * bug fix: SB-SPROF could be foiled by foreign code not have a frame
543 pointer, leading to memory faults. (thanks to Bart Botta)
544 * bug fix: better floating point exception handling on x86/OpenBSD.
545 (thanks to Josh Elsasser)
546 * bug fix: exit status from QUIT when called under --script was lost
547 (reported by Hubert Kauker)
548 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
549 same implementation of FILL to initialize the array, even if a more
550 efficient one was available (reported by Stas Boukarev, thanks to
552 * bug fix: potential miscompilation of array stack allocation on x86 and
553 x86-64. (reported by Time Tossavainen)
554 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
555 result in their subforms being treated as top level forms. (reported by
557 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
558 than a machine register could leave garbage in the high bits of the
559 result register (bug 316325).
560 * bug fix: disable address space randomization Linux/x86-64 as well,
561 not just x86-64. (reported by Ken Olum)
562 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
563 generate incorrect code.
564 * bug fix: #201; type inference for CONS and ARRAY types could derive
565 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
566 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
567 could not be assigned to or bound with PROGV. (reported by Lorenz
569 * bug fix: the value of CL:- in the inspector was the previous expression
570 evaluated rather than the expression being evaluated.
571 * bug fix: constants can no longer be locally declared special.
572 * bug fix: signals delivered to threads started from foreign land (read:
573 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
574 thread by blocking all signals and resignalling.
575 * bug fix: SHARED-INITIALIZE initialized unbound :ALLOCATION :CLASS slots
576 from :INITFORM, if any.
578 changes in sbcl-1.0.28 relative to 1.0.27:
579 * a number of bugs in cross-compilation have been fixed, with the ultimate
580 result that building under (at least) clisp should be much more reliable.
581 * minor incompatible changes: echo-streams now propagate unread-char to the
582 underlying input stream, and no longer permit unreading more than one
584 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
585 frames, allowing for instance more reliable backtraces.
586 * improvement: the debugger REPL can now reference lexical variables
587 by name directly for code compiled with (DEBUG 3).
588 * improvement: errors from malformed declarations now have better source
589 paths associated with them. (thanks to Tobias Rittweiler)
590 * optimization: faster local calls on x86/x86-64
591 * bug fix: some error messages for out-of-bound array indexes confused the
592 index and the bound. (thanks to Stas Boukarev)
593 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
595 * bug fix: running regressions tests in shells without OSTYPE set now works.
596 (reported by Harald Hanche-Olsen)
597 * bug fix: more robust static space exhaustion signalling from
598 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
599 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
600 docstring away instead of storing it under names such as (LAMBDA (X)).
601 (reported by Leslie Polzer)
602 * bug fix: timers could go off in the wrong order, be delayed indefinitely
603 (thanks to Ole Arndt for the patch)
604 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
605 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
606 potentially-invalid effective methods in its cache.
607 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
608 instances as well (thanks to Paul Khuong)
609 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
610 SB-POSIX:WAITPID and their C equivalents.
611 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
613 changes in sbcl-1.0.27 relative to 1.0.26:
614 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
615 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
616 * improvement: the system either recovers from stack exhaustion or dies
617 properly as opposed to leaving the user uncertain of whether the handler
618 trampled on some random memory next to the stack or having to rely on
619 --lose-on-corruption (which is still a good idea to use in production
620 because stack exhaustion can happen in signal handlers which will likely
622 * bug fix: fix gc related interrupt handling bug on ppc (regression from
623 1.0.25.37, reported by Harald Hanche-Olsen)
624 * bug fix: work around signal delivery bug in darwin (regression from
625 1.0.25.44, reported by Sidney Markowitz)
626 * bug fix: fix ERROR leaking memory (reported by David Thompson)
628 changes in sbcl-1.0.26 relative to 1.0.25:
629 * incompatible change: an interruption (be it a function passed to
630 INTERRUPT-THREAD or a timer function) runs in an environment where
631 interrupts can be enabled. The interruption can use
632 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
633 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
634 potentially running out of stack. Keep in mind that in the absence
635 of WITHOUT-INTERRUPTS some potentially blocking operation such as
636 acquiring a lock can enable interrupts.
637 * incompatible change: GC-OFF and GC-ON are removed, as they were
638 always unsafe. Use WITHOUT-GCING instead.
639 * new feature: runtime option --disable-ldb
640 * new feature: runtime option --lose-on-corruption to die at the
641 slightest hint of possibly non-recoverable errors: running out of
642 memory, stack, alien stack, binding stack, encountering a memory
643 fault, etc. In the absence of --lose-on-corruption a warning is
645 * enhancement: detect binding stack exhaustion
646 * enhancement: detect alien stack exhaustion on x86/x86-64
647 * improvement: generally more stable and reliable interrupt handling
648 * improvement: there is a per thread interruption queue,
649 interruptions are executed in order of arrival
650 * improvement: a repeating timer reschedules itself when the it has
651 finished, but expiration times are spaced equally. If an
652 expiration time is in the past it will trigger after a short grace
653 period that may give a chance to other things to run.
654 * optimization: slightly faster gc on multithreaded builds
655 * optimization: faster WITHOUT-GCING
656 * bug fix: when JOIN-THREAD signals an error, do it when not holding
657 important locks so that the debugger/handler doesn't produce
658 recursive errors or deadlock.
659 * bug fix: real-time signals are not used anymore, so no more
660 hanging when the system wide real-time signal queue gets full.
661 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
663 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
664 * bug fix: fix random memory faults related to interrupts on alpha
665 * bug fix: fix random memory faults related to interrupts on sparc
666 * bug fix: fix deadlocks related to starting threads
667 * bug fix: fix deadlines on locks on futex platforms
668 * bug fix: restore errno in signal handlers
669 * bug fix: fix deadlocks related to hash tables
670 * bug fix: fix deadlocks in pcl
672 changes in sbcl-1.0.25 relative to 1.0.24:
673 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
674 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
675 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
676 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
677 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
678 better name and does not return values so stale on multiprocessor systems.
679 Also, HOLDING-MUTEX-P was added for about the only sane usage of
681 * improvement: unithread builds keep track of MUTEX-VALUE.
682 * improvement: reading from a TWO-WAY-STREAM does not touch the output
683 stream anymore making it thread safe to have a concurrent reader and
684 a writer, for instance, in a pipe.
685 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
686 arguments, making compiler-macros for SETF-functions able to inspect
687 their constant arguments.
688 * improvement: COMPILE-FILE reports times with millisecond accuracy
689 (thanks to Luis Oliveira)
690 * optimization: CHAR-CODE type derivation has been improved, making
691 TYPEP elimination on subtypes of CHARACTER work better. (reported
692 by Tobias Rittweiler, patch by Paul Khuong)
693 * bug fix: setting alien structure fields of type struct by value now
694 computes the right offset for the memory copy.
695 * bug fix: compilation problem involving inlined calls to aliens with
696 result type VOID. (reported by Ken Olum)
697 * bug fix: #235a; sequential inline expasion in different policies no
698 longer reuses the functional from the previous expansion site.
699 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
700 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
701 * bug fix: named ENUMs in multiply-referenced alien record types no longer
702 cause an implied type redefinition. Regression from 1.0.21.29.
703 * bug fix: sign-extension of small signed return values now works with C
704 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
705 * improvements to the Windows port:
706 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
707 Westfold, thanks to Rudi Schlatte)
709 changes in sbcl-1.0.24 relative to 1.0.23:
710 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
711 vector of a multidimensional SIMPLE-ARRAY.
712 * new feature: the system now signals a continuable error if standard
713 readtable modification is attempted. (thanks to Tobias Rittweiler)
714 * new feature: DIRECTORY has been extended with a non-standard keyword
715 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
716 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
717 has been robustified and documented better. (thanks to Robert Goldman)
718 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
719 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
720 * tradeoff: constant FORMAT control strings are now compiled unless
721 SPACE > SPEED (previously only when SPEED > SPACE.)
722 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
724 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
726 * bug fix: lexical type declarations are now correctly reported by
727 SB-CLTL2. (reported by Larry D'Anna)
728 * bug fix: STRING-TO-OCTETS did not handle :START properly when
729 using UTF-8 as external format. (thanks to Luis Oliveira)
730 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
731 are signalled correctly. (thanks to Stas Boukarev)
732 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
733 designator. (thanks to Tobias Rittweiler)
734 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
735 designator, and returns T instead of the function. (thanks to
737 * bug fix: direct superclasses of STANDARD-CLASS and
738 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
739 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
740 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
741 when no class with the named slot yet exists no longer causes a
742 compile-time style-warning.
743 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
744 in safe code. (reported by Didier Verna)
745 * bug fix: #430; nested structure constructors can stack allocate.
746 * bug fix: on some 64-bit platforms dynamic space size was truncated
747 to #xffff0000 bytes. (reported by Benjamin Lambert)
748 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
749 REPL unusable. (reported by Daniel Herring)
750 * bug fix: return values of READ-SEQUENCE did not take :START into
751 account on file streams, regressions since 1.0.12.22. (reported by
752 Thomas Russ, patch by Paul Khuong)
753 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
754 method specializer used to confuse permuation vector optimization.
755 * bug fix: system inserted bogus implicit type declarations for local
756 special variables in DEFMETHOD bodies.
757 * bug fix: #354; duplicated frames in backtraces due to
758 non-tail-call-optimized XEPs to functions with return type NIL
760 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
761 initializes structure object slots according to DEFSTRUCT initforms,
762 and DEFSTRUCT forms :INCLUDEind structure classes defined using
763 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
764 (reported by Bruno Haible and Stephen Wilson)
765 * bug fix: #395; fill-pointer output streams used now support
766 element-type BASE-CHAR as well.
767 * bug fix: compiler error when attempting to derive return value of
768 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
770 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
772 * bug fix: Mac OS X binaries should now be portable between Leopard
775 changes in sbcl-1.0.23 relative to 1.0.22:
776 * enhancement: when disassembling method functions, disassembly
777 for the associated fast function is also produced.
778 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
780 * optimization: printing with *PRINT-PRETTY* true is now more
781 efficient as long as the object being printed doesn't require
782 special handling by the pretty printer.
783 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
784 now interact correctly with type declarations.
785 * partial bug fix: PCL detects infinite recursion during wrapper
786 validation. (thanks to Attila Lendvai)
787 * bug fix: #426; nested function calls are inlined properly.
788 Previously if FOO was an inline function, in calls of the form
789 (FOO (FOO ...)) the outer call was not inlined.
790 * bug fix: long long arguments passed in registers to alien
791 functions on PowerPC Linux and NetBSD are now handled correctly
792 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
794 changes in sbcl-1.0.22 relative to 1.0.21:
795 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
796 for the shared object in the current directory, but passes the native
797 namestring of the designated pathname to the operation system's shared
798 object loading function as-it.
799 * minor incompatible change: --disable-debugger toplevel option now takes
800 effect before processing of initialization files and --eval or --load
802 * new feature: new commandline argument: --script, which supports
803 shebang lines. See documentation for details. (based on work by
805 * new feature: SAVE-LISP-AND-DIE can save current values of
806 --dynamic-space-size and --control-stack-size in the executable core,
807 causing it to skip normal runtime option processing. See documentation
808 for details. (thanks to Zach Beane)
809 * enhancement: inoccous calls to EVAL or generic functions dispatching
810 on subclasses of eg. STREAM no longer cause compiler notes to appear.
811 * enhancement: the system no longer resignals errors from --load and
812 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
813 associated with the original error to be lost. (thanks to Ariel
815 * enhancement: :DONT-SAVE keyword argument has been added to
816 LOAD-SHARED-OBJECT for controlling interaction with
817 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
818 the effects of an earlier LOAD-SHARED-OBJECT call.
819 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
820 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
821 provided. (reported by Cedric St-Jean)
822 * bug fix: circularity handling in the reader did not treat raw
823 structure slots correctly. (reported by Cedric St-Jean)
824 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
825 file descriptors when there were none.
826 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
827 was broken. (reporteed by Willem Broekema)
828 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
829 pathnames without a directory.
830 * bug fix: COMPILE-FILE did not warn about undefined variable
831 references at toplevel, and LOAD of the resulting fasl did
833 * bug fix: functions with non-required arguments used to end up with
834 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
835 * bug fix: redefining a function with non-required arguments didn't
836 update the system's knowledge about its call signature properly.
837 * bug fix: fixed #431; incompatible alien record type redefinitions
838 are detected and handled. (reported by Neil Haven)
839 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
840 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
842 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
843 after alien stack frames.
844 * bug fix: strings passed to mk{s,d}temp lacked null termination
846 changes in sbcl-1.0.21 relative to 1.0.20:
847 * new feature: the compiler is able to track the effective type of a
848 generic function across method addition and removal.
849 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
850 appropriately typed structure slots without locking.
851 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
852 information like those gathered by TIME using a programming-friendly
854 * new feature: TIME reports time taken even if the form performs a
855 non-local transfer of control.
856 * enhancement: reduced conservativism on GENCGC platforms: on
857 average 45% less pages pinned (measured from SBCL self build).
858 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
859 mutate constant symbols or violate declaimed type of the symbol.
860 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
861 sleeping on the semaphore in heavily contested semaphores.
862 * bug fix: semaphores and condition variables were not interrupt
864 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
865 owned by other threads anymore.
866 * bug fix: FIND on lists called KEY outside the specified
867 subsequence. (reported by budden)
868 * bug fix: LOG doesn't use single-float intermediate results when
869 given mixed integer and double-float arguments, leading to better
870 precision. (reported by Bob Felts)
871 * bug fix: LOG with base zero returned values of inconsistent type.
873 changes in sbcl-1.0.20 relative to 1.0.19:
874 * minor incompatible change: OPTIMIZE qualities
875 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
876 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
877 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
878 * documentation: some slot access efficiency guidelines have been
879 added to the user manual.
880 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
881 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
882 as ASSOC and MEMEBER.
883 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
884 to more efficient EQ-comparison versions more often.
885 * optimization: enhanced derivation of DOLIST iteration variable type
887 * optimization: constant folding of simple (LIST ...) forms as DOLIST
889 * optimization: runtime lookup of function definitions can be
890 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
891 * optimization: compiler is able to derive the return type of
892 (AREF (THE STRING X) Y) as being CHARACTER.
893 * optimization: CLRHASH on empty hash-tables no longer does pointless
894 work. (thanks to Alec Berryman)
895 * optimization: REPLACE deftransforms don't punt when :START1 or
897 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
898 no longer emits a code deletion note. (reported by Andrew
900 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
901 functions with an odd number of &OPTIONAL arguments, a &REST
902 argument, and one or more &KEY arguments at the call site.
903 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
904 arguments appeared at call sites as well.
905 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
906 READ-LINE &co can no longer read from them afterwards. (reported
908 * bug fix: fixed #427: unused local aliens no longer cause compiler
909 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
911 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
912 alien-stack leakage. (reported by Andy Hefner)
913 * bug fix: PROGV signals an error when an attempt to violate declared
914 type of a variable or bind a constant is made.
915 * bug fix: SET signals an error when an attempt to violate declared
916 type of a variable is made.
917 * bug fix: restart computation during the execution of a restart
918 test function no longer causes infinite recursion. (reported by
920 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
921 signals an error. (thanks to Pierre Mai)
922 * bug fix: EXPORT left symbol unexported in conflict situations.
923 (thanks to Michael Weber)
924 * bug fix: correctly handle name conflicts involving CL:NIL.
925 (thanks to Michael Weber)
926 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
927 conflicts arising from USEing package with conflicting symbols
928 correctly. (thanks to Michael Weber)
930 changes in sbcl-1.0.19 relative to 1.0.18:
931 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
932 warnings that go otherwise unhandled will be muffled if they are
933 of the type that's the value of this variable.
934 * optimization: stack allocation is slightly more efficient on x86
936 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
937 where a raw slot always is initialized using the initform whose
938 type is not know sufficiently well a compile-time are now compiled
939 correctly. (reported by John Morrison)
940 * bug fix: compiler no longer makes erronous assumptions in the
941 presence of non-foldable SATISFIES types.
942 * bug fix: stack analysis missed cleanups of dynamic-extent
943 arguments in non-let-converted calls to local functions.
944 * improvements to the Windows port:
945 ** adjusted address spaces for building on both Win32 and
946 Win64. (thanks for John Connors)
947 * fixed some bugs revealed by Paul Dietz' test suite:
948 ** interval arithmetic during type derivation used inexact integer
949 to single-float coercions.
950 ** arithmetic operations involving large integers and single
951 floats give the same results in compiled and interpreted code.
952 ** deriving the result type of COERCE no longer signals an error
953 if the derived type of the second argument is a MEMBER type
954 containing invalid type specifiers.
955 ** ADJOIN with constant NIL as second argument works correctly.
956 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
958 changes in sbcl-1.0.18 relative to 1.0.17:
959 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
960 profiles only the current thread.
961 * minor incompatible change: changes to SYMBOL-VALUE of constants
962 defined with DEFCONSTANT now signal an error.
963 * enhancement: SB-SPROF now has support for wallclock profiling,
964 and is also able to profile specific threads. REPORT output
965 has also additional sorting options.
966 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
968 * optimization: structure allocation has been improved
969 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
970 ** out of line constructors are ~10% faster.
971 ** inline constructors are ~15% faster.
972 ** inline constructors are capable of dynamic extent allocation
973 (generally on x86 and x86-64, in some cases on other platforms
975 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
977 * optimization: file compiler is now able to coalesce non-circular
978 lists, non-base strings, and bit-vectors. Additionally, constants
979 are never referenced through SYMBOL-VALUE at runtime.
980 * optimization: code defining methods on PRINT-OBJECT (and other
981 generic functions in the COMMON-LISP package) now loads faster.
982 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
984 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
985 fasl is now deleted. (reported by Attila Lendvai)
986 * bug fix: READ-LINE always returned NIL for the last line in files.
987 (reported by Yoshinori Tahara)
988 * bug fix: more accurate disassembly annotations of foreign function
989 calls. (thanks to Andy Hefner)
990 * bug fix: trimming non-simple strings and non-string string
991 designators when the there is nothing to trim works properly.
992 (thanks to James Knight)
993 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
994 (thanks to Travis Cross)
995 * fixed some bugs revealed by Paul Dietz' test suite:
996 ** NIL is a valid function name (regression at 1.0.13.38)
997 ** FILL on lists was missing its return value (regression at 1.0.12.27)
998 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
999 fill pointers properly (regression at 1.0.12.23)
1000 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
1001 displacement indices properly (regression at 1.0.12.23)
1003 changes in sbcl-1.0.17 relative to 1.0.16:
1004 * temporary regression: user code can no longer allocate closure
1005 variable storage on stack, due to bug 419 without explicitly
1006 requesting it. Please consult sbcl-devel for advice if you need to
1007 use this feature in the meanwhile.
1008 * new feature: runtime argument --control-stack-size can be used to
1009 adjust thread default control stack size.
1010 * enhancement: improved TIME output
1011 ** all times are reported using the measured accuracy (milliseconds
1012 for real and GC times, microseconds for everything else.)
1013 ** processor cycle counts on x86 and x86-64.
1014 ** interpreted forms are counted for both evaluator modes.
1015 ** number of lambdas converted by the compiler is reported.
1016 ** CPU percentage report (computed from real and total run time.)
1017 ** more comprehensive run time reporting, using a condenced format
1018 ** interperted form, lambda, and page fault counts are omitted
1020 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
1022 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
1023 * optimization: two argument forms of LAST are upto ~10% faster
1024 in normal SPEED policies.
1025 * optimization: NCONC no longer needs to heap cons its &REST list
1026 in normal SPEED policies.
1027 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
1028 to Sidney Markowitz)
1029 * bug fix: UNION and NUNION work with :TEST-NOT once more,
1030 regression since 1.0.9.1. (thanks to Eric Marsden)
1031 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
1032 since 1.0.15.36. (thanks to Paul Khuong)
1033 * bug fix: LAST when always returned the whole list when given a bignum
1034 as the second argument.
1035 * bug fix: dynamic extent allocation of nested lists and vectors
1036 could leak to otherwise accessible parts.
1037 * bug fix: invalid optimization of heap-allocated alien variable
1039 * bug fix: fasl header checking is less vulnerable to different
1040 platform word lengths.
1041 * bug fix: more correct assembler syntax for GNU binutils
1042 2.18.50.0.4 support. (thanks to Marijn Schouten)
1043 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
1045 * bug fix: the fopcompiler can handle LOCALLY forms (with no
1046 declarations) successfully. (reported by Attila Lendvai)
1048 changes in sbcl-1.0.16 relative to 1.0.15:
1049 * minor incompatible change: revert the changes to sb-posix's error
1050 signaling added in 1.0.14.
1051 * minor incompatible change: change PROBE-FILE back to returning
1052 NIL whenever we can't get a truename, as was the case before 1.0.14.
1053 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
1054 inherits from ERROR instead of just CONDITION.
1055 * new feature: SB-INTROSPECT can provide source locations for instances
1056 as well. (thanks to Tobias Rittweiler)
1057 * optimization: binding special variables now generates smaller code
1058 on threaded platforms.
1059 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
1060 and cases where no :TEST is given but the compiler can infer that
1061 the element to search is of type (OR FIXNUM (NOT NUMBER)).
1062 * optimization: better LOGNOT on fixnums.
1063 * optimization: modular arithmetic for a particular requested width
1064 is implemented using a tagged representation unless a better
1065 representation is available.
1066 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
1067 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
1068 for wellformedness and returns a specialized vector. (reported by
1069 Francois-Rene Rideau)
1070 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
1071 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
1072 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
1073 unscheduling timer at the same time as another timer fires.
1074 * bug fix: don't reschedule timers for dead threads.
1075 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
1076 * bug fix: copying output from RUN-PROGRAM to a stream signalled
1077 bogus errors if select() was interrupted.
1078 * enhancement: add support for fcntl's struct flock to SB-POSIX.
1080 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
1081 * enhancement: cleaner backtraces for interactive interrupts, as
1082 well as other cases where the interesting frames used to be
1083 obscured by interrupt handling frames.
1084 * enhancement: untracing a whole package using (UNTRACE "FOO") is
1085 now supported, and tracing a whole package using (TRACE "FOO") now
1086 traces SETF-functions as well.
1087 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
1088 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
1089 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
1090 * unreadably printed representation of hash-tables now includes
1092 * bug fix: partially fixed #188: type propagation from assignments
1093 is now more efficient.
1094 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
1095 'DOUBLE-FLOAT) are not flushable.
1096 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
1097 full address of the object, and none of the tag bits.
1098 * bug fix: readably printing hash-tables now respects other printer
1099 control variables. (reported by Cedric St-Jean)
1100 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
1101 keyword with MAKE-HASH-TABLE.
1102 * bug fix: export SB-POSIX:MKSTEMP.
1103 * bug fix: SORT was not interrupt safe.
1104 * bug fix: XREF accounts for the last node of each basic-block as
1106 * bug fix: MAKE-INSTANCE optimizations interacted badly with
1107 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
1108 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
1109 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
1110 create obsolete instances in the case of redefinition or
1111 obsoletion of a superclass. (thanks to Andy Hefner)
1112 * bug fix: Support for the Alpha architecture has been revived; it had
1113 suffered somewhat from lack of maintenance since sbcl-1.0.
1114 * improvements to the Windows port:
1115 ** The system detects the codepage to use at startup, instead of
1116 using the value from the saved core. (thanks to Kei Suzuki)
1118 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
1119 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
1120 (see documentation for details.)
1121 * revived support for OpenBSD (contributed by Josh Elsasser)
1122 * partially fixed bug #108: ROOM no longer suffers from occasional
1123 (AVER (SAP= CURRENT END)) failures.
1124 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
1125 bodies no longer cause a WARNING to be signalled. (reported by
1127 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
1128 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
1129 the bug in TRUNCATE.) (reported by Michael Weber)
1130 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
1131 no samples. (reported by Andy Hefner)
1132 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
1133 no longer appear as (NIL ...) frames in backtraces.
1134 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
1135 is now actually exported from SB-EXT as documented. (reported by
1137 * bug fix: sb-aclrepl now correctly understands how to inspect
1138 single-floats on 64-bit platforms where single-floats are not boxed.
1139 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
1140 yet been finalized. (reported by Levente Meszaros)
1141 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
1142 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
1143 SBCL-specific optimize qualities.
1145 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
1146 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
1147 an executable in the search path, and does so in the child
1148 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
1149 been removed; it can be found in the CVS history, for somebody who
1150 needs that search behavior (see the manual).
1151 * minor incompatible change: compiler policy re. weakening type
1152 checks has changed: now type checks are weakened only if SAFETY < 2
1154 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
1155 unparsing of directory pathnames as files. Analogously,
1156 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
1157 filename to parse into a directory pathname.
1158 * enhancement: implicit generic function creation now signals a
1159 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
1160 users can bind handlers for and muffle around calls to LOAD.
1161 * enhancement: RUN-PROGRAM allows unicode arguments and environments
1162 to be used (using the default stream external format), and allows
1163 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
1164 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
1165 strings and vectors whose element-type or simplicity is not fully
1166 known at compile-time.
1167 * optimization: STRING-TRIM and related functions no longer allocate
1168 a new string when no trimming needs to be performed. These functions
1169 are also faster than before when the input string has been declared
1171 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
1172 * optimization: READ-LINE is significantly faster for files containing
1174 * optimization: non-open coded uses of character comparison operators
1175 (e.g. char=) no longer cons when called with more than one parameter
1176 on platforms supporting dynamic-extent allocation.
1177 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
1178 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
1179 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
1181 * bug fix: some sequence functions elided bounds checking when
1183 * bug fix: too liberal weakening of union-type checks when SPEED >
1185 * bug fix: more bogus fixnum declarations in ROOM implementation
1188 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
1189 * new feature: MAKE-HASH-TABLE now experimentally accepts a
1190 :SYNCHRONIZED argument, which makes the hash-table safe for
1191 concurrent accesses (but not iteration.) See also:
1192 SB-EXT:WITH-LOCKED-HASH-TABLE, and
1193 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
1194 * optimization: CONCATENATE on strings is an order of magnitude faster
1195 in code compiled with (> SPEED SPACE).
1196 * optimization: SUBSEQ is ~50% faster on lists.
1197 * bug fix: bug 417 fixed -- source location reporting is now more robust.
1198 * bug fix: SUBSEQ on a list will now correctly signal an error if if
1199 END is smaller then START.
1200 * bug fix: SB-PROFILE will no longer report extra consing for nested
1201 calls to profiled functions.
1202 * bug fix: ROOM implementation had bogus fixnum declarations which
1203 could cause type-errors when calling ROOM on large images.
1204 * bug fix: if file compilation is aborted, the partial fasl is now
1205 deleted, and COMPILE-FILE returns NIL as the primary value.
1206 * bug fix: number of thread safety issues relating to SBCL's internal
1207 hash-table usage have been fixed.
1208 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
1209 be returned from its body when the values were being returned
1210 using unknown-values return convection and the W-P-O was wrapped
1211 inside an UNWIND-PROTECT.
1212 * bug fix: sb-posix should now compile again under Windows, enabling
1213 slime to work again.
1215 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
1216 * incompatible change: hash-table accessor functions are no longer
1217 automatically protected by locks. Concurrent accesses on the same hash-table
1218 from multiple threads can give inconsistent results or even corrupt the
1219 hash-table completely. Multi-threaded applications should do their own
1220 locking at the correct granularity. In the current implementation it is
1221 still safe to have multiple readers access the same table, but it's not
1222 guaranteed that this property will be maintained in future releases.
1223 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
1224 and will signal an error at runtime.
1225 * enhancement: SB-THREAD package now exports a semaphore interface.
1226 * enhancement: CONS can now stack-allocate on x86 and
1227 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
1229 * enhancement: nested lists can now be stack allocated on
1230 platforms providing stack allocation support.
1231 * enhancement: dynamic-extent support has been extended to support
1232 cases where there are multiple possible sources for the stack
1234 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
1235 if the mutex is uncontested on Linux.
1236 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
1237 as specified by the CLHS. (thanks to Tobias Rittweiler)
1238 * bug fix: NaN comparison now works on x86-64.
1239 * bug fix: CLOSE :ABORT T on a stream with pending output now
1241 * bug fix: instances of non-standard metaclasses using standard
1242 instance structure protocol sometimes missed the slot type checks
1244 * bug fix: known functions can be inlined in high-debug code.
1246 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
1247 * minor incompatible change: the MSI installer on Windows no longer
1248 associates .lisp and .fasl files with the installed SBCL.
1249 * minor incompatible change: :UNIX is no longer present in *FEATURES*
1250 on Windows. (thanks to Luis Oliviera)
1251 * new feature: SB-CLTL2 contrib module now implements
1252 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
1253 * optimization: scavenging weak pointers is now more efficient,
1254 requiring O(1) instead of O(N) per weak pointer to identify
1255 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
1256 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
1257 method bodies using constant slot-names when the first argument is
1258 a specializer parameter for the method.
1259 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
1260 names, when the class of the instance is a direct instance
1261 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
1262 as slow as the constant slot-name case.)
1263 * optimization: member type construction is now O(N) instead
1265 * optimization: UNION and NUNION are now O(N+M) for large
1266 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
1268 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
1269 now more readable in environments like Slime which display it.
1270 (thanks to Tobias C. Rittweiler)
1271 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
1272 was able to infer the correct argument type for the object on which
1273 the CAS operation was being performed.
1274 * bug fix: copy propagation interfered with parallel assignment
1275 semantics in local calls. (reported by Paul Khuong)
1276 * bug fix: the signed modular fixnum shift compiled to wrong code on
1277 x86 and x86-64. (spotted by a slight modification to some of
1280 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
1281 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
1282 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
1283 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
1285 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
1286 and generic function now print the object identity for anonymous
1287 instances. (thanks to Attila Lendvai)
1288 * enhancement: as an extension to MOP, SBCL now supports SETF of
1289 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
1290 (thanks to Attila Lendvai)
1291 * enhancement: sb-cover has an alternate annotation mode which puts
1292 the form annotations at the car of the form, instead of annotating
1293 it completely. The benefit of this mode is that it shows better
1294 which forms were instrumented by the compiler.
1295 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
1296 by printer control variables. (reported by Dan Corkill)
1297 * bug fix: system leaked memory when delayed output was performed by
1298 the OS in smaller chunks then expected. (thanks to David Smith)
1299 * bug fix: system leaked memory when file streams were not closed
1301 * bug fix: large objects written to slow streams that were modified
1302 after the write could end up with the modified state written to
1303 the underlying file descriptor.
1304 * bug fix: multiple threads operating in parallel on the same stream
1305 could cause buffer-overflows.
1306 * bug fix: source location information is stored correctly
1307 (broken since 1.0.6). This bug would generally show up as the
1308 Slime debugger highlighting the wrong form when the sldb-show-source
1310 * bug fix: Branch forms are again annotated as branches in the sb-cover
1312 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
1313 on x86-64. (reported by Gregory Vanuxem)
1314 * bug fix: EQUALP could return wrong results for structures with raw
1315 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
1316 word). (reported by Vjacheslav Fyodorov)
1317 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
1320 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
1321 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
1322 atomic compare-and-swap operations on threaded platforms.
1323 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
1324 allows assigning a global minimum value to optimization qualities
1325 (overriding proclamations and declarations).
1326 * enhancement: closed over variables can be stack-allocated on x86
1328 * performance bug fix: GETHASH and (SETF GETHASH) are once again
1330 * optimization: slot definition lookup is now O(1). This speeds up
1331 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
1332 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
1333 * optimization: ASSOC and MEMBER can now be open-coded for all
1334 combinations of keyword arguments when second argument is constant
1335 and SPEED >= SPACE. In other cases a specialized version is
1337 * bug fix: using obsoleted structure instances with TYPEP and
1338 generic functions now signals a sensible error.
1339 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
1340 (reported by Kristoffer Kvello)
1341 * bug fix: backtrace construction is now more careful when making
1342 lisp-objects from pointers on the stack, to avoid creating bogus
1343 objects that can be seen by the GC.
1344 * bug fix: defaulting of values in contexts expecting more than 7
1345 variables now works on x86-64. (reported by Christopher Laux)
1346 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
1348 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
1349 as the property-list of a symbol.
1350 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
1351 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
1352 situations requested, are once again file-compileable. (reported
1355 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
1356 * MOP improvement: support for user-defined subclasses of
1357 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
1358 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
1359 the expansion of DEFMETHOD, is responsible for generating a form
1360 which creates a list of specializers when evaluated. Additional
1361 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
1362 debugging and introspective support.
1363 * minor incompatible change: the (unsupported) spinlock interface
1364 has changed: free spinlock now has the value NIL, and a held spinlock
1365 has the owning thread as its value.
1366 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
1367 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
1369 * enhancement: name of a socket-stream is now "a socket" instead of
1370 "a constant string".
1371 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
1372 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
1373 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
1374 for generic functions. (Thanks to Tobias C. Rittweiler)
1375 * optimization: bignum printing speed has been improved by 20-40%
1376 (depending on the bignum size.)
1377 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
1379 * bug fix: the cache used by the CLOS to store precomputed effective
1380 methods, slot offsets, and constant return values is now thread and
1382 * bug fix: generic function dispatch function updating is now thread
1383 and interrupt safe (in the sense that the known issues have been
1385 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
1386 * bug fix: interning EQL-specializers is now thread and interrupt safe.
1387 * bug fix: asdf systems with dependencies to the SB-POSIX or
1388 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
1389 * bug fix: interrupt safety of applicable method computation has been
1392 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
1393 * new contrib: sb-cover, an experimental code coverage tool, is included
1394 as a contrib module.
1395 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
1396 significantly faster.
1397 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1398 produce more efficient code on x86-64 if the compiler can determine
1399 that (+ INDEX <constant>) does not require a bounds check and FOO
1400 has an element type at least 8 bits wide.
1401 * enhancement: a new, experimental synchronous timeout facility is
1402 provided. Refer to SB-SYS:WITH-DEADLINE for details.
1403 * enhancement: when a symbol name conflict error arises, the
1404 conflicting symbols are always printed with a package prefix.
1405 (thanks to Kevin Reid)
1406 * enhancement: stepping is now once again supported on the SPARC and
1407 MIPS platforms. (It is also now more likely to work on CheneyGC
1409 * enhancement: sb-sprof can now also track and report accurate call
1411 * bug fixes: the treatment of non-standard subclasses of
1412 SB-MOP:SPECIALIZER is more correct.
1413 * incompatible change: PURIFY no longer copies the data from the
1414 dynamic space into the static and read-only spaces on platforms
1415 that use the generational garbage collector
1416 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
1418 * bug fix: GC race condition occasionally resulting in crashes with
1419 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
1421 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
1423 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
1424 a STYLE-WARNING to be signalled (regression from 1.0.4.)
1425 * bug fix: an asynchronous interrupt could previously leave the
1426 system running with GC inhibited.
1427 * bug fix: a DECLARE form evaluated at top-level now causes an error
1428 rather than silently (or verbosely) returning NIL.
1429 * bug fix: trying to dissassemble functions compiled at high DEBUG could
1430 result in a error being signalled due to source form lookup errors.
1431 (reported by Peter Graves)
1433 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
1434 * incompatible change: removed writer methods for host-ent-name,
1435 host-ent-addresses -- changing the values did not update the DNS
1437 * minor incompatible change: changed experimental JOIN-THREAD interface
1438 * documentation: the manual now lists reader and writer methods
1439 in class slot documentation sections. (thanks to Richard M Kreuter)
1440 * documentation: unwinding from asyncronous events has been
1441 documented as unsafe.
1442 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
1443 in multithreaded application code.
1444 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
1445 platforms. (thanks to James Anderson for the optimization hint)
1446 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
1448 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
1449 variants no longer cons.
1450 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
1451 their NOT- variants no longer cons.
1452 * optimization: EQUAL hash tables no longer use SXHASH for objects
1453 of all data types, but instead use an EQL hash for types for which
1454 EQUAL is the same as EQL.
1455 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
1456 are significantly faster.
1457 * optimization: new STRING-OUTPUT-STREAM implementation is much
1458 faster for large outputs and conses 30% less on average.
1459 * enhancement: XREF information is now collected for references made
1460 to global variables using SYMBOL-VALUE and a constant argument.
1461 * enhancement: SIGINT now causes a specific condition
1462 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
1463 * bug fix: FILE-POSITION used to signal an error in some cases where
1464 ANSI requires it to return NIL.
1465 * bug fix: ADJUST-ARRAY is now interrupt-safe.
1466 * bug fix: adding and removing fd-handlers is now interrupt-safe.
1467 * bug fix: inlined calls to C now ensure 16byte stack alignment on
1469 * bug fix: bad type declaration in the CLOS implementation has
1470 been fixed. (reported by James Anderson)
1471 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
1472 (thanks to Lutz Euler)
1473 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
1474 required by the CLHS. (reported by Samium Gromoff)
1475 * bug fix: dead unbound variable references now signal an error.
1476 (reported by Marco Monteiro)
1477 * bug fix: / with an unused value was being deleted in safe code.
1478 (thanks to Marco Monteiro and Kevin Reid)
1479 * bug fix: number of characters that can be written onto a single
1480 line in a file is unlimited.
1481 * bug fix: some GC deadlocks caused by asynchronous interrupts have
1482 been fixed by inhibiting interrupts for when GC is disabled.
1483 * bug fix: some interrupt safety issues with GETHASH, (SETF
1484 GETHASH), CLRHASH and REMHASH have been fixed.
1485 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
1486 specifier no longer causes infinite recursion.
1487 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
1488 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
1489 * bug fix: an &environment argument with the correct variable information
1490 is passed to macros that are expanded during byte compilation.
1491 (reported by Samium Gromoff)
1492 * bug fix: a logic error was causing the hash function for bignums to
1493 have excessive amounts of collisions. (reported by Faré Rideau)
1494 * bug fix: modifying the contents of an array could change the return
1495 value of SXHASH on that array, which is only allowed for strings
1496 and bit vectors. (bug introduced in 0.9.16)
1497 * improvement: the x86-64/darwin port now passes all tests (except
1498 for the debugger tests) but should still be considered
1499 experimental until this is fixed.
1500 * improvement: a style-warning is signaled for CASE (etc) clauses with
1501 duplicate keys. (thanks to Kevin Reid)
1502 * improvement: macos/x86 and macos/x86-64 now use mach exception
1503 handlers for dealing with illegal instructions (for trapping and
1504 error handling) and memory protection violations (for GC).
1506 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
1507 * new platform: experimental support for x86-64/darwin (MacOS).
1508 * incompatible change: the thread-safe (on most platforms) getaddrinfo
1509 and getnameinfo sockets functions are used instead of gethostbyaddr
1510 and gethostbyname, on platforms where the newer functions are available.
1511 As a result, the ALIASES field of HOST-ENT will always be NIL on these
1513 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
1514 don't print the startup banner, but behave as if --noinform was passed
1515 as a command line argument. (thanks to Kevin Reid)
1516 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1517 * optimization: code using alien values with undeclared types is much faster.
1518 * optimization: the compiler is now able to open code SEARCH in more cases.
1519 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1520 * bug fix: using standardized COMMON-LISP special variables as loop
1521 variables no longer signals bogus package lock violations. (reported
1523 * bug fix: declaring local loop variables to be of a range-limited type
1524 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1525 (reported by Andras Simon)
1526 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1527 bugs remain on x86-64.)
1528 * bug fix: the #= and ## reader macros now interact reasonably with
1529 funcallable instances.
1530 * bug fix: type-checks for function arguments were compiled using the
1531 compiler policy settings of the wrong lexical environment (in 1.0.2
1533 * bug fix: SHADOW accepts characters as string designators, as required
1534 by the spec (thanks to Eric Marsden)
1535 * bug fix: fixed GC safety issues when foreign functions are called with
1536 non-base strings as arguments
1537 * bug fix: more consistent error output for fatal-compiler-warnings, like
1539 * bug fix: functions with proper names were showing up as NIL in
1540 backtraces (reported by Edi Weitz, regression in 1.0.2)
1542 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1543 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1544 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1545 produce more efficient code on the x86 if the compiler can determine
1546 that (+ INDEX <constant>) does not require a bounds check and FOO
1547 has an element type at least 8 bits wide.
1548 * bug fix: references to undefined variables are handled the same way
1549 in toplevel forms as in the normal compiler
1550 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1552 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1553 (thanks to Jon Buller)
1554 * improvement: faster compilation times for complex functions
1555 * improvement: added readlink support to SB-POSIX (thanks to Richard
1558 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1559 * improvement: experimental support for mach exception handling on
1560 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1562 * improvement: support for GBK external format.
1563 (thanks to Chun Tian (binghe))
1564 * improvement: the debugger now displays variables that have been closed
1565 over, in code compiled with (DEBUG 2) or higher
1566 * improvement: support for executable cores on NetBSD (thanks to
1568 * new feature: added a RESTART-FRAME debugger command
1569 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1570 be used to provide an implementation for FILE-POSITION on Gray streams
1571 (thanks to Eric Marsden)
1572 * new feature: add syslog, openlog and closelog support to SB-POSIX
1573 (thanks to Richard Kreuter)
1574 * optimization: the function call overhead in code compiled with
1575 a high DEBUG optimization setting is significantly
1576 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1577 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1578 (thanks to Tony Martinez)
1579 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1580 evaluated once (reported by Kevin Reid)
1581 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1582 works again (thanks to Kilian Sprotte)
1583 * bug fix: an error is signaled for tagbodies with duplicate tags
1584 (thanks to Stephen Wilson)
1585 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1586 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1587 stack frames from alien callbacks.
1588 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1589 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1590 65k pages (thanks to David Woodhouse)
1591 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1593 changes in sbcl-1.0.1 relative to sbcl-1.0:
1594 * new platform: FreeBSD/x86-64, including support for threading.
1595 * new feature: the compiler stores cross-referencing information
1596 about function calls (who-calls), macroexpansion (who-macroexpands)
1597 and special variables (who-binds, who-sets, who-references) for code
1598 compiled with (< SPACE 3). This information is available through the
1599 sb-introspect contrib.
1600 * new feature: users may subclass SEQUENCE, and have instances of
1601 these classes interoperate with standard Common Lisp functions if
1602 a number of methods are defined. (This feature is experimental
1603 and the interface subject to change based on feedback from SBCL
1604 users and the general community)
1605 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1606 x86/x86-64, rather than the previous fixed depth of 8
1607 * improvement: another pthread back-end of mutex "pthread-futex".
1608 * bug fix: non-ascii command-line arguments are processed correctly
1609 (thanks to Yaroslav Kavenchuk)
1610 * bug fix: non-required arguments were not passed correctly when a method
1611 defined using DEFMETHOD was called from a mop-generated method using
1612 CALL-NEXT-METHOD (reported by Pascal Costanza)
1613 * bug fix: recursion is now permitted in accessors through
1614 SLOT-UNBOUND. (reported by Pascal Costanza)
1615 * bug fix: an error was signaled at startup if the HOME environment
1616 variable was defined, but had an empty value (reported by Peter Van Eynde)
1617 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1618 signal a PROGRAM-ERROR, not a BUG.
1619 * bug fix: SB-POSIX:READDIR works when built with large file support.
1620 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1622 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1623 for better automatic indentation support. (thanks to Matt Pillsbury)
1624 * optimization: loading generic functions no longer takes O(n^2) time,
1625 proportional to the amount of methods in the generic function
1626 (reported by Todd Sabin and Jeremy Brown)
1627 * optimization: the FIND and POSITION family of sequence functions
1628 are significantly faster on arrays whose element types have been
1630 * improvements to the Windows port:
1631 ** Intermittent heap corruption problems have been fixed. (thanks
1632 to Alastair Bridgewater)
1633 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1635 ** Lisp is able to unwind foreign exception frames from alien
1636 callbacks. (thanks to Alastair Bridgewater)
1638 changes in sbcl-1.0 relative to sbcl-0.9.18:
1639 * improvement: experimental support for threading on FreeBSD/x86.
1640 (thanks to NIIMI Satoshi)
1641 * improvement: runtime option --dynamic-space-size can be used
1642 to set the size of the dynamic space reserved on startup.
1643 * improvement: floating point modes in effect are now saved in
1644 core, and restored on startup.
1645 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1646 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1647 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1648 threads are not running after *SAVE-HOOKS* have run.
1649 * improvement: writes to CLOS instance slots are type-checked in code
1650 compiled with (SAFETY 3)
1651 * improvement: floating-point exception handling on FreeBSD (thanks to
1653 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1654 (thanks to Zach Beane)
1655 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1657 * improvement: added support for the Shift-JIS external format.
1658 (contributed by NIIMI Satoshi)
1659 * improvement: callbacks are supported on Linux/PPC. (thanks to
1661 * bug fix: compiler bug triggered by a (non-standard) VALUES
1662 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1663 * bug fix: file compiler no longer confuses validated and already
1664 dumped structurres. (reported by Kaersten Poeck)
1665 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1666 fill-pointer signals a type-error as required. (thanks to
1668 * bug fix: disassemly of funcallable instances works.
1669 * bug fix: single stepping on PPC.
1670 * bug fix: fix thread-safety problems in the type system (generally
1671 manifesting as nonsensical errors like "STRING is a bad type specifier
1672 for sequences" or "The value 1 is not of type FIXNUM")
1673 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1674 (reported by Josip Gracin)
1675 * bug fix: an error is signaled for attempts to displace arrays with
1676 incompatible element types (thanks to Mario Mommer)
1677 * bug fix: more correct handling of wide characters in debug info
1678 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1679 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1680 and don't cause extra consing
1681 * optimization: MAP and MAP-INTO are significantly faster on vectors
1682 whose elements types have been declared.
1683 * Improvements to SB-SPROF:
1684 ** Support for allocation profiling
1685 ** Reduced profiling overhead, especially for long profiling runs
1686 * Improvements to the Windows port:
1687 ** floating point exceptions are now reported correctly.
1688 ** stack exhaustion detection works partially.
1689 ** more accurate GET-INTERNAL-REAL-TIME.
1690 ** better RUN-PROGRAM behaviour in case of errors in the child
1692 ** PROBE-FILE now simplifies pathnames correctly.
1693 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1694 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1695 1st 1970 (although time timezone data will be unreliable).
1696 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1698 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1699 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1700 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1702 * bug fix: two potential GC deadlocks affecting threaded builds.
1703 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1704 returns T (reported by Anton Kazennikov)
1705 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1706 with non-variable places
1707 * bug fix: remove a race condition in the setting of
1708 funcallable-instance functions, this should make threaded CLOS
1709 code more stable against memory faults.
1710 * bug fix: corruption of specials when unbinding is interrupted by an
1711 asynchronous unwind (reported by Hannu Koivisto)
1712 * improvement: the debugger will now also display local variables that
1713 are only used once, for code compiled with a DEBUG optimization quality
1716 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1717 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1718 * incompatible change: External-format support for FFI calls. The
1719 SB-ALIEN:C-STRING no longer implies an ASCII
1720 external-format. Instead the string is subject to external-format
1721 conversion. Additionally return values of type SB-ALIEN:C-STRING
1722 are by default of type (SIMPLE-ARRAY CHARACTER), not
1723 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1724 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1725 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1726 * incompatible change: SB-EXT package no longer contains the
1727 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1728 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1730 * incompatible change: the single-stepper is no longer available
1731 on Alpha, Mips and Sparc platforms.
1732 * minor incompatible change: the direct superclasses of
1733 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1734 not (STANDARD-OBJECT FUNCTION). This makes the
1735 class-precedence-lists of GENERIC-FUNCTION and
1736 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1738 * minor incompatible change: the default stream external format on
1739 non-unicode SBCL is no longer detected from the locale, but is always
1741 * new feature: Add a version of evaluator that uses an interpreter instead
1742 of the compiler. EVAL still uses the compiler by default, to switch it
1743 to use the interpreter, set the value of the variable
1744 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1745 * minor incompatible change: the single-stepper REPL has been merged
1746 with the normal debugger (see the "Stepping" heading of the debugger help
1747 for more details). The debugger command STEP will no longer switch
1748 to the single-stepper REPL.
1749 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1750 for a type now works.
1751 * bug fix: loading of default sysinit file works. (thanks to Leonid
1753 * bug fix: better detection of circularities in the file-compiler.
1754 (reported by Marco Monteiro)
1755 * bug fix: the CL pathname functions now work with files that have
1756 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1757 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1758 systems with Unix98 pty semantics.
1759 * bug fix: ASDF-INSTALL will now work with bsd tar.
1760 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1762 * bug fix: timers expiring in dead threads no longer cause a
1763 type-error (reported by Paul "Nonny Mouse").
1764 * bug fix: thanks to more lightweight single-stepper instrumentation,
1765 code compiled with (DEBUG 3) will compile and execute significantly faster,
1766 and will have more accurate type-inferencing than before
1767 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1768 whose bindings are modified
1769 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1770 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1771 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1772 CL:SOFTWARE-VERSION with proper version
1774 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1775 * feature: implemented the READER-METHOD-CLASS and
1776 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1777 as specified by AMOP.
1778 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1780 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1781 (thanks to Lutz Euler)
1782 * optimization: hashing of general arrays and vectors has been
1783 improved. (reported by Any Fingerhut)
1784 * enhancement: SB-INTROSPECT is now able to find definitions of
1785 profiled functions. (thanks to Troels Henriksen)
1786 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1787 as well, allowing compiler-macros for SETF-functions to expand.
1788 * enhancement: step-instrumentation no longer wraps known
1789 single-value functions with multiple-value context, allowing
1790 better type inference.
1791 * fixed bug #337: use of MAKE-METHOD in method combination now works
1792 even in the presence of user-defined method classes. (reported by
1793 Bruno Haible and Pascal Costanza)
1794 * fixed bug #339(c): if there are applicable methods not part of any
1795 long-form method-combination group, call INVALID-METHOD-ERROR.
1796 (reported by Bruno Haible)
1797 * fixed bug #361: the :FUNCTION initarg in the protocol for
1798 initialization of methods can now be used to override
1799 internally-produced optimized functions. (reported by Bruno
1801 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1802 system-provided lambda expression no longer cause warnings about
1803 unbound #:|pv-table| symbols.
1804 * bug fix: improved the handling of type declarations and the
1805 detection of violations for keyword arguments with non-constant
1807 * bug fix: potentially erronous calls to PATHNAME and
1808 MERGE-PATHNAMES were being flushed in some cases.
1809 (reported by Richard Kreuter)
1810 * bug fix: compiled calls to TYPEP were mishandling obsolete
1811 instances. (reported by James Bielman and Attila Lendvai)
1812 * bug fix: format strings with ~> without matching ~< no longer
1813 trigger an AVER, but signal an understandable error instead.
1814 (reported by Antonio Martinez)
1815 * bug fix: specifying an output-file without a pathname-type for
1816 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1817 of input-file instead of "fasl". (reported by Robert Dodier)
1818 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1819 (reported by James Y Knight).
1820 * bug fix: compiler-macros-function did not consider the environment
1821 argument for shadowing by local functions.
1822 * bug fix: compiler-macros expansion was inhibited by local INLINE
1824 * bug fix: inline expansions of known functions were subject to
1825 step-instrumentation in high debug policies, leading to problems
1826 with type-inference.
1827 * bug fix: compiler failed to differentiate between different CONS
1828 types in some cases.
1829 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1830 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1831 * bug fix: #368: incorrect use of expressed vs. upgraded array
1833 * bug fix: #306a: more precise unions of array types.
1834 * thread-safety improvements:
1835 ** CONDITION-WAIT could return early on Linux, if the thread was
1836 interrupted and subsequently continued with SIGCONT.
1837 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1839 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1840 * added support for the ucs-2 external format. (contributed by Ivan
1842 * minor incompatible change: pretty printing of objects of type
1843 (cons symbol) is, in the default pprint-dispatch-table, now
1844 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1846 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1847 called later in a class's lifetime, possibly as late as when the
1848 first instance of the class is created. Previously,
1849 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1850 class became finalizeable.
1851 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1852 for multibyte external-format streams. (thanks to "vbzoli")
1853 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1854 with :ALLOCATION :CLASS inherited from superclasses of the
1856 * fixed bug: anonymous classes can now be created using the :NAME
1857 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1858 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1859 * fixed bug: core-files saved with :EXECUTABLE T can again be
1860 executed when SBCL_HOME isn't set. (reported by James Knight)
1861 * fixed bug: toplevel LOCALLY forms with declarations could
1862 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1863 * fixed bug: printing from several different threads using different
1864 values of *print-case* could cause invalid output, due to
1865 some internal special variables of the printer not being bound
1866 thread-locally (reported by Max Mikhanosha)
1867 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1868 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1869 (reported by Pascal Costanza)
1870 * fixed bug: FINALIZE-INHERITANCE is called from
1871 REINITIALIZE-INSTANCE on classes when the class has previously
1872 been finalized, as required by AMOP.
1873 * minor code generation optimizations:
1874 ** better register allocation in CLOS dispatching functions
1875 ** overflow detection when coercing signed bytes to fixnums on x86-64
1876 ** is now implemented with one IMUL instruction instead of three shifts
1877 ** more efficient bit-vector access on x86 and x86-64
1878 ** more efficient access to raw structure slots on x86 and x86-64
1879 * fixed some bugs revealed by Paul Dietz' test suite:
1880 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1881 return its argument.
1883 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1884 * feature: thread support on Solaris/x86, and experimental thread support
1886 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1888 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1889 default method for SLOT-UNBOUND.
1890 * minor incompatible change: prevent the user from specializing the
1891 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1892 somewhat counter to the intent of the protocol, I (CSR) think, and
1893 additionally it just doesn't work in SBCL as currently
1894 implemented, thanks to optimizations (that are always valid for
1895 the other three SLOT-VALUEish functions, but not for the setter).
1896 * minor incompatible change: the :SB-LDB feature is now enabled by
1897 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1898 the low-level debugger.
1899 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1900 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1901 down to an SB-SYS:FD-STREAM, allowing direct communication in
1902 more cases. Temporary files and pipes are still used as a fallback
1904 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1905 on the non-reentrant C functions localtime(3) and gmtime(3).
1906 * bug fix: global optimization policy was not visible in LOCALLY and
1908 * bug fix: class objects can be used as specializers in methods.
1909 (reported by Pascal Costanza)
1910 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1911 with a type error. (reported by blitz_ on #lisp)
1912 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1913 failed with a type error. (reported by Pascal Bourguignon)
1914 * bug fix: merging pathnames against defaults with :DIRECTORY
1915 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1916 (reported by James Y Knight)
1917 * bug fix: saving large (>2GB) cores on x86-64 now works
1918 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1919 constant index and a (simple-array (signed-byte 32)) array
1920 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1921 error (patch by Robert J. Macomber)
1922 * feature: TIME now displays the amount of run-time spent in GC
1923 * bug fix: The debugger now does a better job of respecting
1924 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1925 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1926 workaround for bug 403.)
1927 * bug fix: floating point exception handling now works on FreeBSD
1928 (thanks to NIIMI Satoshi)
1929 * fixed some bugs revealed by Paul Dietz' test suite:
1930 ** MISC.641: LET-conversion were not supposed to work in late
1932 * improvements to the Win32/x86 port:
1933 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1934 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1936 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1937 * new feature: source path information is generated for macro-expansion
1938 errors for use in IDE's like Slime (thanks to Helmut Eller)
1939 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1940 * bug fix: compiler does not loop forever on an invalid type in
1942 * improvement: compilation of most CLOS applications is significantly
1944 * optimization: added a limited bytecode compiler for simple toplevel
1945 forms, speeding up compilation and FASL loading
1946 * bug fix: the statistical profiler now properly distinguishes anonymous
1949 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1950 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1951 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1952 into subclasses of the finalized class. Now user calls to
1953 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1954 system return before any subclasses are finalized.
1955 * minor incompatible change: The reader no longer ignores errors
1956 regarding non-existent packages in #+ and #- feature tests.
1957 * new feature: command line options --no-sysinit, --no-userinit to
1958 inhibit loading the corresponding init files
1959 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1960 for sending data through UDP sockets (thanks to François-René Rideau)
1961 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1962 error is signalled instead (thanks to François-René Rideau)
1963 * minor incompatible change: Error signalling behaviour of lexical
1964 operations violating package locks has changed slightly. Refer to
1965 documentation on package locks for details.
1966 * enhancement: EVAL can process IF-expressions without resorting to the
1968 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1969 constant-folded properly due to missing out-of-line functions.
1970 (reported by Mika Pihlajamäki)
1971 * bug fix: LISTEN sometimes returned T even in cases where no data was
1972 immediately available from the stream
1973 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1974 were reversed. (reported by Levente Mészáros)
1975 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1976 when given a vector displaced to an adjustable array. (reported
1978 * bug fix: derivation of float boundaries from numbers outside the
1979 appropriate float range (reported by John Wiseman)
1980 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1982 * bug fix: precomputing cache entries for generic functions with
1983 some subclasses of specializers as yet invalid does not attempt to
1984 fill a cache line with a negative offset. (reported by Levente
1986 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1987 allow condition class objects as arguments to DOCUMENTATION and
1988 (SETF DOCUMENTATION); only find and set documentation for
1989 structure names for the STRUCTURE doc-type. (suggested by Gary
1991 * improvements to the Win32/x86 port:
1992 ** (user-homedir-pathname) and default initialization file
1993 locations now know about the user's "Documents and Settings"
1994 directory (thanks to Yaroslav Kavenchuk)
1995 ** run-program is implemented (thanks to Mike Thomas)
1996 ** sockets support (thanks to Timothy Ritchey)
1997 ** better backtrace support (thanks to Alastair Bridgewater)
1998 ** sb-grovel supported
1999 ** asdf-install and sb-posix work somewhat
2000 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
2001 * improvements to the Solaris/x86 port:
2002 ** works on Solaris 11/Solaris Express
2003 ** floating-point exception handling support
2004 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
2005 * fixed some bugs revealed by Paul Dietz' test suite:
2006 ** REMOVE-METHOD returns its generic function argument even when
2007 the method is not one of the generic functions' methods.
2008 ** objects declared as MEMBER types can be admissible arguments to
2009 structure accessors.
2010 ** printing characters should simply be printed by the FORMAT ~:C
2012 ** compiler failure when compiling functions with hairy constant
2013 defaults for optional parameters.
2014 ** compiler produces wrong code when MAYBE-INLINE-expanding a
2015 function, which is already optimized.
2017 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
2018 * new platform: experimental support for SBCL x86/Darwin, including
2019 MacOS X 10.4.5 on Intel.
2020 * new feature: Unicode character names are now known to the system
2021 (through CHAR-NAME and NAME-CHAR).
2022 * new feature: the filesystem location of SBCL's core file is
2023 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
2024 * minor incompatible change: the contrib modules SB-POSIX and
2025 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
2026 this change is to make it easier to distribute
2027 location-independent binaries.
2028 * bug fix: as implied by AMOP, standardized classes no longer have
2029 slots named by external symbols of public packages. (reported by
2031 * optimization: calling functions via constant symbols -- as in
2032 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
2033 via the function object as in (FUNCALL #'FOO).
2034 * optimization: on x86 and x86-64, the calling convention for the
2035 case of dealing with an unknown number of values has been altered
2036 to be friendlier to the prediction heuristics implemented,
2037 particularly with reference to CALL and RET pairing. (thanks to
2038 Alastair Bridgewater)
2039 * optimization: on x86, the code for access to thread-local symbol
2040 values and binding/unbinding of thread-local symbols is smaller.
2041 (thanks to Alastair Bridgewater)
2042 * enhancement: CONSTANTP is now able to determine constantness of
2043 more complex forms, including calls to constant-foldable standardized
2044 functions and some special forms beyond QUOTE.
2045 * fixed bug: occasional GC crashes on Solaris/x86
2046 * optimization: x86-64 supports stack allocation of results of simple
2047 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
2048 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
2049 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
2050 (thanks to James Knight)
2051 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
2052 by the standard (thanks to Stephen Compall)
2054 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
2055 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
2056 be used for bundling the runtime and the core file into one
2057 executable binary. This feature is not currently supported on all SBCL
2058 platforms. (thanks to James Bielman and NIIMI Satoshi)
2059 * new feature: a generational or ephemeral garbage collector is now
2060 the default on the PowerPC platform (both Linux and Darwin). The
2061 old Cheney (stop and copy) collector is a build-time option.
2062 (thanks to Cyrus Harmon, after Raymond Toy)
2063 * minor incompatible change: the method by which SBCL finds its
2064 contributed modules has changed; it no longer relies on symbolic
2065 links from an $SBCL_HOME/systems directory, but searches directly
2066 in the subdirectories of $SBCL_HOME.
2067 * enhancement: the dynamic heap size on the Linux/PPC platform is
2068 markedly larger, even using the older Cheney garbage collector.
2069 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
2070 certain complicated string types. (reported by Gary King)
2071 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
2072 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
2073 * fixed bug: DOCUMENTATION on structures no longer signals an
2074 error if no structure type of the right name exists. (reported by
2076 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
2077 applicable non-standard primary or :AROUND method on
2078 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
2079 no longer cause unbound variable errors. (reported by Kevin Reid)
2080 * fixed bug: metacircle resolution in cases where methods have slots
2081 added before the slots from STANDARD-METHOD. (reported by Jean
2083 * fixed bug: the Power PC assembly code for calling into lisp
2084 sometimes computed its return address wrongly (depending on the
2085 15th bit of the address, fixed at link time). This used to
2086 manifest itself in a segmentation violation while building PCL.
2087 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
2088 many others over the years)
2089 * fixed bug: ROOM no longer reports silly numbers for stack usage if
2090 the stack pointer (treated as a signed integer) is negative.
2091 (thanks to Peter van Eynde)
2093 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
2094 * new platform: experimental support for the Windows operating
2095 system has been added. (thanks to Alastair Bridgewater)
2096 * fixed several bugs in and robustified the PPC FFI (including
2097 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
2098 * bug fix: allow non-simple string symbol names (reported by Paul
2100 * bug fix: interrupt handling on NetBSD (thanks to Richard M
2102 * bug fix: saving a core corrupted callbacks on x86/x86-64
2103 * bug fix: closed a loophole in metacircularity detection and
2104 grounding in the PCL implementation of CLOS.
2105 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
2106 (reported by tomppa on #lisp)
2107 * bug fix: clear the direction flag on Lisp -> C transitions, as
2108 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
2109 (reported by Andrej Grozin and Hendrik Maryns)
2110 * optimization: major improvements to GC efficiency on GENCGC platforms
2111 * optimization: faster implementation of EQUAL
2112 * optimization: emit more efficient opcodes for some common
2113 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
2114 * optimization: several other minor code-generation improvements on x86-64
2115 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
2117 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
2118 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2119 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
2120 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
2121 by ANSI as a generic function, and for consistency (SETF
2122 GENERIC-FUNCTION-NAME) is treated likewise.
2123 * fixed bug #233.b: make constraint propagation notice when a variable
2124 value is changed after it is referenced but before it is used
2125 * fixed bug #296: no more arbitrary behaviour from filenames with
2126 odd characters as --load arguments. (reported by Adam Warner)
2127 * fixed bug #390: :CHARACTER-SET pathname components now work as
2128 expected. (reported by Tim Daly Jr)
2129 * fixed bug #391: complicated :TYPE intersections in slot
2130 definitions no longer cause an error in PCL internals.
2131 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
2132 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
2133 returning the number of octets which would be written to the
2134 file-stream. (thanks to Robert J. Macomber)
2135 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
2136 arguments to the internal stream functions. (thanks to David
2138 * bug fix: stack exhaustion is now handled on the x86/SunOS
2139 platform. (thanks to Daisuke Homma)
2140 * bug fix: more accurate ROOM results on GENCGC platforms
2141 * optimization: improved type inference for arithmetic-for
2142 index variables in LOOP
2143 * optimization: faster floating-point SQRT on x86-64
2144 * fixed some bugs revealed by Paul Dietz' test suite:
2145 ** DOCUMENTATION returns NIL instead of "" for method combinations
2146 that don't have a docstring
2148 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
2149 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2150 GENERIC-FUNCTION-NAME) are no longer generic functions, and
2151 therefore are not customizeable by user code (as seems to be at
2152 least permitted and maybe required by AMOP). As a consolation,
2153 however, the SBCL implementation of these functions now calls
2154 REINITIALIZE-INSTANCE as specified by AMOP.
2155 * bug fix: slot-definition documentation is propagated as per ANSI
2156 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
2157 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
2158 argument, as do eslotd initialization methods. (from Pascal
2159 Costanza's "Closer" project)
2160 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
2161 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
2163 * bug fix: it is now possible to have more than one subclass of
2164 STANDARD-GENERIC-FUNCTION without causing stack overflow.
2165 (reported by Bruno Haible, Pascal Costanza and others)
2166 * bug fix: the dependent update protocol now works for generic
2167 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
2168 and Pascal Costanza)
2169 * bug fix: condition-class instances corresponding to
2170 DEFINE-CONDITION forms are now created eagerly. (reported by
2171 Kalle Olavi Niemitalo on comp.lang.lisp)
2172 * bug fix: floating point printing is more accurate in some
2173 circumstances. (thanks to Simon Alexander)
2174 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
2175 merged with *DEFAULT-PATHNAME-DEFAULTS*.
2176 * bug fix: callbacks on OS X now preserve stack-alignment.
2177 (thanks to Cyrus Harmon)
2178 * enhancement: the x86-64 disassembler is much better at
2179 disassembling SSE instructions. (thanks to Lutz Euler)
2180 * enhancement: CHECK-TYPE on a variable now allows type-inference
2181 (thanks to Vincent Arkesteijn)
2182 * optimization: improved performance of EUC-JP external format.
2183 (thanks to NIIMI Satoshi)
2184 * optimization: performance improvements to IO on file streams of
2185 :ELEMENT-TYPE CHARACTER
2186 * optimization: much faster memory allocation on x86-64
2187 * optimization: faster garbage collections (latency and throughput)
2189 * optimization: faster bignum subtraction and division on x86 and x86-64
2190 (thanks to Lutz Euler)
2191 * optimization: more accurate type-inference for loops that use a
2192 floating point index variable or a negative step.
2194 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
2195 * bug fix: add a workaround to SBCL looping infinitely at startup on
2196 Linux kernels with apparently buggy implementations of personality().
2197 (thanks to Svein Ove Aas)
2198 * bug fix: Unicode symbols are correctly printed in LDB backtraces
2199 (thanks to David Lichteblau)
2200 * bug fix: local bindings shadow global symbol macros for
2201 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
2202 * optimization: non-open coded uses of numeric comparison operators
2203 (e.g. >) no longer cons when called with more than one parameter
2204 on platforms supporting dynamic-extent allocation.
2205 * enhancement: saving cores with foreign code loaded is now supported
2206 on MIPS/Linux in addition to the previously supported platforms.
2207 * bug fix: FIND-RESTART now tests for activity, not applicability when given
2208 a restart object as identifier. (reported by Helmut Eller for CMUCL)
2209 * bug fix: division by zero in sb-sprof when no samples were collected
2210 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
2211 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
2212 (thanks to Alexander Kjeldaas)
2213 * new feature: ignoring the return values of destructive functions
2214 like NREVERSE signals a compile-time style-warning.
2215 (thanks to Kevin Reid)
2216 * new feature: the alignment of alien structure fields can be
2217 explicitly specified. (thanks to Cyrus Harmon)
2218 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
2219 and dump core on SIGQUIT
2221 ** incompatible change: threads do not inherit values of specials
2222 from their parents (see manual)
2223 ** bug fix: threads stacks belonging to dead threads are freed by the
2224 next exiting thread, no need to gc to collect thread stacks anymore
2225 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
2226 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
2227 ** bug fix: dynamic variable and thread start related gc lossage
2228 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
2230 * fixed some bugs revealed by Paul Dietz' test suite:
2231 ** SUBTYPEP is slightly more accurate on heinously complicated
2232 CONS types where some of the members have uncertain (in the
2233 NIL, NIL sense) type relationships to each other.
2235 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
2236 no longer purified unless :PURIFY T is explicitly specified.
2237 ** Non-purified cores are significantly smaller than before
2239 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
2240 * new feature: timers based on Zach Beane's excellent timer package
2241 * added support for the following external formats: koi8-u,
2242 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
2243 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
2244 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
2245 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
2246 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
2247 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
2248 * incompatible change: a threaded SBCL will no longer revert to
2249 non-threaded mode on non-NPTL systems, but refuse to start entirely.
2250 * bug fix: interrupts are disabled until startup is complete; no
2251 more sigsegvs when receiving a signal to soon
2252 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
2254 * bug fix: add a workaround for the memory randomization features in
2255 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
2256 workaround will only be in effect on systems with the proc filesystem
2258 * bug fix: printing objects of type HASH-TABLE signals a
2259 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
2261 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
2262 have been running for over 50 days. (reported by Gilbert Baumann)
2263 * bug fix: the logic for getting names of functions gets less
2264 confused when confronded with alternate-metaclass
2265 funcallable-instances. (reported by Cyrus Harmon)
2266 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
2267 consistent, even on internal alternate-metaclass objects.
2268 * bug fix: flush closure information collected by physical
2269 environment analysis prepass before the main pass. (bug reported
2271 * bug fix: compiler pack phase does not modify a hashtable, which is
2272 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
2273 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
2274 may be expanded. (reported by Alexander <malishev> on c.l.l)
2275 * bug fix: no more style-warnings when compiling code using LAST
2276 under high-SPEED low-SPACE settings. (reported by David Wallin
2278 * fixed some incompatibilities between SBCL's MOP and the MOP
2280 ** the METAOBJECT class is now implemented;
2281 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
2282 STANDARD-OBJECT, as required;
2283 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
2284 now compatible (as required by VALIDATE-SUPERCLASS); there
2285 remains a consistency requirement over the presence or absence
2286 of the FUNCTION class in the superclass list of finalized
2287 classes; see the manual for more details;
2288 ** the :AROUND method for COMPUTE-SLOTS on
2289 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
2290 requested slot ordering.
2292 ** bug fix: parent thread now can be gc'ed even with a live
2294 ** bug fix: binding a special with PROGV to no value is not
2295 the same as that symbol not having been bound (thanks to
2297 ** bug fix: binding specials is thread safe (thanks to
2299 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
2300 and SB-SPROF do not die with 'no handler for signal XX in
2301 interrupt_handle_now(..)' anymore
2302 ** bug fix: WITH-TIMEOUT works with multiple running threads
2303 * fixed some bugs revealed by Paul Dietz' test suite:
2304 ** ENSURE-GENERIC-FUNCTION should take a method class object for
2305 the :method-class keyword argument.
2307 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
2308 * new port: the Solaris operating system on x86 processors is now
2309 mostly supported, though some rough edges in the environment
2310 remain. (thanks to Daisuke Homma)
2311 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
2312 space for its heap. (thanks to Thiemo Seufer)
2313 * optimization: SBCL on MIPS platforms now supports dynamic-extent
2314 closures. (thanks to Thiemo Seufer)
2315 * minor incompatible change: eof selects abort in the debugger.
2316 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
2317 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
2319 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
2320 deprecated, and will go away in a future revision of SBCL.
2321 * minor incompatible change: GC-ON and GC-OFF are no longer
2322 implemented with a counter, it does not matter how many times gc
2323 is switched on or off
2324 * bug fix: discriminating functions for generic function classes
2325 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
2326 make invalid assumptions about method precedence order. (reported
2328 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
2329 platforms. (thanks to Thiemo Seufer)
2330 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
2331 * bug fix: correct pathname printing with printer escaping is on.
2332 (thanks to Kevin Reid)
2333 * bug fix: complex VOP definitions in "user-space" no longer trigger
2334 package locks. (reported by Zach Beane)
2335 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
2336 for funcallable-instance functions, and is no different from
2338 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
2339 characters in the current readtable. (reported by Nicholas Neuss)
2340 * bug fix: SUBTYPEP on various CONS types returns more a more
2341 accurate acknowledgment of its certainty.
2342 * optimizations: REMOVE-DUPLICATES now runs in linear time on
2343 lists in some cases. This partially fixes bug 384.
2344 * flush all standard streams before prompting in the REPL and the
2346 * bug fix: signal handling and triggering gc do not conflict
2347 directly anymore, in particular a high frequency sb-sprof does
2348 not prevent gc from running
2349 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
2350 approximation for timezone and DST information between the
2351 universal time epoch and the smallest negative 32-bit time_t.
2352 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
2353 year 1899 so long as the encoded time is non-negative.
2354 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
2355 the correct answer for non-constant rotate amounts when performing
2356 an inline 32-bit rotation.
2358 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
2359 there is only one thread in the session
2360 ** bug fix: memory leak for streams created in one thread and
2361 written to in another
2362 ** bug fix: lockup when compiled with gcc4
2363 ** bug fix: race that allows the gc to be triggered when gc is
2365 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
2366 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
2368 ** bug fix: finished threads are gc'ed properly
2369 * fixed some bugs revealed by Paul Dietz' test suite:
2370 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
2371 has arguments with a different set of applicable methods from
2372 the orignal arguments.
2373 ** The type error thrown by MAP now has a correct expected-type
2375 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
2376 name a compiled function.
2377 ** Three MISC tests where a large bignum was improperly coerced to
2378 a float that couldn't represent that bignum during type
2379 derivation were fixed.
2380 ** SUBTYPEP can now handle types involving the negation of a
2381 list-form FUNCTION type.
2382 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
2383 as (COMPLEX (AND RATIO (NOT FIXNUM))).
2384 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
2386 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
2387 * New feature: Experimental support for bivalent streams: streams
2388 opened with :element-type :default now allow character and binary
2389 (unsigned-byte 8) I/O
2390 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
2391 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
2393 * bug fix: no more highly sporadic "couldn't check whether ~S is
2394 readable" when reading a stream and an interrupt hits in the middle
2395 of a select system call
2396 * compiler better recognizes complex arrays (reported by Eduardo
2398 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
2399 for variables on SBCL built with linkage-tables. (reported by Luis
2401 * various error reporting improvements.
2402 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
2403 (Thanks to Thiemo Seufer)
2404 * bug fix: nested reader invokations work correctly
2405 * bug fix: it is possible to have simultaneous references to foreign
2406 code and foreign data with the same name.
2408 ** added x86-64 support
2409 ** incompatible change: the threading api now works with thread
2410 objects instead of thread ids
2411 ** bug fix: threads are protected from signals and interruption when
2412 starting up or going down
2413 ** bug fix: a race where an exiting thread could lose its stack to gc
2414 ** bug fix: don't halt on infinite error in threads if possible
2415 ** fixed numerous gc deadlocks introduced in the pthread merge
2416 ** bug fix: fixed thread safety issues in read and print
2417 ** bug fix: debugger doesn't hang on session lock if interrupted at
2418 an inappropriate moment
2419 ** bug fix: run-program is now thread safe(r)
2420 ** bug fix: inner with-recursive-lock no longer releases the mutex
2421 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
2422 * fixed some bugs revealed by Paul Dietz' test suite:
2423 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
2424 EXPECTED-TYPE slots filled.
2425 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
2427 ** MAKE-CONDITION accepts classes as type-designators.
2428 ** COMPILE may never return NIL.
2429 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
2430 range before calling Unix time functions
2432 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
2433 * numerous signal handling fixes to increase stability
2434 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
2435 * minor incompatible change: we now correctly canonize default
2436 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
2437 as per the MOP, rather than the historical (INITARG INITFUNCTION
2438 INITFORM). (reported by Bruno Haible)
2439 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
2440 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
2441 for more information.
2442 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
2443 pathname is a directory pathname.
2444 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
2445 * dynamic space size on PPC has been increased to 768Mb. (thanks to
2447 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
2448 :METACLASS argument in addition to a class name. (reported by
2449 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
2450 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
2451 (reported by Bruno Haible; patch from Gerd Moellmann)
2452 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
2454 * bug fix: more cleanups to the floating point exception handling on
2455 x86-64 (thanks to James Knight)
2456 * bug fix: the compiler does not try to constant fold calls of
2457 COERCE and BIT-* functions when they return freshly constructed
2458 objects. (reported by Nikodemus Siivola and Paul Dietz)
2459 * optimization: improved the MIPS versions of generic (in the
2460 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
2461 * optimization: direct conversion of (unsigned-byte 32) to floats on
2462 the PowerPC platform.
2463 * optimization: structure instances with raw slots now use less
2464 memory, and probably show better memory locality. (thanks to
2466 * optimization: DYNAMIC-EXTENT declarations for lists and closures
2467 are treated as requests for stack allocation on the x86-64,
2468 Alpha-32, PPC and SPARC platforms.
2469 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
2470 the wrong file descriptor; implementation of SOCKET-OPEN-P.
2471 (thanks to Tony Martinez)
2473 ** gcing a dead thread can no longer lead to lockups
2474 ** threads block signals until they are set up properly
2475 ** errno is no longer shared by threads
2476 ** interrupt-thread restores the eflags register on x86
2477 ** fixed some lockups due to gc/thread interaction
2478 * fixed some bugs revealed by Paul Dietz' test suite:
2479 ** invalid dotted lists no longer raise a read error when
2480 *READ-SUPPRESS* is T
2481 ** signal an error if a symbol that names a declaration is used
2482 as the name of a type, or vice versa
2483 ** allow using the (declare (typespec var*)) abbreviation for
2484 (declare (type typespec var*)) with all type specifiers
2485 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
2486 ** always return NIL from PPRINT-POP when OBJECT is NIL
2487 ** don't signal errors when pretty-printing malformed LABELS,
2488 FLET or MACROLET forms
2489 ** declarations in a DOLIST body are also in scope for the
2491 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
2493 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
2496 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2497 * fixed cross-compiler leakages that prevented building a 32-bit
2498 target with a 64-bit host compiler.
2499 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
2500 opened with :IF-EXISTS :OVERWRITE.
2501 * fixed bug 281: error for an invalid qualifier in a short-form method
2502 combination method is not signalled until the faulty method is called.
2503 * bug fix: iteration variable type inferrer failed to deal with open
2504 intervals. (reported by Alan Shields)
2505 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
2506 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
2507 or more runtime options were provided to the sbcl binary.
2508 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
2509 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
2510 * optimization: there's now a fast-path for fixnum arguments in the
2511 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
2512 * optimization: the code generated on x86-64 is more compact thanks
2513 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
2514 * bug fix: floating-point exception handling now partly works on
2515 x86-64. (Thanks to James Knight)
2516 * improvement to the MIPS backend from Thiemo Seufer: C-style
2517 64-bit long long arguments and return values to alien functions
2519 * fixed some bugs revealed by Paul Dietz' test suite:
2520 ** the type-error signalled from WARN has a filled-in DATUM slot.
2521 ** the type-error required when a stream is not associated with
2522 a file has the stream as its datum.
2523 ** type-errors on single-floats on x86-64 no longer have
2524 :INVALID-OBJECT as the datum
2525 ** the type-errors signalled for invalid function names now have
2526 a correct expected type
2527 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2528 for structures defined with a :TYPE. Documentation strings for
2529 typed structures are no longer immediately discarded
2530 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2531 broadcast streams and synonym streams. FILE-LENGTH now also works
2532 on broadcast streams.
2534 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2535 * incompatible change: the --noprogrammer option, deprecated since
2536 version 0.7.5, has been removed. Please use the equivalent
2537 --disable-debugger option instead.
2538 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2540 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2541 inoperational for a while now) has been completely removed.
2542 * null lexical environments are now printed as #<NULL-LEXENV>,
2543 significantly reducing the amount of clutter in typical backtraces.
2544 * documentation on weak pointers, finalization, and after GC hooks
2545 has been added to the manual.
2546 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2547 as other specialized array types, is much faster. SUBSEQ and
2548 COPY-SEQ on such arrays have also been sped up.
2549 * optimization: EQL is now more efficient when at least other argument
2550 is known to be of type (OR FIXNUM (NOT NUMBER)).
2551 * fixed bug: the runtime is now less vulnerable to changes in the
2552 size of the SBCL object on OS X, and virtual memory is reserved for
2553 all spaces that need to be at a fixed address.
2554 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2555 * fixed bug: finalizers and after GC hooks that cause consing are now
2557 * fixed bug: compiler error messages and summaries are now printed to
2558 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2559 * fixed inference of the upper bound of an iteration variable.
2560 (reported by Rajat Datta).
2561 * fixed bug 211e: calling local functions with duplicated constant
2562 keyword argument no longer causes a bogus style warning about an
2564 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2565 ftype declaration to be disregarded. (reported by Dave Roberts)
2566 * fixed bug 373: caused by erronous compilation of references to alien
2567 variables in the runtime on ppc/darwin.
2568 * fixed bug 376: CONJUGATE type deriver.
2569 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2570 (reported by Baughn on #lisp)
2571 * a cross-compiler bug on non-x86 platforms has been identified and
2572 fixed. (thanks to Bruno Haible)
2573 * improvements to the MIPS runtime code for increased stability.
2574 (thanks to Thiemo Seufer)
2575 * increased the maximimum compact environment size to allow
2576 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2577 * partial workaround for bug 135: don't name the function we're
2578 calling for hairy cases of EVAL, so as not to accumulate one environment
2579 entry per call to EVAL. (reported by Kevin Reid)
2580 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2581 * optimization: type testing for non-vector arrays should be faster.
2582 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2584 * bug fix: setting 31st element of a bit vector to zero did not work
2586 * bug fix: redefining a class definition which failed due to a
2587 previous accessor / function clash now works (but see BUGS entry
2588 #380 for more problems in this area). (thanks to Zach Beane)
2589 * the long form of DEFINE-METHOD-COMBINATION disables method group
2590 checking when given a single method group with pattern *. (thanks
2592 * on x86 compiler supports stack allocation of results of simple
2593 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2594 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2595 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2596 * fixed some bugs related to Unicode integration:
2597 ** the restarts for recovering from input and output encoding
2598 errors only appear when there is in fact such an error to
2600 * fixed some bugs revealed by Paul Dietz' test suite:
2601 ** MISC.549 and similar: late transformation of unsafe type
2602 assertions into derived types caused unexpected code
2604 ** SCALE-FLOAT type deriver is less wrong.
2605 ** type derivers for EXP, LOG and similar functions compute result
2606 types for complex arguments better.
2607 ** (MISC.563) CONJUGATE type deriver works for very restricted
2609 ** out-of-line type testers for character strings are available.
2610 ** EQUAL compiler transform understands specialness of objects
2612 ** accessing double-floats stored on the stack now works on x86-64.
2613 ** debugger internals could sometimes create invalid lispobjs,
2614 resulting in GC crashes.
2615 ** MISC.548: type check weakening can convert required type into
2617 ** initialization forms for bindings are not in scope of free special
2620 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2621 * incompatible change: thread support for non-NPTL systems has
2622 been removed - locking is buggy and unreliable. A threaded
2623 SBCL build will now warn at startup and refuse to create
2624 new threads, unless futex support is detected in the kernel
2625 * incompatible change: the top level REPL now has only an ABORT
2626 restart associated with it, not TOPLEVEL and ABORT as it used to.
2627 TOP and TOPLEVEL are now available as debugger commands for
2628 returning to the top level.
2629 * incompatible change: forms evaluated in the REPL now use the
2630 global optimization policy.
2631 * incompatible change: user- and system-initialization files are
2632 no longer processed with LOAD, but by READ and EVAL; hence the
2633 global optimization policy, startup package, readtable, etc,
2635 * The .fasl file format number has been incremented because of
2636 various incompatible changes.
2637 * internal entry point details and argument counts no longer appear
2638 in backtraces unless explicitly requested by setting
2639 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2640 * built-in and standard functions no longer have names like "top
2641 level local call to FOO".
2642 * fixed bug 32: functions defined in non-null lexical environments
2643 now have more legible printed representation
2644 * fixed bug 33: functions defined in non-null lexical environemnts
2645 are now more amenable to inspection by INSPECT.
2646 * workaround for bug 354: XEPs no longer appear in backtraces unless
2647 explicitly requested.
2648 * fixed bug: receiving the signal which results from attempting to
2649 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2650 SIGBUS on other *BSDs) on architectures where the C stack is also
2651 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2652 to debugger confusion or wild instability. (reported by Cheuksan
2654 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2655 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2656 (reported by Lutz Euler)
2657 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2658 compiled code. (reported by Paul Dietz)
2659 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2660 specialized parameter to a method to something that is not TYPEP
2661 the specializer is now possible.
2662 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2663 face of package deletion.
2664 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2665 pathname. (reported by David Tolpin on c.l.l)
2666 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2667 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2668 STRING-STREAM and FILE-STREAM, respectively.
2669 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2670 than previously; in addition, its test suite is now run on build.
2671 (thanks to Luke Gorrie)
2672 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2673 * optimization: added a immediate representation for single-floats
2675 * fixed some bugs revealed by Paul Dietz' test suite:
2676 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2677 ** attempting to create a package with a colliding nickname causes
2678 correctable errors to be signalled.
2679 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2680 ** several x86-64 backend bugs related to sign-extension of immediate
2683 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2684 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2685 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2687 * fixed loading of multiply forward-referenced layouts.
2688 (thanks to Cheuksan Wang)
2689 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2690 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2691 (see COMPILE-FILE documentation for details of :PRINT option)
2692 * fixed bugs 19 and 317: fixed-format floating point printing is
2693 more accurate. This also fixes a bug reported by Adam Warner
2694 related to the ~@F format directive.
2695 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2697 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2698 dispatch table if the from-char is a dispatch macro character.
2699 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2700 by the wordsize no longer ignore the last word. (reported by Lutz
2702 * fixed bug in type checking of dynamic-extent variables. (reported
2704 * optimization: sequence traversal functions use their freedom to
2705 coerce function designators to functions.
2706 * optimization: code with many calls to CLOS methods specialized on
2707 CLOS classes has had redundant type checks removed.
2708 * optimization: added declarations to speed up operations that access
2709 the internal character database (for example STRING-UPCASE)
2710 * optimization: comparison operations between floats and sufficiently small
2711 fixnums no longer create extra rationals
2712 * fixed some bugs related to Unicode integration:
2713 ** portions of multibyte characters at the end of buffers for
2714 character-based file input are correctly transferred to the
2715 start of the buffer at the next read.
2716 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2717 passing it through to OPEN.
2718 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2719 argument given to internal calls to OPEN.
2720 ** the built-in comment readers (introduced by character sequences
2721 ";" and "#|") are more forgiving to encoding errors; they will
2722 STYLE-WARN and then attempt to resync the stream at a character
2723 boundary. (thanks to Teemu Kalvas)
2724 * fixed some bugs in the x86-64 port:
2725 ** Negative short int return values from c-calls are sign-extended
2727 ** The stack is aligned to 16-bytes for c-calls, as required by
2728 the ABI. (thanks to Cheuksan Wang)
2729 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2730 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2731 for enums. (thanks to Vincent Arkesteijn)
2732 ** Multiple small optimizations and bugfixes for floating point
2734 * fixed some bugs revealed by Paul Dietz' test suite:
2735 ** Space, Tab, Linefeed, Return and Page have the invalid
2736 secondary constituent character trait.
2737 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2739 ** WITH-INPUT-FROM-STRING should only update the index place on
2741 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2742 works more reliably.
2743 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2744 with one-dimensional arrays with fill pointers.
2745 ** TYPE-OF failed on a complex with an integer realpart and a
2747 ** compiler failure during type inference for the code of form
2748 (IF (EQL X (THE ...)) ...) (MISC.535).
2750 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2751 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2752 platform. The port supports 61-bit fixnums, large memory spaces
2753 and reloading shared object files.
2754 * enhancement: saving cores with foreign code loaded is now
2755 supported on ppc/Darwin in addition to the previously supported
2757 * enhancement: the statistical profiler now walks deeper into the
2758 call stack for more meaningful call-graphs and accrued time
2759 reports (x86/x86-64 only). It also now reports time spent in
2761 * enhancement: it is now possible to trace most individual methods
2762 of a generic function in addition to tracing the generic function
2764 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2766 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2767 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2768 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2769 produces an error. (thanks to Vincent Arkesteijn)
2770 * bug fix: NAMESTRING on pathnames with :WILD components in their
2771 directories works correctly. (thanks to Artem V. Andreev)
2772 * fixed bug 125: compiler preserves identity of closures. (reported
2774 * bug fixed: functions with &REST arguments sometimes failed with
2775 "Undefined function" when compiled with (DEBUG 3). (reported by
2777 * bug fix: overflow during compiling of setting element of a bit
2778 vector with constant index and value. (reported by Timmy Douglas)
2779 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2780 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2781 * on x86 compiler partially supports stack allocation of dynamic-extent
2783 * GO and RETURN-FROM do not check the extent of their exit points
2784 when compiled with SAFETY 0.
2785 * fixed some bugs related to Unicode integration:
2786 ** encoding and decoding errors are now much more robustly
2787 handled; it should now be possible to recover even from invalid
2788 input or output to the terminal. (thanks to Teemu Kalvas)
2789 ** provided a first cut at implementing STRING-TO-OCTETS and
2790 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2791 ** altered the SB-MD5 contributed module slightly, changing the
2792 interface just enough for it to be supportable for builds where
2793 lisp characters are not eight bits.
2794 * fixed some bugs revealed by Paul Dietz' test suite:
2795 ** the FORMATTER-generated functions for ~V[ conditionals require
2796 the correct number of arguments.
2797 ** READ-FROM-STRING returns the mandated second value when applied
2798 to displaced strings.
2799 ** the #\Rubout and #\Backspace characters are treated as invalid
2800 constituent characters by the tokenizer.
2802 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2803 * new feature: reloading changed shared object files with
2804 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2805 * new feature: references to foreign variables and functions
2806 can now be compiled and loaded before the corresponding shared
2807 object file is loaded, as long as the foreign definitions are
2808 available at runtime.
2809 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2810 supported. (thanks to Dan Debertin)
2811 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2812 just keywords. (thanks to Vincent Arkesteijn)
2813 * fixed bug #331: structure-class instances corresponding to
2814 DEFSTRUCT forms are now created eagerly.
2815 * fixed bug #345: backtraces from calls to undefined functions work
2816 on x86 as well. Related bug #61 is now also partially fixed on x86
2817 (backtraces from throws to unknown catch tags.)
2818 * bug fix: lambda-list parsing is now stricter vrt. order and number
2819 of lambda-list keywords.
2820 * bug fix: as specified by AMOP, an error is signalled if a
2821 class-option appears multiple times in a DEFCLASS form. (reported
2823 * bug fix: RANDOM can be compiled when the compiler derives the type
2824 of its numeric argument as a disjoint set of small integers.
2825 (reported by Paul Dietz)
2826 * bug fix: starting a core saved with shared objects loaded when
2827 those objects are not available no longer causes threaded SBCL to
2828 hang. (reported by Sean Ross)
2829 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2831 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2832 argument, which must always be NIL. (reported by Kalle Niemitalo)
2833 * bug fix: printing 1.0d+23 no longer results in an error.
2834 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2835 * bug fix: structure slot setters preserve evaluation order. (thanks
2837 * fixed some bugs related to Unicode integration:
2838 ** RUN-PROGRAM can allow its child to take input from a Lisp
2839 stream. (reported by Stefan Scholl)
2840 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2841 locales. (reported by Ken Causey)
2842 * fixed some bugs revealed by Paul Dietz' test suite:
2843 ** INCF, DECF and REMF evaluate their place form as specified in
2845 ** FORMATTER expands ~{ ~} iteration directives with V or #
2846 parameters correctly.
2847 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2848 consequent uses no arguments correctly.
2849 ** the system has a partial understanding of the (COMPLEX RATIO)
2851 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2854 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2855 * new feature: a build-time option (controlled by the :SB-UNICODE
2856 keyword feature, enabled by default) for building the system with
2857 support for the entire 21-bit character space defined by the
2859 * new feature: the system now has rudimentary external-format
2860 support; the primary user-visible change at this time is that
2861 characters with the high bit set (such as lower-case-e-acute) will
2862 print correctly to a terminal in a UTF-8 environment.
2863 * minor incompatible change: BASE-CHAR no longer names a class;
2864 however, CHARACTER continues to do so, as required by ANSI.
2865 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2866 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2867 has been moved to the SB-EXT package (temporarily re-exported from
2869 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2870 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2872 * fixed bug #351: better error-handlind and reporting for malformed
2874 * fixed bug #350: bignum-printing is now more memory-efficient,
2875 allowing printing of very large bignums, eg. (expt 2 10000000).
2876 (reported by Bruno Haible)
2877 * fixed bug #302: better primitive-type selection for intersection
2879 * fixed bug #308: non-graphic characters now all have names, as
2880 required. (reported by Bruno Haible)
2881 * bug fix: redefining a class with different superclasses now correctly
2882 removes it from the direct-subclasses of its previous superclasses.
2883 (reported by David Morse)
2884 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2885 new value now works. (reported by Bruno Haible)
2886 * bug fix: correct canonicalization of multiple non-standard slot
2887 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2888 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2889 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2890 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2891 now exists, an signals an error.
2892 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2893 during macroexpansion for non-iteration variables as well. (reported
2894 by Bruno Haible for CMUCL)
2895 * bug fix: Cyclic structures and unprintable objects in compiler
2896 messages no longer cause errors. (reported by Bruno Haible)
2897 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2898 and READ-FROM-STRING all now return a primary value of NIL if
2899 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2900 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2901 (reported by Bruno Haible for CMUCL)
2902 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2903 do not designate a condition. (reported by Bruno Haible for
2905 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2906 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2907 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2908 permitting resolution in favour of any of the conflicting symbols.
2909 (reported by Bruno Haible for CMUCL)
2910 * bug fix: EQUAL compiler optimizations is less aggressive on
2911 strings which can potentially compare true despite having distinct
2912 specialized array element types.
2913 * bug fix: unit enumerations can be defined without dividing by
2914 zero. (thanks to Vincent Arkesteijn)
2915 * FORMAT compile-time argument count checking has been enhanced.
2916 (report from Bruno Haible for CMUCL)
2917 * a partial workaround for the bug 262: the compiler does not try to
2918 inline-expand a local function doing RETURN-FROM from a deleted
2919 BLOCK. (thanks to Peter Denno for the bug report and to David
2920 Wragg for the simple test case)
2921 * fixed some bugs revealed by Paul Dietz' test suite:
2922 ** RENAME-PACKAGE allows all package designators as new package
2924 ** constraint propagation and lambda variable substitution are
2925 more cautious in dealing with partially deleted code.
2926 ** compiler sometimes lost reoptimization passes.
2927 ** CERROR, given a condition as condition designator, passes its
2928 remaining arguments to the continue format control without
2930 ** Case-altering FORMAT directives work correctly on non-ASCII
2932 ** The REAL type specifier handles bounds outside the floating
2933 point ranges without signalling FLOATING-POINT-OVERFLOW.
2934 ** Functions with IR1-transformations can create intercomponent
2935 references to global functions.
2936 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2938 ** FORMAT ~R treats a nil value for its first parameter correctly.
2940 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2941 * enhancement: saving cores with foreign code loaded is now
2942 supported on x86/NetBSD and sparc/Linux in addition to the previously
2943 supported platforms.
2944 * bug fix: on some platforms repeated installations caused multiple
2945 copies of HTML documentation to be installed -- should not happen
2946 any more. (reported by Stefan Scholl)
2947 * bug fix: parsing self-recursive alien record types multiple times
2948 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2949 original patch by Helmut Eller for CMUCL)
2950 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2951 (thanks to Richard Kreuter)
2952 * bug fix: defining classes whose accessors are methods on existing
2953 generic functions in other (locked) packages no longer signals
2954 bogus package lock violations. (reported by François-René Rideau)
2955 * bug fix: special variables as DEFMETHOD parameters no longer have
2956 associated bogus type declarations. (reported by David Wragg and
2958 * bug fix: read-write consistency on streams of element-type
2959 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2960 * bug fix: redefiniton of the only method of a generic function with
2961 no DEFGENERIC no longer emits a full WARNING. In addition,
2962 redefinition of generic functions with no DEFGENERIC to an
2963 incompatible lambda list now signals an error. (thanks to Zach
2965 * bug fix: DEFGENERIC now works even when there's a function of the
2966 same name in an enclosing lexical environment. (thanks to Zach
2968 * fixed compiler failure, caused by instrumenting code during
2969 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2970 * optimization: added loop analysis and improved register allocation
2971 * fixed some bugs revealed by Paul Dietz' test suite:
2972 ** POSITION on displaced vectors with non-zero displacement
2973 returns the right answer.
2974 ** (SIMPLE-STRING) is a valid type specifier for sequence
2976 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2978 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2979 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2981 ** The justification version of the FORMAT ~< directive treats
2982 non-zero minpad parameter correctly.
2984 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2985 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2986 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2987 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2988 the supported interface.
2989 * new feature: Single-stepping of code compiled with DEBUG 2 or
2990 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2991 * new feature: saving cores with foreign code loaded is now
2992 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2993 Timothy Moore's work for CMUCL)
2994 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2995 parameters to * instead of NIL if no initform is supplied.
2996 (reported by Johan Bockgaard)
2997 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2998 correctly when FUNCALL appears as the car of the form. Note:
2999 despite this FUNCALL forms are not currently subject to
3000 compiler-macro expansion. (port of Raymond Toy's fix for the
3001 same from CMUCL, reported by Johan Bockgaard)
3002 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
3003 (thanks for Teemu Kalvas)
3004 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
3005 case instead of terminating the iteration (thanks for Julian
3006 Squires, Sean Champ and Raymond Toy)
3007 * bug fix: incorrect expansion of defgeneric that caused a style
3008 warning. (thanks for Zach Beane)
3009 * bug fix: slot accessor effective method computation works properly
3010 for classes with multiple non-standard applicable methods on
3011 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
3012 * on x86 compiler supports stack allocation of results of LIST and
3013 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
3014 CMUCL implementation by Gerd Moellmann)
3015 * fixed some bugs revealed by Paul Dietz' test suite:
3016 ** FORMAT strings with both the ~<~:;~> form of the justification
3017 directive and pretty-printing directives cause an error.
3019 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
3020 * incompatible change: the internal functions
3021 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
3022 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
3023 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
3024 instead of the old functions.
3025 * new feature: on platforms where "dladdr" is available foreign
3026 function names now appear in backtraces. (based on Helmut Eller's
3028 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
3029 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
3031 * documentation: ASDF manual is now distributed with SBCL.
3032 * bug fix: SBCL can now load its contributed modules with REQUIRE
3033 even if the system-provided entries have been removed from
3034 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
3036 * bug fix: backtraces involving undefined functions or assembly
3037 routines are more informative. (thanks to Brian Downing)
3038 * bug fix: mutually referent alien structures now work correctly.
3039 (reported by Rick Taube)
3040 * bug fix: structures defined by WITH-ALIEN can be referred to
3041 within other definitions in the same WITH-ALIEN.
3042 * bug fix: division operators (MOD, TRUNCATE and the like) with
3043 constant zero divisors and integer dividends no longer generate
3045 * bug fix: provide default methods for INPUT-STREAM-P and
3046 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
3047 * bug fix: improve the detection and resolution of MOP metacycles
3048 (where effective method computation depends on the generic function
3049 having its effective method computed). (reported by Bruno Haible)
3050 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
3051 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
3052 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
3053 * bug fix: pathnames with relative directory components can be
3054 represented relative to default pathnames.
3055 * optimization: in taking the GCD of bignums, reduce the two bignums
3056 to approximately the same size (using Euclid's algorithm) before
3057 applying the more sophisticated binary GCD. (thanks to Juho
3059 * optimization: COUNT on bitvectors now operates word-at-a-time.
3060 * optimization: ASH with a positive, but not necessarily constant,
3061 (leftwards) shift, when in a modular context, is compiled to a
3063 * fixed some bugs revealed by Paul Dietz' test suite:
3064 ** FORMAT variable parameters ("~V<char>") are defaulted properly
3065 if the corresponding argument is NIL.
3066 ** FORMAT directives accepting a minpad parameter treat negative
3068 ** literal commas as character parameters to FORMAT directives are
3070 ** literal spaces directly after ~<Newline> directives within a
3071 format-logical-block (~:< ~@:>) do not induce :FILL-style
3072 conditional newlines.
3073 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
3074 as specified: it no longer includes conditional newlines).
3075 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
3077 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
3078 :PER-LINE-PREFIX argument does not evaluate to a string.
3080 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
3081 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
3082 the manual for details; for now, package locks can be disabled by
3083 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
3084 if no major problems are found then it is likely that they will be
3085 compiled in unconditionally.
3086 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
3087 unsupported operators on all platforms. To load a shared library
3088 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
3089 object file, link it into a shared library outside of SBCL and
3090 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
3092 * bug workaround: ROOM T and the GC were not working together
3093 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
3094 reporting facility aren't preserved. That reporting has been
3095 disabled, so now until and unless someone figures out how to make it
3096 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
3097 * minor incompatible change: as threatened around sbcl-0.8.0, the
3098 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
3099 The symbols are also exported from SB-PCL for backwards
3100 compatibility, but more so than before SB-PCL should be treated as
3101 an implementation-internal package.
3102 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
3104 * the SB-SPROF contrib now works on (most) non-x86 architectures.
3105 It is known as of this release not to work on the Alpha, however.
3106 * fixed bug #167: errors signalled due to illegal syntax in method
3107 bodies are now more legible.
3108 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
3109 designators and can hence be used with TYPEP.
3110 * fixed bug #333: CHECK-TYPE now ensures that the type error
3111 signalled, if any, has the right object to be accessed by
3112 TYPE-ERROR-DATUM. (reported by Tony Martinez)
3113 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
3114 5.1.2.3 for multiple-value place subforms. (reported by Kalle
3116 * fixed bug #334: programmatic addition of slots using specialized
3117 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
3118 and :ALLOCATION :CLASS slots.
3119 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
3120 float exponents. (rereported by Peter Seibel)
3121 * fixed a bug: #\Space (and other whitespace characters) are no
3122 longer considered to be macro characters in standard syntax by
3123 GET-MACRO-CHARACTER.
3124 * fixed bug: initialization of condition class metaobjects no longer
3125 causes an instance of the condition to be created. (reported by
3127 * fixed bug: it is now possible to have slots such that
3128 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
3129 description is neither :INSTANCE nor :CLASS.
3130 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
3131 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
3132 SLOT-VALUE-USING-CLASS), and no longer causes errors with
3133 non-standard SLOT-DEFINITION-ALLOCATION values.
3134 * fixed bugs: various race conditions handling exiting threads.
3135 CL-PPCRE's thread torture test now passes.
3136 * fixed arguably-a-bug: GC time proportional to number of threads in
3137 system even when most of them are idle
3138 * optimization: improved performance of BIT and SBIT on bit-vectors.
3139 * .fasl file incompatibility: The fasl file version number has
3140 been incremented because of changes associated with package locks.
3142 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
3143 * minor incompatible change: the system no longer provides
3144 optimization hints (conditions which are TYPEP
3145 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
3146 mode; these hints are still emitted when higher SPEED optimization
3148 * new contrib module: a sampling profiler (profiling by statistical
3149 sampling, rather than by instrumenting functions) is available as
3150 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
3151 * the behaviour of the standard function ED is now customizeable by
3152 third parties through a hook variable: see ED's documentation
3153 string for information on the protocol.
3154 * the compiler no longer emits efficiency notes for (FUNCALL X)
3155 when the type of X is uncertain under default optimization
3157 * fixed bug 276: mutating a binding of a specialized parameter to a
3158 method to something that is not TYPEP the specializer is now
3160 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
3161 exhibit the required behaviour on the x86 platform. (thanks to
3162 Peter van Eynde, Eric Marsden and Bruno Haible)
3163 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
3164 even for difficult arguments. (reported by Peter Graves)
3165 * fixed bug 141a: the backquote printer now descends quoted
3167 * fixed another bug in backquote printing: no more destructive
3168 modification of the form's list structure. (reported by Brian
3170 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
3171 move between its address being taken and the call to
3172 interrupt_thread, fixing a crashing race condition.
3173 * the SB-POSIX contrib implementation has been adjusted so that it
3174 no longer exhibits ridiculously poor performance when constructing
3175 instances corresponding to C structs.
3177 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
3178 * minor incompatible change: the sb-grovel contrib now treats C
3179 structures as alien (in the sense of SB-ALIEN) objects rather than
3180 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
3181 has implications for memory management of client code
3182 (sb-grovel-returned objects must now be manually managed) and for
3183 type safety (alien objects now have full types).
3184 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
3185 used to control emission of compiler diagnostics, rather than the
3186 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
3187 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
3188 quality should be considered deprecated.
3189 * install.sh now installs the user manual as well
3190 * (not quite a new documentable feature, but worth considering in
3191 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
3192 beginnings of a semantically meaningful condition hierarchy is
3193 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
3195 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
3196 designator as the defaults argument.
3197 * fixed bug: Displaced arrays whose displaced-to array has become
3198 too small now cause ARRAY-DIMENSION to signal an error, providing
3199 for safer bounds-checking. (reported by Bruno Haible)
3200 * fixed bug: DEFCLASS slot definitions with identical :READER and
3201 :WRITER names now signal a reasonable error. (reported by Thomas
3203 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
3205 * fixed bug: Invalid slot specification errors now print correctly.
3206 (thanks to Zach Beane)
3207 * fixed bug 320: Shared to local slot value transfers in class
3208 redefinitions now happen corrently from superclasses as
3209 well. (reported by Bruno Haible)
3210 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
3212 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
3213 improper lists correctly. (reported by Bruno Haible)
3214 * fixed bug 313: source-transform for <fun-name> was erroneously
3215 applied to a call of a value of a variable with name <fun-name>.
3216 (reported by Antonio Menezes Leitao)
3217 * fixed bug 307: The obsolete instance protocol ensures that
3218 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
3219 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
3220 remove dead unknown-values globs from the middle of the stack.
3221 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
3222 easier to resume long computations after using *BREAK-ON-SIGNALS*
3223 to diagnose and fix failures. (thanks to Nikodemus Siivola)
3224 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
3225 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
3226 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
3227 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
3228 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
3230 * fixed a (fairly theoretical) bug in string printing: if
3231 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
3232 does not have array-element-type equal to the most general string
3234 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
3235 function name. (thanks to Nikodemus Siivola)
3236 * fixed bug: compile-time format string checker failed on
3237 non-closed ~{. (reported by Thomas F Burdick)
3238 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
3239 don't warn on legal loop code involving more than one aggregate
3240 boolean. (thanks to Nikodemus Siivola)
3241 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
3242 in keyword position to MAKE-PACKAGE was not being reported as
3244 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
3245 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
3247 * fixed bugs 280 and 312: the checking for multiple definitions in a
3248 file is less likely to become confused by uses of inline
3250 * fixed bug: the #S reader macro performs the keyword coercion
3251 specified for slot names. (reported by Kalle Niemitalo)
3252 * fixed bug: lambda lists may contain symbols whose names start with
3253 & but are not lambda-list-keywords; their occurrence triggers a
3255 * fixed bug 321: define-method-combination argument lambda lists do
3256 not cause invalid code to be generated when &optional or &aux
3257 variables have default values. (reported by Bruno Haible)
3258 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
3259 have CLOS classes; this appears also to have fixed reported
3260 instability in stack exhaustion detection.
3261 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
3262 MUFFLE-WARNING when no associated restart is present is now
3264 * optimization: rearranged the expansion of various defining macros
3265 so that each expands into only one top-level form in a
3266 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
3268 * optimization: used a previously little-used slot in symbols to
3269 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
3271 * fixed some bugs revealed by Paul Dietz' test suite:
3272 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
3274 ** Fixed an optimization bug related to inheritance of initforms
3275 from local to shared slots.
3276 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
3277 ** CAST optimizer forgot to flush argument derived type.
3278 ** print/read consistency on floats is now orders of magnitude
3279 more likely. (thanks also to Bruno Haible for a similar report
3281 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
3282 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
3283 for objects of type REAL. Make it so.
3284 ** SXHASH is specified (infelicitously) to respect similarity,
3285 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
3286 it so. (thanks to Markus Ziegler)
3287 ** on the Alpha, the compiler succeeds in compiling functions
3288 returning a known number of arguments greater than 63.
3289 ** fixed handling of invalid NIL arguments in keyword position
3291 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
3293 ** provided a readably-printable representation for RANDOM-STATE
3295 ** ensured that pathnames, where they have a namestring, always
3296 print using #P"..." syntax.
3298 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
3299 * Support for the forthcoming 2.0 version of the NetBSD kernel
3300 running on x86 hardware has been added. (thanks to Perry
3301 E. Metzger most immediately, and others for their past work)
3302 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
3303 current release version; SBCL's OpenBSD support had been broken
3304 since about the time of OpenBSD's switch to ELF binary format.)
3305 * [placeholder for DX summary]
3306 ** user code with &REST lists declared dynamic-extent, under high
3307 speed or space and low safety and debug optimization policy.
3308 * The manual has been converted to Texinfo format and the debugger
3309 chapter from the cmucl manual has been added.
3310 * A facility has been added to extract documentation strings from
3311 sbcl and store them as Texinfo-formatted snippets for inclusion in
3312 the manual (via Texinfo's @include directive)
3313 * bug fix: compiler emitted division in optimized DEREF. (thanks for
3314 the test case to Dave Roberts)
3315 * bug fix: multidimensional simple arrays loaded from FASLs had fill
3316 pointers. (reported by Sean Ross)
3317 * bug fix: PROFILE output is printed nicely even for large numerical
3318 values. (thanks to Zach Beane)
3319 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
3320 greater than 32 handle EOF correctly.
3321 * bug fix: on X86 an immediate argument of the IMUL instruction is
3322 correctly printed in disassembly. (thanks to Lutz Euler)
3323 * bug fix: class slots in redefined classes preserve their old
3324 values. (thanks to Bruno Haible and Nikodemus Siivola)
3325 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
3326 succeeds. (reported by Marco Baringer)
3327 * fixed some bugs revealed by Paul Dietz' test suite:
3328 ** READ-SEQUENCE now works on ECHO-STREAMs.
3329 ** RATIONALIZE works more according to its specification. (thanks
3332 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
3333 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
3334 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
3335 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
3336 should matter to you only if you rebind the printer control
3337 variables and then find you want different bindings in the
3338 debugger than in the ordinary execution of your program.)
3339 * The runtime build system has been tweaked to support building
3340 (on SPARC/SunOS) using a C compiler which invokes Sun's own
3341 assembler and linker. (thanks to Nikodemus Siivola)
3342 * Unbound, undefined, undeclared variables now trigger full
3343 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
3344 more often programmer error than deliberate exploitation of undefined
3346 * optimization: the hash algorithm for strings has changed to one
3347 that is less vulnerable to spurious collisions. (thanks to Juho
3349 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
3350 needless bounds checking. (thanks to Juho Snellman)
3351 * optimization: implemented multiplication as a modular
3352 (UNSIGNED-BYTE 32) operation on the PPC backend.
3353 * fixed some bugs revealed by Paul Dietz' test suite:
3354 ** ADJUST-ARRAY now copies the datum in a zero rank array if
3356 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
3357 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
3358 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
3360 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
3362 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
3364 ** ECHO-STREAMs no longer attempt to echo the end of file value to
3365 their output stream on EOF from read.
3366 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
3367 have been read to end-of-file.
3368 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
3370 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
3371 description of determination of which consecutive characters
3373 ** Printing the "Space" character with escaping on now yields "#\\ ",
3374 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
3375 ** Reading floating-point numbers with *READ-BASE* set to a number
3376 less than 10 works correctly.
3377 ** Reading floating-point numbers with *READ-BASE* set to a number
3378 more than 10 works correctly.
3379 ** Printing with *PRINT-READABLY* targets the standard readtable, not
3380 the readtable currently in effect.
3382 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
3383 * minor incompatible change: parsing of namestrings on a physical
3384 (Unix) host has changed; numbers after the final #\. in a
3385 namestring are no longer interpreted as a version field. This is
3386 intented to be largely invisible to the user, except that the
3387 meaning of the namestring "*.*.*" has changed: it now refers to a
3388 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
3389 should usually be replaced by
3390 (make-pathname :name :wild :type :wild :version :wild)
3391 with the added benefit that this is more likely to be portable.
3392 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
3393 option to OPEN now signals an error if the file being opened
3394 exists; this may have an impact on existing code.
3395 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
3396 signals received, on the PowerPC platforms (both Linux and
3397 Darwin). (thanks to Pierre Mai for pointing out the location of
3399 * several fixes on OS X: The system now builds and runs cleanly on
3400 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
3401 (thanks to Brian Mastenbrook)
3402 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
3403 second-resolution: integer multiples of 1/3600 between -24 and 24.
3404 (thanks to Vincent Arkesteijn)
3405 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
3406 type when called with 1 argument; PEEK-CHAR checked type of
3407 PEEK-TYPE only after having read first character from a
3408 stream. (reported by Peter Graves)
3409 * bug fix: the garbage collector now has much better locality
3410 behaviour, and in particular no longer treats all memory as being
3411 exhausted when many small objects point to each other in a deeply
3413 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
3414 recognized as being TYPEP their class.
3415 * bug fix: the PUSHNEW documentation string has been corrected.
3416 (thanks to Vincent Arkesteijn)
3417 * bug fix: defaulting of the value for the last of an atypically
3418 large number of multiple values being bound was not being
3419 performed correctly on the Alpha or PPC platforms
3420 * optimization: implemented multiplication as a modular
3421 (UNSIGNED-BYTE 32) operation on the x86 backend.
3422 * optimization: SEARCH on simple-base-strings can now be open-coded.
3423 (see also contrib/compiler-extras.lisp for inspiration for
3424 teaching the compiler about the Boyer-Moore algorithm).
3425 * value, returned by MAX (and MIN) called with several EQUALP, but
3426 not EQL, arguments now does not depend on compiler settings.
3427 * fixed some bugs revealed by Paul Dietz' test suite:
3428 ** in stack analysis liveness information is propagated from
3429 non-local entry points.
3430 ** pathwise CAST removing failed when the CAST node did not start
3432 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
3433 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
3435 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
3436 host is already defined.
3437 ** RENAME-FILE works on streams instead of signalling an internal
3439 ** PEEK-CHAR uses the current readtable when determining whether
3440 or not a character is whitespace.
3441 ** MERGE-PATHNAMES handles the case when the pathname does not
3442 specify a name while the default-pathname specifies a version.
3443 ** Pathnames now stand a better chance of respecting print/read
3445 ** Attempting to use standardized file system operators with a
3446 pathname with invalid :DIRECTORY components signals a
3448 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
3450 ** DIRECTORY on logical pathnames is more correct.
3451 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
3452 signal a TYPE-ERROR if their argument is not a stream.
3453 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
3454 designator argument does not designate a stream.
3455 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
3456 examining the synonym.
3457 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
3459 ** OPEN and WITH-OPEN-STREAM allow opening streams with
3460 element-type larger than ([UN]SIGNED-BYTE 32).
3462 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
3463 * When built with the :SB-FUTEX feature, threaded builds now take
3464 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
3465 for faster/more reliable mutex and condition variable support.
3466 * Incompatible change (but one you probably shouldn't have been using
3467 anyway): the interface and code for arbitrating between multiple
3468 threads in the same user session has been redesigned.
3469 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
3470 error when called without an explicit environment argument.
3471 (thanks to Nikodemus Siivola)
3472 * bug fix: buffered :DIRECTION :IO streams are less likely to become
3473 confused about their position. (thanks to Adam Warner and Gerd
3475 * bug fix: Pretty printing backquoted forms with unquotations in the
3476 argument list position of various code constructs such as LAMBDA
3477 now works correctly. (reported by Paul Dietz)
3478 * bug fix: Pretty printing unquotations no longer loses all
3479 stream position information.
3480 * optimization: performance of string output streams is now less
3481 poor for multiple small sequence writes.
3482 * optimization: performance of CSUBTYPEP in the presence of complex
3483 expressions involving CONS and NOT many times has been improved.
3484 (reported by Paul Dietz)
3485 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
3487 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
3488 streams for the REPL, for the debugger, and so on. (thanks to
3490 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
3491 * fixed some bugs revealed by Paul Dietz' test suite:
3492 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
3493 is ignored if it is too small, rather than propagating through
3494 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
3495 ** extremely complex negations of CONS types were not being
3496 sufficiently canonicalized, leading to inconsistencies in
3498 ** VALUES tranformer lost derived type.
3500 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
3501 * fixed a bootstrapping bug: the build process no longer assumes
3502 that the various BOOLE-related constants have the same value in
3503 host and target lisps. (noted by Paul Dietz' test suite on an
3504 SBCL binary built from CLISP)
3505 * The system can now be dynamically linked on the MIPS platform,
3506 which enables dynamic loading of foreign code from Lisp. (thanks
3507 to Ralf Baechle for discussions on the MIPS ABI)
3508 * The system now records debugging information for its own source
3509 files in a filesystem-position-independent manner, relative to
3510 the "SYS" logical host.
3511 * fixed a compiler bug: MV-LET convertion did not check references
3512 to the "max args" entry point. (reported by Brian Downing)
3513 * tweaked disassembly notes to be less confident about proclaiming
3514 some instruction as an LRA. (thanks to Brian Downing)
3515 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
3516 now each have their own history, command character, and other
3517 characteristics. (thanks to David Lichteblau)
3518 * fixed some bugs revealed by Paul Dietz' test suite:
3519 ** compiler failure in compiling LOGAND expressions including a
3521 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3522 shift greater than 32.
3523 ** FLUSH-DEST did not mark blocks for type check regeneration.
3524 ** HANDLER-CASE failed to accept declarations in handler clauses
3525 in some circumstances.
3527 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3528 * New code in contrib/sb-introspect (still probably not entirely
3529 stable yet) provides some support for smart Lisp development
3530 environments like SLIME.
3531 * The conditions signalled for errors occurring when loading .fasl
3532 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3533 in a way which should help ASDF recover gracefully.
3534 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3535 stuff has been cleaned up. If you code contrib/ stuff, this might
3536 affect you, and you can look at contrib/README, contrib/STANDARDS,
3537 and/or the 0.8.4.27 diff to check.
3538 * In full calls the compiler now does not generate checks for declared
3539 argument types for all arguments.
3540 * various threading fixes
3541 ** and some experimental patches which didn't make it into
3542 the main tree for this release, but which are shipped in
3543 contrib/experimental-thread.patch as a possible fix for some
3544 failures (deadlock, spinning...) in GC-intensive multithreaded
3546 * fixed PPC build problem (source code incompatibility of different
3547 library versions): added offsetof() hackery which attempts to divine
3548 where glibc maintainers put uc_mcontext today
3549 * fixed bug 282: compiler does not trust type assertions while passing
3550 arguments to a full call.
3551 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3552 &OPTIONAL VALUES type parameter.
3553 * fix bug 214: algorithm for noting rejected templates is now more
3554 similar to that of template seletion. (also reported by rydis on
3556 * fixed bug 141b: printing backquoted information readably and prettily
3557 inserts a space where necessary.
3558 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3559 (CHARACTER 10) are now reported as errors, rather than propagated
3560 as unknown types. (reported by piso on #lisp)
3561 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3562 with values NIL and :ERROR. (thanks to Milan Zamazal)
3563 * fixed bug 191c: CLOS now does proper keyword argument checking as
3564 described in CLHS 7.6.5 and 7.6.5.1.
3565 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3566 longer raise an error; further, using a list as a for-as-arithmetic
3567 counter now raises a meaningful error.
3568 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3569 now understood by sequence creation functions such as MAKE-SEQUENCE
3571 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3572 a STRING-INPUT-STREAM.
3573 * compiler enhancement: SIGNUM is now better able to derive the type
3575 * type declarations inside WITH-SLOTS are checked. (reported by
3577 * fixed some bugs revealed by Paul Dietz' test suite:
3578 ** incorrect optimization of TRUNCATE for a positive first
3579 argument and negative second.
3580 ** compiler failure in let-convertion during flushing dead code.
3581 ** compiler failure while deriving type of TRUNCATE on an
3582 interval, containing 0.
3583 ** ASH of a negative bignum by a negative bignum count now returns
3585 ** intersection of CONS types now canonicalizes properly, fixing
3586 inconsistencies in SUBTYPEP.
3588 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3589 * incompatible change: The --disable-debugger command line
3590 option now clobbers the debugger at a more fundamental
3591 level, by redefining #'INVOKE-DEBUGGER instead of by
3592 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3593 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3594 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3595 enter the debugger REPL and then suffer recursive errors
3596 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3597 while under the new system, BREAK in --disable-debugger mode
3598 terminates the system just as an unhandled error would.
3599 * fixed compiler performance when processing loops with a step >1;
3600 * bug fix: DOCUMENTATION now retrieves generic function
3601 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3602 support has been systematized, and now supports the methods
3603 specified by ANSI, along with a default method and a method for
3604 slot documentation. (reported by Nathan Froyd)
3605 * bug fix: effective methods associated with a generic function are
3606 no longer cached over a change of that generic function's method
3607 combination. (reported by Andreas Fuchs)
3608 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3609 names a non-existent directory. (thanks to Andreas Fuchs)
3610 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3611 the PPC platform, lead to stack corruption; this has been fixed.
3612 (reported by Rainer Joswig)
3613 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3614 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3615 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3616 arithmetic in general, is now much more reliable on the PPC
3618 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3619 platform now returns the right answer.
3620 * optimization: restored some effective method precomputation in
3621 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3622 precomputation is now tunable.
3623 * optimization: compiler-internal data structure use has been
3624 reviewed, and changes have been made that should improve the
3625 performance of the compiler by about 20%.
3626 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3627 simple-base-strings and simple-bit-vectors is improved.
3628 * optimization: the optimization of 32-bit logical and arithmetic
3629 functions introduced in version 0.8.3 on the x86 has been
3630 implemented on the mips, ppc and sparc platforms; an
3631 implementation of the same facility, but for 64-bit arithmetic,
3632 has been added for the alpha.
3633 * microoptimization: the compiler is better able to make use of the
3634 x86 LEA instruction for multiplication by constants.
3635 * bug fix: in some situations compiler did not report usage of
3636 generic arithmetic in (SPEED 3) policy.
3637 * bug 145b fix: compiler used wrong type specifier while converting
3638 MEMBER-types to numeric.
3639 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3641 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3642 index argument now works properly on the Alpha platform.
3643 * bug fix: floating point exception treatment on the Alpha platform
3645 * bug fix: FILE-POSITION works much better on string input and
3646 output streams. (thanks to Nikodemus Siivola)
3647 * bug fix: many threading/garbage collection symptoms sorted.
3648 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3649 might be pseudo-atomic.
3650 * internal change: Stopping for GC is now done with signals not
3651 ptrace. GC is now done in whichever thread wanted it, instead of
3653 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3655 * bug fix: result form in DO is not contained in the implicit
3657 * incompatible change: ICR structure is changed; the value part of
3658 CONTINUATION is now called LVAR; corresponding functions are
3659 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3660 * added type deriver for ISQRT (thanks to Robert E. Brown).
3661 * bug fix: better support for loading from the command line when an
3662 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3664 * fixed some bugs revealed by Paul Dietz' test suite:
3665 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3666 ** ROUND and FROUND now give the right answer when given very
3667 small float arguments.
3668 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3670 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3671 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3672 ** LCM should return a non-negative integer.
3673 ** PARSE-INTEGER returned the index of a terminator instead of the
3674 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3675 ** PARSE-INTEGER returned an incorrect index being applied to a
3677 ** LCM with two arguments of 0 returns 0 rather than signalling
3679 ** unsigned addition of a 32-bit constant with the high bit set no
3680 longer causes an internal compiler error on the x86.
3681 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3682 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3683 with negative last argument.
3684 ** byte specifiers with zero size and position no longer cause
3685 an error during type derivation.
3686 ** bignum multiplication on the Alpha platform now returns the
3688 * porting: The system now builds on SuSE AMD64, although it still
3689 generates a 32-bit binary.
3690 * .fasl file incompatibility: The fasl file version number has
3691 been incremented (because of the changes to internal compiler
3692 data structures referred to above).
3694 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3695 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3696 more accurately, on the Darwin kernel running on PowerPC hardware.
3697 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3698 * Compiler code deletion notes now signal a condition of type
3699 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3700 an associated MUFFLE-WARNING restart.
3701 * The compiler now performs limited argument count validation of
3702 constant format strings in FORMAT, and where appropriate in ERROR,
3703 CERROR and WARN. (thanks to Gerd Moellmann)
3704 * New ASDF-INSTALL contrib can be used for automatic download and
3705 installation of third-party Lisp code from CCLAN or other sites
3707 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3708 which forces another thread to execute a function supplied by the
3710 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3711 now accept and act upon their :ELEMENT-TYPE keyword argument.
3712 (reported by Martin Atzmueller, Edi Weitz)
3713 * bug fix: FILE-POSITION now accepts position designators up to
3714 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3715 is the greater. (thanks to Patrik Nordebo)
3716 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3717 Kalle Olavi Niemitalo)
3718 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3719 printable. (reported by Eric Marsden)
3720 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3721 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3722 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3723 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3724 (thanks to Stig Sandoe)
3725 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3726 UNREAD-CHAR now correctly includes the unread character in the
3727 target sequence. (thanks to Gerd Moellmann)
3728 * bug fix in threaded builds: the system can now be suspended and
3729 resumed by shell job control with minimal disruption.
3730 * bug fixes in times and timezones >2038AD
3731 * better handling of "where is GNU make?" problem in build scripts
3732 (thanks to Nikodemus Siivola)
3733 * new optimization: inside a named function any reference to a
3734 function with the same name is considered to be a self-reference;
3735 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3736 optimization quality.
3737 * new optimization on x86: logical functions and + now have
3738 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3739 used when the result is truncated to 32 bits.
3740 * VALUES declaration is partially enabled.
3741 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3742 * fixed some bugs revealed by Paul Dietz' test suite:
3743 ** The system now obeys the constraint imposed by
3744 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3745 types form a lattice under type intersection.
3746 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3747 ** ASSOC now ignores NIL elements in an alist.
3748 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3749 and (1+ MOST-POSITIVE-FIXNUM) answers.
3750 ** The addition of a method with invalid qualifiers to a generic
3751 function does not cause an error to be signalled immediately;
3752 a warning is signalled, and the error is generated only on
3753 calling the generic function.
3754 * changed .fasl file version number, in part to add type codes for
3755 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3756 obscure ANSI requirements
3758 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3759 * fixed bug 148: failure to inline-expand a local function left
3760 garbage, confusing the compiler.
3761 * fixed bugs 3cd: structure slot readers perform type check if the
3762 slot can have an invalid value (i.e. it is either not initialized
3763 or can be written with a less specific slot writer).
3764 * bug fix: the compiler now traps array references to elements off
3765 the end of an array; previously, the bounds checking in some
3766 circumstances could go off-by-one.
3767 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3769 * type declarations for array element types now obey the description
3770 on the CLHS page "Declaration TYPE", as per discussions on
3771 sbcl-help around 2003-05-08. This means that a declaration
3772 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3773 declaration, all references to BAR will be asserted or assumed
3774 (with THE, so dependent on compiler policy) to involve objects of
3775 type FOO. Note that no such declaration is implied in
3776 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3777 * declared types of functions from the "Conditions"
3778 chapter. (reported by Paul Dietz)
3779 * bug fix: CERROR accepts a function as its first argument.
3780 * bug fix: NTH an NTHCDR accept a bignum as index
3781 arguments. (reported by Adam Warner)
3782 * optimization: character compare routines now optimize comparing
3783 against a constant character. (reported by Gilbert Baumann)
3784 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3785 argument now works properly on the MIPS platform.
3786 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3787 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3788 declared to have a functional type, does not cause an error
3790 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3791 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3792 * fixed bug: if last continuation of a deleted block has a
3793 destination, this destination should be deleted too. (reported by
3795 * fixed a bug in the bootstrap process: the host compiler's values
3796 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3797 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3798 test case from Patrik Nordebo)
3799 * improved the ability of the disassembler on the PPC platform to
3800 provide helpful disassembly notes.
3801 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3802 the class in more cases than previously.
3803 * bug fix: FILE-POSITION now understands :START and :END for
3804 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3805 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3806 8) data. (thanks to David Lichteblau)
3807 * bug fix: it is possible to add a method to a generic function
3808 without lambda list.
3809 * bug fix: reader failed to signal END-OF-FILE inside an
3810 object representation. (reported by Nikodemus Siivola)
3811 * fixed some bugs revealed by Paul Dietz' test suite:
3812 ** LAST and [N]BUTLAST should accept a bignum.
3813 ** condition slot accessors are methods.
3814 ** (VECTOR NIL) is a subtype of STRING.
3816 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3817 * minor incompatible change: some nonsensical specialized lambda
3818 lists (used in DEFMETHOD) which were previously ignored now signal
3820 * minor incompatible change: the system is now aware of the types of
3821 variables in the COMMON-LISP package, and will signal errors for
3822 most violations of these type constraints (where previously they
3823 were silently accepted).
3824 * minor incompatible change: COMPILE-FILE now uses the freedom
3825 afforded (ANSI 3.2.2.3) to use derived function types for
3826 functions defined in the same file. This also permits the system
3827 to warn on static type mismatches and function
3828 redefinition. (Currently it does not work with high DEBUG level.)
3829 * minor incompatible change: VALUES declaration is disabled.
3830 * When issuing notes, the compiler now signals a condition of type
3831 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3832 restart for use in user handlers. It is expected that the
3833 COMPILER-NOTE condition will eventually become a condition
3834 supertype to a hierarchy of note types, which will then be
3835 handleable in a similar fashion. However, at the moment, no such
3836 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3837 but it's an implementation detail, not a classification for the
3839 * Changes in type checking closed the following bugs:
3840 ** type checking of unused values (192b, 194d, 203);
3841 ** template selection based on unsafe type assertions (192c, 236);
3842 ** type checking in branches (194bc).
3843 * A short form of VALUES type specifier has ANSI meaning (it has
3844 increased the number of situations when SBCL cannot perform type
3846 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3847 keywords or constants is permissible.
3848 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3849 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3850 classes. (thanks to Antonio Martinez)
3851 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3852 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3853 argument to a pretty-print stream. Instead, it leaves any such
3854 operations to DESCRIBE-OBJECT methods.
3855 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3857 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3858 lambda lists. (thanks to David Lichteblau)
3859 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3860 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3861 (thanks to Gerd Moellmann)
3862 * fixed bug 246: increased compilation speed of long
3863 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3865 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3867 * DEFCONSTANT now throws a condition of type
3868 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3869 constant to a non-EQL value; CONTINUE and ABORT restarts
3870 respectively change and preserve the value.
3871 * fixed bug 63: The code walker, part of the implementation of CLOS,
3872 is now better at handling symbol macros.
3873 * bug fix: There is no longer an internal implementation type named
3874 CL:LENGTH. (reported by Raymond Toy)
3875 * bug fix: In macro-like defining macros/special operators the
3876 implicit block does not enclose the lambda list.
3877 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3878 enclosed in parentheses) are not suitable as type specifiers, and
3879 their use properly signals an error now.
3880 * bug fix: An argument count mismatch for a type specifier in code
3881 being compiled no longer causes an unhandled error at compile
3882 time, but signals a compile-time warning.
3883 * fixed simple vector readable printing
3884 * bug fix: DESCRIBE takes more care over whether the class
3885 precedence list slot of a class is bound before accessing it.
3886 (reported by Markus Krummenacker)
3887 * bug fix: FORMATTER can successfully compile pretty-printer format
3888 strings which use variants of the ~* directive inside.
3889 * bug fix: SEARCH now applies its TEST predicate to the elements of
3890 the arguments in the correct order. (thanks to Wolfhard Buss)
3891 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3892 in type inference. (thanks to Robert E. Brown)
3893 * bug fix: Reading in symbols with an explicit package name of ""
3894 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3895 * fixed some bugs revealed by Paul Dietz' test suite:
3896 ** NIL is now allowed as a structure slot name.
3897 ** Arbitrary numbers, not just REALs, are allowed in certain
3898 circumstances in LOOP for-as-arithmetic clauses.
3899 ** Multiple class redefinitions before slot access no longer
3900 causes a type error.
3901 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3902 association between the name and a class.
3903 ** Generic functions with non-standard method-combination and over
3904 five methods all of which return constants no longer return NIL
3905 after the first few invocations. (thanks to Gerd Moellmann)
3906 ** CALL-NEXT-METHOD with no arguments now passes the original
3907 values of the arguments, even in the presence of assignment.
3908 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3910 ** DISASSEMBLE works with closures and funcallable instances.
3911 ** ADD-METHOD now returns the generic function, not the new method.
3912 ** FIND-METHOD signals an error if the lengths of the specializers
3913 is incompatible with the generic function, even if the ERRORP
3915 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3916 which its argument is a member.
3917 ** DEFCLASS only redefines the class named by its class-name
3918 argument if that name is the proper name of the class;
3919 otherwise, it creates a new class.
3920 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3921 of the UNBOUND-SLOT condition to the name of the slot.
3922 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3923 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3924 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3925 SLOT-MAKUNBOUND in the specified fashion.
3927 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3928 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3929 cross-compilation host. As a consequence, we can now bootstrap our
3930 way up to SBCL starting with a bare gcc toolchain and human-readable
3931 source code (first the source to CLISP, then the source to SBCL).
3932 * A contributed module containing a partial implementation of the
3933 simple-streams interface has been included. (thanks to Rudi
3935 * A contributed module implementing the RFC1321 Message Digest
3936 Algorithm, known as MD5, has been included.
3937 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3938 no longer has any effect, as the code controlled by this feature
3939 has been deleted. (As far as we know, no-one has ever built using
3940 this feature, and its semantics were confused in any case).
3941 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3942 work on conditions (as required by the ANSI specification),
3943 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3944 expected behaviour on conditions. Users should note, however,
3945 that such behaviour is not required by the ANSI specification,
3946 and so use of this behaviour may render their code unportable.
3947 * fixed some bugs revealed by Paul Dietz' test suite:
3948 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3950 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3951 ** SXHASH on condition objects no longer returns NIL.
3952 ** :ALLOCATION :CLASS slots are better treated; their values are
3953 updated on class redefinition, and initforms inherited from
3954 superclasses are applied.
3955 ** REMOVE-METHOD returns its generic function argument even when
3956 no method was removed.
3957 ** SHARED-INITIALIZE now initializes the values of the requested
3958 slots, including those with :ALLOCATION :CLASS.
3959 ** ALLOCATE-INSTANCE now works on structure classes defined via
3960 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3962 ** SLOT-EXISTS-P now works on conditions, as well as structures
3964 ** MAKE-LOAD-FORM now has the required methods on
3965 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3966 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3967 keyword, and now implements the SLOT-NAMES argument.
3968 ** methods with &OPTIONAL arguments no longer allow too many
3969 arguments to be passed in the call without error.
3970 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3971 option is consistent with the required arguments of the generic
3972 function lambda list.
3973 * bug fix: REQUIRE accepts a string designator. (Thanks to
3975 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3976 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3977 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3978 * bug fix: The compiler now checks for duplicated variables in macro
3980 * bug fix: SETQ on globals returns the correct value.
3981 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3982 not a COMPILER-ERROR (followed by some other strange error on
3983 choosing the CONTINUE restart).
3984 * bug fix: make.sh and friends are now more consistent in the way that
3985 they look for GNU "make".
3987 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3988 * experimental native threads support (on x86 Linux >=2.4 only).
3989 This is not compiled in by default: you need to add :SB-THREAD to
3990 the target features. See the "Beyond ANSI" chapter of the manual
3992 * fix for longstanding nonANSIism: The old distinction between
3993 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3994 The return value from CL:FIND-CLASS is now a CLOS class, and
3995 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3996 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3997 * An interface to the MetaObject Protocol, as described in Kiczales,
3998 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3999 MIT Press, 1991, is available from the SB-MOP package.
4000 * incompatible change: the SB-PCL package should now be considered
4001 a private implementation detail, and no longer a semi-private MOP
4003 * minor incompatible change: due to rearrangement for threads, the
4004 control stack and binding stack are now allocated at arbitrary
4005 addresses instead of being hardcoded per-port. Users affected by
4006 this probably have to be doing advanced things with shared
4007 libraries, and will know who they are.
4008 * minor incompatible change: Previously, all --eval forms used were
4009 processed with READ before any of them were processed with EVAL.
4010 Now each --eval form is processed with both READ and EVAL before
4011 the next --eval form is processed. (Thus package operations like
4012 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
4013 work as the user might reasonably expect.)
4014 * minor incompatible change: *STANDARD-INPUT* is now only an
4015 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
4017 * minor incompatible change: Y-OR-N-P is now character-oriented,
4018 not line oriented. Also, YES-OR-NO-P now works without errors.
4019 (thanks to Antonio Martinez)
4020 * sb-aclrepl module improvements: an integrated inspector, added
4021 repl features, and a bug fix to :trace command.
4022 * Known functions, which cannot be open coded by the backend, are
4023 considered to be able to check types of their arguments. (fixing
4024 a bug report by Nathan J. Froyd)
4025 * fixed a bug in computing method discriminating functions: It is
4026 now possible to define methods specialized on classes which have
4027 forward-referenced superclasses. (thanks to Gerd Moellmann)
4028 * fixed evaluation order in optional entries (reported by Gilbert
4030 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
4031 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
4032 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
4033 required-by-AMOP NAME argument, as well as CLASS and
4034 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
4035 * fixed bug 20: DEFMETHOD can define methods using names that are
4036 not the proper names of classes to designate class specializers.
4037 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
4038 Unix file descriptors, instead of blowing up. (thanks to Antonio
4040 * Garbage collection refactoring: user-visible change is that a
4041 call to the GC function during WITHOUT-GCING will not do garbage
4042 collection until the end of the WITHOUT-GCING. If you were doing
4043 this you were probably losing anyway.
4044 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
4045 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
4046 TYPEP the latter but not the former.
4047 * The compiler issues a full WARNING for calls to undefined functions
4048 with names from the CL package.
4049 * MAP-INTO for a vector destination is open coded. (reported by
4050 Brian Downing on c.l.l)
4051 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
4052 documentation string.
4053 * fixed some bugs revealed by Paul Dietz' test suite:
4054 ** COPY-ALIST now signals an error if its argument is a dotted
4056 ** Condition slots are now accessed more correctly in the presence
4057 of multiple initargs for a given slot.
4058 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
4059 exclude restarts of the same name associated with a different
4061 ** DEFCLASS of forward-referenced classes with another
4062 forward-referenced class in the superclasses list no longer
4064 ** Condition slots are now initialized once each, not multiple
4065 times. (thanks to Gerd Moellmann)
4066 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
4068 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
4069 ** NSET-EXCLUSIVE-OR does not return extra elements when its
4070 arguments contain duplicated elements.
4071 ** RESTART-CASE understands local macros.
4072 ** RESTART-CASE associates exactly its own restarts with a condition.
4073 ** ENDP in safe mode checks its argument to be of type LIST.
4074 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
4075 in question is unbound.
4076 ** Optimized MAKE-INSTANCE functions no longer cause internal
4077 assertion failures in the presence of duplicate initargs.
4078 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
4079 ** Side-effectful :DEFAULT-INITARGS have their side-effects
4080 propagated even in the ctor optimized implementation of
4082 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
4084 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
4085 * a better implementation of SXHASH on (simple) bit vectors,
4086 measured both in execution speed and in distribution of results
4087 over the positive fixnums, has been installed. Likewise, a better
4088 implementation of EQUAL for simple bit vectors is now available.
4089 * fixed CEILING optimization for a divisor of form 2^k.
4090 * fixed bug 240 (emitting extra style warnings "using the lexical
4091 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
4092 by Antonio Martinez)
4093 * fixed SXHASH, giving different results for NIL depending on type
4094 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
4095 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
4096 variable. (found by Rolf Wester)
4097 * SBCL does not ignore type declarations for special
4098 variables. (reported by rif on c.l.l 2003-03-05)
4099 * some bug fixes in contrib/sb-aclrepl/
4100 * fixed some bugs revealed by Paul Dietz' test suite:
4101 ** a bug in the CONS type specifier, whereby the CAR and CDR
4102 types got intertwined, has been fixed;
4103 ** the type system is now able to reason about the interaction
4104 between INTEGER and RATIO types more completely;
4105 ** APPEND, [N]REVERSE and NRECONC check that those their
4106 arguments, which must be proper lists, are really so;
4107 ** An array specialized to be unable to hold elements has been
4108 implemented, as required -- yes, really -- by ANSI;
4109 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
4110 on malformed property lists;
4112 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
4113 * incompatible packaging change: in line with Unix convention,
4114 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
4115 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
4116 * REQUIRE and PROVIDE are now optionally capable of doing something
4117 useful. See the documentation string for REQUIRE.
4118 * infrastructure for a managed SBCL contrib system: contributed
4119 modules in this release include:
4120 ** the ASDF system definition facility;
4121 ** an interface to the BSD Sockets API;
4122 ** an ACL-like convenience interface to the repl;
4123 (thanks to Kevin Rosenberg)
4124 ** an implementation of ROTATE-BYTE, with efficient implementation
4126 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
4127 gives a closer approximation to the right answer than previously.
4128 (thanks to Raymond Toy)
4129 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
4130 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
4131 optional environment arguments, as required by ANSI.
4132 * fixed bugs in other functions taking environment objects, allowing
4133 calls with an explicit NIL environment argument to be compiled
4135 * fixed bug 228: primary return values from
4136 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
4137 COMPILE or FUNCTION.
4138 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
4139 :TYPE will no longer signal errors on innocuous objects.
4140 * fixed bug 231b: SETQ is better at respecting type declarations in
4141 the lexical environment.
4142 * fixed a bug in DEFCLASS: classes named by symbols with no or
4143 unprintable packages can now be defined.
4144 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
4145 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
4146 * fixed compiler failure related to checking types of functions.
4147 (reported by Robert E. Brown)
4148 * the compiler is now much more consistent in its error-checking
4149 treatment of bounding index arguments to sequence functions: in
4150 (SAFETY 3) code, errors will be signalled in almost all cases if
4151 invalid sequence bounding indices are passed to functions defined
4152 by ANSI to operate on sequences.
4153 * fixed a bug in the build procedure: documentation of SBCL-specific
4154 packages is now preserved and available in the final Lisp image.
4155 * lifted FDEFINITION lookup out of loops in the implementation of
4156 many list operations. (thanks to Robert E. Brown)
4157 * fixed a bug in the reader: the #n# reader macro now works for
4158 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
4159 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
4160 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
4161 E. Brown and Tony Martinez respectively)
4162 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
4163 SLOT-BOUNDP: in particular, it is now able to optimize them much
4164 better, and is now not vulnerable to having packages renamed.
4165 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
4166 always called when a slot is not present in an instance. (thanks
4168 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
4169 longer signal an error on unreading a character following EOF on
4170 the previous constituent stream. (thanks to Tony Martinez)
4171 * fixed some bugs revealed by Paul Dietz' test suite:
4172 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
4173 not just nonnegative fixnums;
4174 ** the logical bit-array operators such as BIT-AND now accept an
4175 explicit NIL for their "opt-arg" argument (to indicate a
4176 freshly-consed result bit-array);
4177 ** ELT now signals an error on an invalid sequence index in safe
4179 ** the type system is now cleverer about negations of numeric
4180 types, and consequently understands the BIGNUM and RATIO types
4182 ** the type system is now cleverer about the interaction between
4183 INTEGER and RATIO types: while bugs still remain, many more
4184 cases are accurately computed;
4185 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
4186 if it is in the last clause;
4187 ** CONSTANTLY now correctly returns a side-effect-free function in
4189 ** DECLARE is no longer treated as a special-operator; in
4190 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
4191 * incremented fasl file version number due to the change in the
4192 DEFSTRUCT-SLOT-DESCRIPTION structure.
4194 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
4195 * minor incompatible change: code processed by the "interpreter" or
4196 EVAL now has a compilation optimization policy of (DEBUG 2)
4197 (changed from (DEBUG 1)) to improve debuggability of interactive
4198 development, and to allow the use of the debug RETURN command in
4200 * an experimental implementation of the RETURN command for the
4201 debugger has been included. (thanks to Frederik Kuivinen)
4202 * fixed bug 62: constraints were not propagated into a loop.
4203 * fixed bug in embedded calls of SORT (reported and investigated by
4205 * fixed some bugs revealed by Paul Dietz' test suite:
4206 ** printing and reading of arrays with some dimensions having
4207 length 0 (thanks to Gerd Moellmann);
4208 ** BOA constructor with &AUX argument without a default value does
4209 not cause a type error;
4210 ** CONSTANTP now returns true for all self-evaluating objects.
4212 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
4213 * fixed bug 127: DEFSTRUCT now does not clobber old structure
4214 accessors that are related by inheritance, as specified in the
4215 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
4216 Valtteri Vuorikoski)
4217 * The compiler is now able to inline functions that were defined in
4218 a complex lexical environment (e.g. inside a MACROLET).
4219 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
4221 * Support for the upcoming FreeBSD-5.0 release has been included.
4222 (thanks to Dag-Erling Smorgrav)
4223 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
4224 effect when it is not in a toplevel context.
4225 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
4226 stand a better chance of being correct. (thanks to Gerd
4228 * fixed bug in COERCE, which now signals an error on coercing a
4229 rational to a bounded real type which excludes the expected
4231 * The compiler is now able to derive types more accurately from the
4232 COERCE and COMPILE functions.
4233 * fixed bug 223: functional binding is considered to be constant
4234 only for symbols in the CL package.
4235 * fixed bug 231: SETQ did not check the type of a variable being set
4236 (reported by Robert E. Brown)
4237 * A new optimization for MAKE-INSTANCE has been included, fixing
4238 various bugs (including relating to :ALLOCATION :CLASS slots and
4239 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
4241 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
4242 ** As required by ANSI, LOOP now disallows anonymous collection
4243 clauses such as COLLECT I in conjunction with aggregate boolean
4244 clauses such as THEREIS (= I 1);
4245 ** LOOP now signals an error when any variable is reused in the
4246 same loop (including the potentially useful construct analogous
4247 to WITH A = 1 WITH A = (1+ A);
4248 ** IT is only a special loop symbol within the first clause of a
4249 conditional loop clause;
4250 ** LOOP with a typed iteration variable over a hashtable now
4251 signals a type error iff it should.
4252 * fixed some other bugs revealed by Paul Dietz' test suite:
4253 ** FILE-STREAM now names the class previously known as FD-STREAM;
4254 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
4255 argument) no longer signals an error;
4256 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
4257 of the slot symbol, rather than using the current package
4258 ((:CONC-NAME "") continues to intern the slot's name in the
4260 * incremented fasl file version number, because of the incompatible
4261 change to the DEFSTRUCT-DESCRIPTION structure, and again because
4262 of the new implementation of DEFINE-COMPILER-MACRO.
4264 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
4265 * Support for building SBCL for MIPS platforms running in
4266 little-endian mode has now been checked in, and basic
4267 functionality on said platforms verified.
4268 * minor incompatible change: PCL now records the pathname of a file
4269 in which methods and the like are defined, rather than its
4271 * minor incompatible change: TRUENAME now considers the truename of
4272 a file naming a directory to be the pathname with :DIRECTORY
4273 component indicating that directory.
4274 * minor incompatible change: a NAMED clause in the extended form of
4275 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
4276 reason for the previous behaviour is unclear.
4277 * more systematization and improvement of CLOS and MOP conformance
4278 in PCL (thanks to Gerd Moellman and Pierre Mai):
4279 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
4281 ** DEFINE-METHOD-COMBINATION no longer signals an error for
4282 primary methods with no specializers;
4283 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
4285 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
4286 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
4287 FORWARD-REFERENCED-CLASSes; error reporting on
4288 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
4290 ** SXHASH on CLOS instances now uses a slot internal to the
4291 instance to return different numbers on distinct instances,
4292 while preserving the same return value through invocations of
4294 ** DEFMETHOD signals errors when methods with longer incongruent
4295 lambda lists are added to generic functions;
4296 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
4297 CLASS, as specified in AMOP;
4298 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
4299 on the order returned by the primary method for classes of
4300 class STANDARD-CLASS;
4301 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
4302 * fixed some bugs shown by Paul Dietz' test suite:
4303 ** DOLIST puts its body in TAGBODY;
4304 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
4306 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
4307 value producing form;
4308 ** if more variables are given to PROGV than values, extra
4309 variables are bound and made to have no value;
4310 ** NSUBSTITUTE on list arguments gets the right answer with
4312 ** ELT signals an error of type TYPE-ERROR when the index argument
4313 is not a valid sequence index;
4314 ** LOOP signals (at macroexpansion time) an error of type
4315 PROGRAM-ERROR when duplicate variable names are found;
4316 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
4317 ** FUNCALL of special-operators now cause an error of type
4319 ** PSETQ now works as required in the presence of side-effecting
4320 symbol-macro places;
4321 ** NCONC accepts any object as its last argument;
4322 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
4324 ** loop-for-as-package does not require a package to be explicitely
4326 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
4328 * fixed bug 166: compiler preserves "there is a way to go"
4329 invariant when deleting code.
4330 * fixed bug 172: macro lambda lists with required arguments after
4331 &REST arguments now cause an error to be signalled. (thanks to
4333 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
4334 bug. (thanks to Matthew Danish)
4335 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
4337 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
4338 arguments are lexically rebound. (thanks to Gerd Moellmann and
4340 * fixed bug 194: error messages are now more informative when there
4341 is no primary method applicable in a call to a generic
4342 function. (thanks to Gerd Moellmann)
4343 * fixed bug in command line argument checking (thanks to Julian
4345 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
4346 (thanks to Matthew Danish)
4347 * incremented fasl file version number, because of the
4348 SXHASH-related changes in the layout of CLOS data structures
4350 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
4351 * minor incompatible change: The runtime (the Unix executable named
4352 "sbcl") is now much pickier about the .core files it will load.
4353 Essentially it now requires .core files to descend from the same
4354 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
4355 as the runtime does. (The intent is to prevent the crashes which
4356 can occur, and which can even be reported as mysterious failures,
4357 when people patch the sources or change the build parameters
4358 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
4359 sbcl and .core files.)
4360 * fixed bug: VALUES-LIST is no longer optimized away.
4361 * fixed bug 142: The FFI conversion of C string values to Lisp
4362 string values no longer conses excessively. (thanks to Nathan
4363 Froyd porting Raymond Toy's fix to CMU CL)
4364 * began to systematize and improve MOP conformance in PCL (thanks to
4365 Nathan Froyd, Gerd Moellman and Pierre Mai):
4366 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
4368 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
4369 ** FINALIZE-INHERITANCE is now called on class finalization;
4370 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
4371 argument precedence order.
4372 * fixed bug 202: The compiler no longer fails on functions whose
4373 derived types contradict their declared type.
4374 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
4375 so it can be non-toplevel.
4376 * The fasl file version number has changed (because of the new
4377 implementation of DEFMACRO).
4378 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
4379 safe code, that any length requirement by their type-specifier
4380 argument is valid. The exceptions to this are described in bug
4382 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
4383 functions now check the types of their inputs as required by ANSI.
4384 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
4385 introduced symbol is DECLAREd to be SPECIAL.
4386 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
4387 * fixed bug: :COUNT argument to sequence functions may be negative
4388 * fixed bug: body of DO-SYMBOLS may contain declarations
4389 * fixed bug: PUSHNEW now evaluates its arguments from left to right
4390 (reported by Paul F. Dietz, fixed by Gerd Moellman)
4391 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
4392 symbol macro only once
4393 * fixed printing of call frame when argument list is unavailable
4394 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
4395 * compiler no longer signals WARNING on unknown keyword
4398 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
4399 * A beta-quality port to the mips architecture running Linux,
4400 based on the old CMUCL backend, has been made. It has been tested
4401 on a big-endian kernel, and works sufficiently well to be able to
4402 rebuild itself; it has not been tested in little-endian mode.
4403 * fixed an inconsistency between gencgc.c and purify.c which made
4404 dumping/loading .core files unreliable
4405 * fixed bug 120a: The compiler now deals correctly with IFs where
4406 the consequent is the same as the alternative, instead of
4407 misderiving the return type. (thanks to Alexey Dejneka)
4408 * fixed bug 113: Logical pathnames are now dumpable (the logical
4409 host is resolved at load-time, throwing an error if it is not
4411 * fixed bug 174: FORMAT's error message is slightly clearer when a
4412 non-printing character is used in a format directive.
4413 * fixed several bugs in compiler checking of type declarations, i.e.
4414 violations of the Python "declarations are assertions" principle
4415 (thanks to Alexey Dejneka)
4416 * fixed several bugs in PCL's error checking (thanks to Gerd
4418 * fixed bug: printing of FILE-ERROR (thanks to Antonio
4420 * fixed bug in compilation of functions as first class values
4421 (thanks to Antonio Martinez-Shotton)
4422 * The compiler's handling TYPE-ERRORs which it can prove will
4423 inevitably happen at runtime has been cleaned up and corrected
4424 in several ways. (thanks to Alexey Dejneka)
4425 * improved argument type checking for various basic arithmetic
4426 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
4427 lavished on them in the past that they can be compiled in many
4428 ways in different special cases
4429 * fixed bug 181: compiler checks validity of user supplied type
4431 * cleaned up code flushing in optimization: Function calls which
4432 should signal errors for safety purposes (e.g. which ANSI says
4433 should signal errors when their arguments are of incorrect type)
4434 are no longer optimized away.
4435 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
4436 * incremented fasl file version number, because changes in the
4437 implementation of sequence functions like COERCE caused
4438 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
4439 in old inline expansions) to become undefined. (Actually these
4440 changes were later undone, so we might very well be binary
4441 compatible with 0.7.7 after all, but leaving the version number
4442 incremented seemed like the simplest and most conservative
4445 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
4446 * An alpha-quality port to the parisc architecture running Linux,
4447 based on the old CMUCL backend, has been made. This, even more so
4448 than the other backends, should be considered still a work in
4449 progress; known problems include that the Linux kernel in 64-bit
4450 mode does not propagate the correct sigcontext structure to
4451 userspace, and consequently SBCL on a parisc64 kernel will not
4453 * fixed bug 189: The compiler now respects NOTINLINE declarations for
4454 functions declared in FLET and LABELS. (I.e. "LET conversion" is
4455 suppressed.) Also now that the compiler is looking at declarations
4456 in the environment, it checks optimization declarations as well,
4457 and suppresses inlining when (> DEBUG SPEED).
4458 * More fixes have been made to treatment of floating point exception
4459 treatment and other Unix signals. In particular, floating point
4460 exceptions no longer cause Bus errors on the SPARC/Linux platform.
4461 * The detection and handling of control stack exhaustion (infinite
4462 or very deeply nested recursion) has changed. Stack exhaustion
4463 detection is now done by write-protecting pages at the OS level
4464 and applies at all optimization settings; when found, a
4465 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
4466 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
4467 be caught using IGNORE-ERRORS.
4468 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
4469 correctly when transforming recursive LABELS functions to LETs.
4470 (thanks to Alexey Dejneka)
4471 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
4472 that are names of constants or global variables.
4473 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
4474 alien routines with docstrings.
4475 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
4476 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
4478 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
4479 (thanks to Eric Marsden)
4480 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
4481 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
4482 * Bug fix: The SPARC backend can now compile functions involving
4483 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
4484 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
4485 to a functions expecting a PATHNAME-DESIGNATOR.
4486 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
4487 lambda lists. (thanks to Alexey Dejneka)
4488 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
4489 functions so that EXPORTing the name of the class doesn't cause
4490 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
4491 to Antonio Martinez for reporting this)
4492 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
4493 constant-folded without failing an assertion. (thanks to Einar
4494 Floystad Dorum for reporting this)
4495 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
4496 types are now dealt with more correctly by the compiler.
4497 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
4498 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
4499 behaviour analogous to RENAME-FILE. This puts its behaviour more
4500 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
4502 * The fasl file version number has changed again. (because of the
4503 bug fix involving the names of PCL MAKE-INSTANCE functions)
4505 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
4506 * bug fix: Floating point exceptions are treated much more
4507 consistently on the x86/Linux and PPC/Linux platforms.
4508 * Array initialization with :INITIAL-ELEMENT is now much faster for
4509 cases when the compiler cannot open code the array creation, but
4510 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
4511 array accesses have also seen a speed increase.
4512 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
4513 specified. (This was at the root of some bad interactions between
4514 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
4516 * bug fix: Internal error arguments for undefined functions are now
4517 computed correctly on the PPC/Linux platform.
4518 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4519 Raymond Toy's patch for CMU CL)
4520 * Support for the Solaris 9 operating environment has been included
4521 (thanks to Daniel Merritt)
4522 * A very ugly but hopefully complete draft of the missing FFI chapter
4523 of the manual has been created by reformatting the corresponding
4524 CMU CL manual chapter into (currently very ugly and incoherent)
4525 DocBook and bringing it up to date for SBCL behavior. Thus, the
4526 manual is now essentially complete, at least by my extreme
4527 once-and-only-once standards, whereby it's acceptable to refer to
4528 the doc strings of SB-EXT functions as the primary documentation.
4529 * The fasl file version number has changed again, due to cleanup of
4530 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4531 is no longer a static symbol.)
4533 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4534 * SBCL now builds with OpenMCL (version 0.12) as the
4535 cross-compilation host; also, more progress has been made toward
4536 bootstrapping under CLISP.
4537 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4539 * bug 158 fixed: The compiler can now deal with integer loop
4540 increments different from 1; fixing this turned out also to fix
4542 * bug 169 fixed: no more bogus warnings about using lexical bindings
4543 despite the presence of perfectly good SPECIAL declarations (thanks
4544 to David Lichteblau)
4545 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4546 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4547 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4549 * bug 180 fixed: Method combination specifications no longer ignore
4550 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4551 * bug fix: Structure type predicate functions now check their argument
4552 count as they should.
4553 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4554 correctly. (thanks to Pierre Mai)
4555 * minor incompatible change: The --noprogrammer option is deprecated
4556 in favor of the new --disable-debugger option, which is very similar.
4557 (The major difference is that it takes effect at a slightly different
4558 time at startup, causing handling of errors in --sysinit and
4559 --userinit files will be affected differently.) The
4560 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4561 been added to allow this functionality to be controlled from ordinary
4562 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4563 maintainers, who might want to run non-interactive scripts to
4564 build SBCL cores which will later be used interactively.)
4565 * minor incompatible change: The LOAD function no longer, when given
4566 a wild pathname to load, loads all files matching that pathname.
4567 Instead, an error of type FILE-ERROR is signalled.
4569 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4570 * bug 147 fixed: The compiler preserves its block link/count
4571 invariants more correctly now so that it doesn't crash. (thanks
4573 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4575 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4576 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4577 no-value-for-this-character as NIL (as specified by ANSI).
4578 * HOST-NAMESTRING on physical pathnames now returns a string that is
4579 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4580 (thanks to Christophe Rhodes)
4581 * The Alpha port handles icache flushing more correctly. (thanks to
4583 * More progress has been made toward bootstrapping under CLISP. (thanks
4584 to Christophe Rhodes)
4585 * The fasl file format has changed again, because dynamic loading
4586 on OpenBSD (which has non-ELF object files) motivated some cleanups
4587 in the way that foreign symbols are transformed and passed around.
4588 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4589 is no longer treated as whitespace by the reader, but instead as
4590 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4591 |A
\7fB|, instead of A as it used to.
4593 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4594 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4595 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4596 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4597 this as of 0.7.1.45, but was left out of the previous news section
4598 (thanks to Dan Barlow)
4599 * SBCL now runs on the Solaris operating system on SPARC architectures
4600 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4601 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4602 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4603 * SPARC backend cleanups, allowing builds of cores optimized for V8
4604 and V9 SPARCS, and also emission of code targeted to a particular
4605 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4607 * SBCL is closer to bootstrapping under CLISP, thanks to various
4608 fixes by Christophe Rhodes.
4609 * The fasl file format has changed again, to allow the compiler's
4610 INFO database to support symbol macros.
4611 * The user manual (in doc/) is formatted into HTML more nicely.
4612 (thanks to coreythomas)
4613 * The system is smarter about SUBTYPEP relationships, especially
4614 those involving NOT types (including types such as ATOM which are
4615 represented internally using NOT types). Thus SUBTYPEP is less
4616 likely to return (VALUES NIL NIL) in general, and in particular
4617 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4619 * The fasl file format has changed again, because the internal
4620 representation of types now includes a new slot to support the new
4621 SUBTYPEP-of-NOT-types logic.
4622 * (not a change in the main branch of SBCL, but a related prototype
4623 which can hopefully be merged into the main branch of SBCL in the
4624 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4625 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4626 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4627 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4628 dispatch macro character combinations. (thanks to Alexey Dejneka)
4629 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4630 that superficially look like logical namestrings correctly.
4631 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4633 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4634 default supplied-p) for &optional and &key arguments. (thanks to
4637 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4638 * incompatible change: The compiler is now less aggressive about
4639 tail call optimization, doing it only when (> SPACE DEBUG) or
4640 (> SPEED DEBUG). (This is an incompatible change because there are
4641 programs which relied on the old CMU-CL-style behavior to optimize
4642 away their unbounded recursion which will now die of stack overflow.)
4643 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4644 for non-GENCGC systems has been increased to 20M (since that
4645 seems much closer to the likely performance optimum for modern
4646 systems than the old 4M value was)
4647 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4648 and *DEBUG-PRINT-LEVEL*
4649 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4650 of CMU CL's support for SPARC, and various endianness and other
4651 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4652 * new syntactic sugar for the Unix command line: --load foo.bar is now
4653 an alternate notation for --eval '(load "foo.bar")'.
4655 ** The system now detects stack overflow and handles it gracefully,
4656 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4657 optimization settings. (This is a good thing in general, and
4658 its introduction in this version should be particularly timely
4659 for anyone whose code fails because of suppression of tail
4661 ** The system now hunts for the C variable "environ" in a more
4662 devious way, to avoid segfaults when the C library version differs
4663 between compile time and run time. (thanks to Christophe Rhodes)
4664 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4665 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4666 ** The compiler no longer issues bogus style warnings for undefined
4667 classes in the same source file as the DEFCLASSes which defined
4668 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4670 ** fixes in CONDITION class precedence list for undefined function
4671 errors (thanks to Alexei Dejneka)
4672 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4673 correctly. (thanks to Dan Barlow)
4674 ** portability fixes aiming at bootstrapping under CLISP (thanks
4675 to Dave McDonald and Christophe Rhodes)
4676 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4677 ** fixes in type translation and and type inference (thanks to
4679 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4680 ** various fixes in the new ports (thanks to Dan Barlow)
4681 * several changes related to debugging:
4682 ** suppression of tail recursion, as noted above
4683 ** stack overflow detection, as noted above
4684 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4685 is now the default. (For some time encapsulation has been more
4686 reliable than the breakpoint-based :ENCAPSULATE NIL
4687 implementation, at least on X86 systems; and I just noticed that
4688 encapsulation also seems closer to the spirit of the ANSI
4691 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4693 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4694 up properly again. (There was a packaging bug in 0.7.0 which
4695 left their definitions in SB-SYS::LOAD-FOREIGN and
4696 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4697 vital for most things which interface to C-level interfaces,
4698 like extensions working with sockets or databases or
4699 Perl-compatible regexes or whatever, and the need to fix
4700 this bug is the main reason that 0.7.1 was released so
4702 ** DEFGENERIC is now choosier about the methods it redefines, so that
4703 reLOADing a previously-LOADed file containing DEFGENERICs does
4704 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4705 cycle now works as it should. (thanks to Alexey Dejneka)
4706 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4707 fixed by Christophe Rhodes. (He actually submitted this patch
4708 months ago, and I delayed until after 0.7.0.)
4709 ** Bug 111 (internal compiler confusion about runtime checks on
4710 FUNCTION types) was fixed by Alexey Dejneka.
4711 * Some internal cleanups (getting rid of variables which aren't
4712 needed now that the byte interpreter is gone) caused the fasl
4713 file format number to change again.
4715 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4716 * major incompatible change: The default fasl file extension, i.e. the
4717 default extension for files produced by COMPILE-FILE, has changed
4718 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4720 ** There are many changes in the implementation of the compiler.
4721 SBCL is now essentially a compiler-only implementation of ANSI
4722 Common Lisp. EVAL still "interprets" a few special cases, but
4723 almost all the interesting cases are handled by creating
4724 a LAMBDA expression, calling COMPILE on it, then calling
4725 FUNCALL on the result.
4726 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4727 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4728 Since the code is newer, there might still be some new bugs
4729 (though not as many as before Martin Atzmueller's fixes:-). But
4730 the new code is substantially simpler and clearer, and hopefully
4731 any remaining bugs will be simpler, less fundamental, and more
4732 fixable then the bugs in the old code.
4733 ** The revised compiler is still a little unsteady on its feet.
4735 *** The debugging information it produces (particularly the names
4736 of FUNCTION objects) is sometimes much less useful than what
4737 the old compiler produced.
4738 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4739 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4740 within a MACROLET) has been temporarily weakened.
4741 ** There are new compiler optimizations for various functions:
4742 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4743 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4744 *** the math functions TRUNCATE, FLOOR, and CEILING
4745 *** the function-of-all-trades COERCE
4746 Mostly these should be transparent, but there's one
4747 potentially-annoying problem (bug 117): when the compiler
4748 inline-expands a function and does type analysis on the result,
4749 it can create control paths which have type mismatches, and
4750 when it can't prove that those control paths aren't taken,
4751 it will issue WARNINGs about the type mismatches. This is
4752 a particular problem in practice for the new sequence functions.
4753 It's not clear how this should be fixed, and for now, a
4754 workaround is given in the entry for 117 in the BUGS file.
4755 ** (Because of the interaction between the two previous items --
4756 occasional inlining problems and new inline expansions -- some
4757 of the new sequence function optimizations won't really kick in
4758 completely until debugging information, and then inlining, are
4759 straightened out in some future version.)
4760 * minor incompatible changes:
4761 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4762 DIRECTORY no longer implicitly promotes NIL slots of its
4763 pathname argument to :WILD. In particular, when you ask for the
4764 contents of a directory (which you used to be able to do without
4765 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4766 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4767 ** changes in behavior that ANSI explicitly defines to be
4768 implementation dependent:
4769 *** The new compiler-only implementation still conforms with ANSI,
4770 but acts a little different than before. Besides the obvious
4771 changes in performance tradeoffs (that the cost per form passed
4772 to EVAL has gone up, and the cost per form executed by EVAL
4773 has gone down), the behavior of the system changes a little
4774 because there are no longer any interpreted function objects.
4775 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4776 e.g. doing COMPILE on the output of interactive DEFUN is
4778 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4780 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4781 has changed from (:CL) to NIL.
4782 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4783 e.g. #\Newline and #\Tab, don't have names specified in the
4784 ANSI Common Lisp standard, is now based on their ASCII symbolic
4785 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4786 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4787 are no longer used for output.
4788 ** changes in internal implementation constants:
4789 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4790 4 million. (If your application spends a lot of time GCing and
4791 you have a lot of RAM, you might want to experiment with
4792 increasing it even more.)
4793 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4794 However, almost all old code should still continue to work without
4795 immediate update, as SB-C-CALL is now a (deprecated) nickname
4797 ** Old operator names in the style DEF-FOO are now deprecated in
4798 favor of new corresponding names DEFINE-FOO, for consistency with
4799 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4800 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4801 internal symbols, but a few supported extensions like
4802 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4803 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4804 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4805 etc. as you get deeper into recursive calls to the debugger
4806 command loop, instead of the old "5]", "5]]", "5]]]"
4807 sequence. (I was motivated to do this when squabbles between
4808 ILISP and SBCL left me very deeply nested in the debugger. In the
4809 short term, this change will probably provoke more ILISP/SBCL
4810 squabbles, but hopefully it will be an improvement in the long run.)
4811 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4812 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4813 function, since declaiming return types involving aliens is
4814 (1) annoyingly messy to do by hand and (2) vital to efficient
4815 compilation of code which calls such functions.
4816 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4817 longer reexported by the SB-EXT package. They're solely useful
4818 for alien code, so it seems more logical that you should get
4819 them from the SB-ALIEN package, not in SB-EXT.
4820 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4821 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4822 features. Instead, the code that they used to control is always
4823 built into the system.
4824 * many other bug fixes
4825 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4826 advantage of the new EVAL-WHEN stuff and to clean them up in
4827 general, and they are now more ANSI-compliant in a number of
4828 ways. Martin Atzmueller is responsible for a lot of this.
4829 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4831 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4832 *** correct ERROR type for various file operations
4833 *** some fixes for Lisp streams
4834 *** DEFMETHOD syntax checking
4835 *** changing old weird representation of debug information as
4836 strings (which, among their other deficiencies, don't transform
4837 correctly when you rename packages, and don't change their
4838 print representation when you change things like *PACKAGE*
4839 and *PRINT-LENGTH*) to symbols and lists of symbols
4840 He also made several improvements and fixed several bugs in DESCRIBE.
4841 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4843 *** misbehavior of WRITE-STRING/WRITE-LINE
4844 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4845 and several other LOOP problems as well
4846 *** DIRECTORY when similar filenames are present
4847 *** DEFGENERIC with :METHOD options
4848 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4849 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4850 *** argument ordering in FIND with :TEST option
4851 *** mishandled package designator argument in APROPOS-LIST
4852 *** various problems in the backquote readmacro
4853 *** a bug in APROPOS
4854 *** probably some others that I'm not describing very well here,
4855 since the CVS log documents them by reference to sbcl-devel
4856 messages, and the SourceForge archives aren't working well.:-(
4857 ** Dan Barlow improved the Alpha port (and is making progress on the
4858 PPC port, for those of you who think different).
4859 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4860 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4861 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4862 continues to work on the SPARC port (for those of you in a position
4863 to look down upon our little PC-compatible boxes from a great height).
4864 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4865 and uses that copy, rather than the current dynamic value, when
4866 it's trying to decide whether to truncate output. Thus e.g.
4867 (let ((*print-lines* 50))
4868 (pprint-logical-block (stream nil)
4870 (let ((*print-lines* 8))
4871 (print (aref possiblybigthings i) stream)))))
4872 should now truncate the logical block only at 50 lines, instead of
4873 often truncating it at 8 lines, as it did before.
4874 * The doc/cmucl/ directory, containing old CMU CL documentation
4875 from the time of the fork, is no longer part of the base system.
4876 SourceForge has shut down its anonymous FTP service, and with it
4877 my original plan for distributing the old CMU CL documentation
4878 there. For now, if you need these files you can download an old
4879 SBCL source release and extract them from it.
4880 * The fasl file version number changed again, for dozens of reasons,
4881 some of which are apparent above.
4883 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4884 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4885 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4886 patch, so that DISASSEMBLE output is much nicer.
4887 * The code in the SB-PROFILE package now seems reasonably stable.
4888 I still haven't decided what the final interface should look like
4889 (I'd like PROFILE to interact cleanly with TRACE, since both
4890 facilities use function encapsulation) but if you have a need
4891 for profiling now, you can probably use it successfully with
4892 the current CMU-CL-style interface.
4893 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4894 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4895 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4896 link now returns the dangling link itself, and for similar
4897 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4898 link itself. (In these cases the old code signalled an error and
4899 looped endlessly, respectively.) Thus, DIRECTORY now works even
4900 in the presence of dangling and cyclic symbolic links.
4901 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4902 is now a supported extension again, since the consensus on
4903 sbcl-devel was that it can be useful for ordinary development
4904 work, not just for debugging SBCL itself.
4905 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4906 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4907 that currently-defined functions might be redefined later with
4908 different return types.
4909 * Hash tables can be printed readably, as inspired by CMU CL code
4910 of Eric Marsden and SBCL code of Martin Atzmueller.
4911 * better error handling in CLOS method combination, thanks to
4912 Martin Atzmueller porting Pierre Mai's CMU CL patches
4913 * more overflow fixes for >16Mbyte I/O buffers
4914 * A bug in READ has been fixed, so that now a single Ctrl-D
4915 character suffices to cause end-of-file on character streams.
4916 In particular, now you only need one Ctrl-D at the command
4917 line (not two) to exit SBCL.
4918 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4920 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4921 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4922 does the right thing.
4923 * The implementation of some type tests, especially for CONDITION
4924 types, is now tidier and maybe faster, due to CMU CL code
4925 originally by Douglas Crosher, ported by Martin Atzmueller.
4926 * Some math functions have been fixed, and there are new
4927 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4928 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4929 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4930 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4931 likely to be built into sbcl-0.7.0.)
4932 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4933 CL, ported to SBCL by Martin Atzmueller)
4934 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4935 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4936 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4937 * There's a new slam.sh hack to shorten the edit/compile/debug
4938 cycle for low-level changes to SBCL itself, and a new
4939 :SB-AFTER-XC-CORE target feature to control the generation of
4940 the after-xc.core file needed by slam.sh.
4941 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4942 COMPILE-FILE is no longer supported, so that now every function
4943 gets an entry point, so that block compilation looks a little
4944 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4945 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4946 returns the number of bytes consed since the system started,
4947 rather than the number consed since the first time the function
4948 was called. (The new definition parallels ANSI functions like
4949 CL:GET-INTERNAL-RUN-TIME.)
4950 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4951 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4952 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4953 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4954 * Fasl file version numbers are now independent of the target CPU,
4955 since historically most system changes which required version
4956 number changes have affected all CPUs equally. Similarly,
4957 the byte fasl file version is now equal to the ordinary
4960 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4961 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4962 is no longer recognized. I apologize for this, because it was
4963 listed in SB-EXT as a supported extension, but I found that
4964 its existing behavior was poorly specified, as well as incorrectly
4965 specified, and it looked like too much of a mess to straighten it
4966 out. I have enough on my hands trying to get ANSI stuff to work..
4967 * many patches ported from CMU CL by Martin Atzmueller, with
4968 half a dozen bug fixes in pretty-printing and the debugger, and
4969 half a dozen others elsewhere
4970 * fixed bug 13: Floating point infinities are now supported again.
4971 They might still be a little bit flaky, but thanks to bug reports
4972 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4973 as flaky as they were.
4974 * The --noprogrammer command line option is now supported. (Its
4975 behavior is slightly different in detail from what the old man
4976 page claimed it would do, but it's still appropriate under the
4977 same circumstances that the man page talks about.)
4978 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4979 are now supported, and enabled by default. Thus, the compiler can
4980 handle many floating point and complex operations much less
4981 inefficiently. (Thus e.g. you can implement a complex FFT
4983 * The compiler now detects type mismatches between DECLAIM FTYPE
4984 and DEFUN better, and implements CHECK-TYPE more correctly, and
4985 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4986 thanks to patches from Martin Atzmueller.
4987 * various fixes to make the cross-compiler more portable to
4988 ANSI-conforming-but-different cross-compilation hosts (notably
4989 Lispworks for Windows, following bug reports from Arthur Lemmens)
4990 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4991 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4992 fixed thanks to Pierre Mai's CMU CL patches.
4993 * fixes to keep the system from overflowing internal counters when
4994 it tries to use i/o buffers larger than 16M bytes
4995 * fixed bug 45a: Various internal functions required to support
4996 complex special functions have been merged from CMU CL sources.
4997 (When I was first setting up SBCL, I misunderstood a compile-time
4998 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4999 * improved support for type intersection and union, fixing bug 12
5000 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
5001 more obscure bugs as well
5002 * some steps toward byte-compiling non-performance-critical
5003 parts of the system, courtesy of patches from Martin Atzmueller
5004 * Christophe Rhodes has made some debian packages of sbcl at
5005 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
5006 From his sbcl-devel e-mail of 2001-04-08 they're not completely
5007 stable, but are nonetheless usable. When he's ready, I'd be happy
5008 to add them to the SourceForge "File Releases" section. (And if
5009 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
5010 * new fasl file format version number (because of changes in
5011 internal representation of (OR ..) types to accommodate the new
5012 support for (AND ..) types, among other things)
5014 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
5015 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
5017 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
5019 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
5020 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
5021 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
5022 * DESCRIBE now works on structure objects again.
5023 * Most function call argument type mismatches are now handled as
5024 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
5025 know whether the function will be redefined before the call is
5026 executed. (The compiler could flag local calls with full WARNINGs,
5027 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
5028 it doesn't keep track of enough information to know whether calls
5029 are local in this sense.)
5030 * Compiler output is now more verbose, with messages truncated
5031 later than before. (There should be some supported way for users
5032 to override the default verbosity, but I haven't decided how to
5033 provide it yet, so this behavior is still controlled by the internal
5034 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
5035 src/compiler/ir1util.lisp.)
5036 * Fasl file format version numbers have increased again, because
5037 support for the Gray streams extension changes the layout of the
5038 system's STREAM objects.
5039 * The Gray subclassable streams extension now works, thanks to a
5040 patch from Martin Atzmueller.
5041 * The full LOAD-FOREIGN extension (not just the primitive
5042 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
5043 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
5044 but like most other programs, it defaults to copying the Unix
5045 environment from the original process instead of starting the
5046 new process in an empty environment.
5047 * Extensions which manipulate the Unix environment now support
5048 an :ENVIRONMENT keyword option which doesn't smash case or
5049 do other bad things. The CMU-CL-style :ENV option is retained
5050 for porting convenience.
5051 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
5052 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
5054 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
5056 * A patch from Martin Atzmueller seems to have solved the SIGINT
5057 problem, and as far as we know, signal-handling now works cleanly.
5058 (If you find any new bugs, please report them!)
5059 * The system no longer defaults Lisp source file names to types
5060 ".l", ".cl", or ".lsp", but only to ".lisp".
5061 * The compiler no longer uses special default file extensions for
5062 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
5063 seems to expect a single default extension for all compiled code,
5064 and there's no compelling reason to try to stretch the standard
5065 to allow two different extensions.) Instead, byte-compiled files
5066 default to the same extension as native-compiled files.
5067 * Fasl file format version numbers have increased again, because
5068 a rearrangement of internal implementation packages made some
5069 dumped symbols in old fasl files unreadable in new cores.
5070 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
5071 many fewer weird special cases.
5072 * Bug #17 (differing COMPILE-FILE behavior between logical and
5073 physical pathnames) has been fixed, and some related misbehavior too,
5074 thanks to a patch from Martin Atzmueller.
5075 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
5076 by Tim Moore, ported to SBCL by Martin Atzmueller.
5077 * Martin Atzmueller fixed several filesystem-related problems,
5078 including bug #36, in part by porting CMU CL patches, which were
5079 written in part by Paul Werkowski.
5080 * More compiler warnings in src/runtime/ are gone, thanks to
5081 more patches from Martin Atzmueller.
5082 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
5085 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
5087 * DESCRIBE now works on CONDITION objects.
5088 * The debugger now handles errors which arise when trying to print
5089 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
5091 * The build system now uses an additional file, customize-target-features.lisp,
5092 to allow local modifications to the target *FEATURES* list. (The point of
5093 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
5094 debugging features enabled, without having to worry about propagating it
5095 into everyone's system when I do a "cvs update".) When no
5096 customize-target-features.lisp file exists, the target *FEATURES* list
5097 should be constructed the same way as before.
5098 * fixed bugs in DEFCONSTANT ANSI-compatibility:
5099 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
5100 warn about behavior which is undefined under the ANSI spec. Note: This
5101 is specified by ANSI, but it's not very popular with programmers.
5102 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
5103 macro in the SBCL sources for an example of a workaround which you
5104 might use to make portable ANSI-standard code which does what you want.
5105 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
5106 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
5107 when it's used as a non-toplevel form. (This is required in order
5108 to implement the DEFCONSTANT-EQX macro.)
5109 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
5110 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
5111 funny things, due to bugs in the implementation of EVAL-WHEN
5112 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
5113 non-toplevel DEFCONSTANTs are uncommon.)
5114 * The core file version number and fasl file version number have been
5115 incremented, because the old noncompliant DEFCONSTANT behavior involved
5116 calling functions which no longer exist, and because I also took the
5117 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
5118 * fixed bug 1 (error handling before read-eval-print loop starts), and
5119 redid debugger restarts and related debugger commands somewhat while
5121 ** The QUIT debugger command is gone, since it did something
5122 rather different than the SB-EXT:QUIT command, and since it never
5123 worked properly outside the main toplevel read/eval/print loop.
5124 Invoking the new TOPLEVEL restart provides the same functionality.
5125 ** The GO debugger command is also gone, since you can just invoke
5126 the CONTINUE restart directly instead.
5127 ** The TOP debugger command is also gone, since it's redundant with the
5128 FRAME 0 command, and since it interfered with abbreviations for the
5130 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
5132 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
5133 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
5134 some porting work by Martin Atzmueller)
5135 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
5136 with a boatload of patches which he ported to SBCL. Now that those
5138 ** The system tries to make sure that its low-priority messages
5139 are prefixed by semicolons, to help people who like to use
5140 syntax highlighting in their ILISP buffer. (This patch
5141 was originally due to Raymond Toy.)
5142 ** The system now optimizes INTEGER-LENGTH better, thanks to more
5143 patches originally written by Raymond Toy.
5144 ** The compiler understands coercion between single-value and
5145 multiple-VALUES type expressions better, getting rid of some very
5146 weird behavior, thanks to patches originally by Robert MacLachlan
5147 and Douglas Crosher.
5148 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
5149 lambda lists, thanks to a patch originally by Pierre Mai.
5150 ** The system no longer bogusly warns about "abbreviated type
5152 ** The compiler gets less confused by inlining and RETURN-FROM,
5153 thanks to some patches originally by Tim Moore.
5154 ** The system no longer hangs when dumping circular lists to fasl
5155 files, thanks to a patch originally from Douglas Crosher.
5156 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
5157 undefined function error.
5158 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
5159 in structure slot initforms) for now, documented workaround instead:-|
5160 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
5161 * fixed bug 5: added stubs for various Gray stream functions called
5162 in the not-a-CL:STREAM case, so that even when Gray streams aren't
5163 installed, at least appropriate type errors are generated
5164 * fixed bug 8: better reporting of various PROGRAM-ERRORs
5165 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
5166 consistently in DEFMETHOD forms.
5167 * removed bug 21 from BUGS, since Martin Atzmueller points out that
5168 it doesn't seem to affect SBCL after all
5169 * The C runtime system now builds with better optimization and many
5170 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
5172 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
5174 * The system is now under CVS at SourceForge (instead of the
5175 CVS repository on my home machine).
5176 * The new signal handling code has been tweaked to treat register
5177 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
5178 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
5179 downstream, e.g. in debug-int.lisp, has implicit dependencies
5180 on the unsignedness of integer representation of machine words,
5181 and that caused the system to bomb out with infinite regress
5182 when trying to recover from type errors involving signed values,
5183 e.g. (BUTLAST '(1 2 3) -1).)
5184 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
5185 (This was one of the bugs Peter Van Eynde reported back in July.)
5186 * The system now uses code inspired by Colin Walters' O(N)
5187 implementation of MAP (from the cmucl-imp@cons.org mailing
5188 list, 2 September 2000) when it can't use a DEFTRANSFORM to
5189 inline the MAP operation, and there is more than one
5190 sequence argument to the MAP call (so that it can't just
5191 do ETYPECASE once and for all based on the type of the
5192 single sequence argument). (The old non-inline implementation
5193 of the general M-argument sequence-of-length-N case required
5194 O(M*N*N) time when any of the sequence arguments were LISTs.)
5195 * The QUIT :UNIX-CODE keyword argument has been renamed to
5196 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
5197 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
5199 * Raymond Wiker's patches to port dynamic loading from Linux to
5200 FreeBSD have been added.
5201 * The BUGS file is now more nearly up to date, thanks in large part
5202 to Martin Atzmueller's review of it.
5203 * The debugger now flushes standard output streams before it begins
5204 its output ("debugger invoked" and so forth).
5205 * The core version number and fasl file version number have both
5206 been incremented, because of incompatible changes in the layout
5208 * FINISH-OUTPUT is now called more consistently on QUIT. (It
5209 used to not be called for a saved Lisp image.)
5210 * Martin Atzmueller's version of a patch to fix a compiler crash,
5211 as posted on sbcl-devel 13 September 2000, has been installed.
5212 * Instead of installing Martin Atzmueller's patch for the
5213 compiler transform for SUBSEQ, I deleted the compiler transform,
5214 and transforms for some similar consing operations.
5215 * A bug in signal handling which kept TRACE from working on OpenBSD
5217 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
5218 away by constant folding
5219 * The system now defines its address space constants in one place
5220 (in the Lisp sources), and propagates them automatically elsewhere
5221 (through GENESIS and the sbcl.h file). Therefore, patching the
5222 address map is less unnecessarily tedious and error-prone. The
5223 Lisp names of address space constants have also been systematized.
5224 * CVS tags like dollar-Header-dollar have been removed from
5225 the sources, because they have never saved me trouble and
5226 they've been source of trouble working with patches and other
5227 diff-related operations.
5228 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
5229 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
5231 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
5233 * The system has been ported to OpenBSD.
5234 * The system now compiles with a simple "sh make.sh" on the systems
5235 that it's supported on. I.e., now you no longer need to tweak
5236 text in the target-features.lisp-expr and symlinks in src/runtime/
5237 by hand, the make.sh takes care of it for you.
5238 * The system is no longer so grossly inefficient when compiling code
5239 involving vectors implemented as general (not simple) vectors (VECTOR T),
5240 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
5241 which dares to use the various sequence functions on non-simple
5242 vectors, takes less of a performance hit.
5243 * There is now a primitive type predicate VECTOR-T-P
5244 to test for the (VECTOR T) type, so that e.g.
5245 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
5246 can now be compiled with some semblance of efficiency. (The old code
5247 turned the type declaration into a full call to %TYPEP at runtime!)
5248 * AREF on (VECTOR T) is still not fast, since it's still compiled
5249 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
5250 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
5251 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
5252 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
5253 without making full calls to SUBTYPEP at runtime.
5254 (Some analogous efficiency issues for non-simple vectors specialized to
5255 element types other than T, or for non-simple multidimensional arrays,
5256 have not been addressed. They could almost certainly be handled the
5257 same way if anyone is motivated to do so.)
5258 * The changes in array handling break binary compatibility, so
5259 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
5260 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
5261 instead of (VALUES T T).
5262 * By following the instructions that Dan Barlow posted to sbcl-devel
5263 on 2 July 2000, I was able to enable primitive dynamic object
5264 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
5265 functionality is not implemented (since it calls ld to resolve
5266 library references automatically, requiring RUN-PROGRAM for its
5267 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
5268 try to resolve library references) is now supported.
5269 * The system now flushes the standard output streams when it terminates,
5270 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
5271 them at several other probably-convenient times, e.g. in each pass of
5272 the toplevel read-eval-print loop, and after evaluating a form given
5273 as an "--eval" command-line option. (These changes were motivated by a
5274 discussion of stream flushing issues on cmucl-imp in August 2000.)
5275 * The source transform for TYPEP of array types no longer assumes
5276 that an array whose element type is a not-yet-defined type
5277 is implemented as an array of T, but instead punts, so that the
5278 type will be interpreted at runtime.
5279 * There is now some support for cross-compiling in make.sh: each of
5280 the phases of make.sh has its own script. (This should be transparent
5281 to people doing ordinary, non-cross-compile builds.)
5282 * Since my laptop doesn't have hundreds of megabytes of memory like
5283 my desktop machine, I became more motivated to do some items on
5284 my to-do list in order to reduce the size of the system a little:
5285 ** Arrange for various needed-only-at-cold-init things to be
5286 uninterned after cold init. To support this, those things have
5287 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
5288 symbols with such names are now uninterned after cold init).
5289 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
5290 fasl files for cold load.
5291 ** Remove the old compiler structure pooling code (which used to
5292 be conditional on the target feature :SB-ALLOC) completely.
5293 ** Redo the representation of some data in cold init to be more compact.
5294 (I also looked into supporting byte compiled code at bootstrap time,
5295 which would probably reduce the size of the system a lot, but that
5296 looked too complicated, so I punted for now.)
5297 * The maximum signal nesting depth in the src/runtime/ support code has
5298 been reduced from 4096 to 256. (I don't know any reason for the very
5299 large old value. If the new smaller value turns out to break something,
5300 I'll probably just bump it back up.)
5301 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
5303 * Many, many bugs reported by Peter Van Eynde have been added to
5304 the BUGS list; some have even been fixed.
5305 * While enabling dynamic object file loading, I tried to make the
5306 code easier to understand, renaming various functions and variables
5307 with less ambiguous names, and changing some function calling
5308 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
5309 * While trying to figure out how to do the OpenBSD port, I tried to
5310 clean up some of the code in src/runtime/. In particular, I dropped
5311 support for non-POSIX signal handling, added various comments,
5312 tweaked the code to reduce the number of compilation warnings, and
5313 renamed some files to increase consistency.
5314 * To support the new automatic configuration functionality in make.sh,
5315 the source file target-features.lisp-expr has been replaced with the
5316 source file base-target-features.lisp-expr and the machine-generated
5317 file local-target-features.lisp-expr.
5318 * fixed a stupid quoting error in make.sh so that using CMU CL
5319 "lisp -batch" as cross-compilation host works again
5321 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
5323 * DESCRIBE no longer tries to call itself recursively to describe
5324 bound/fbound values, so that it no longer fails on symbols which are
5325 bound to themselves (like keywords, T, and NIL).
5326 * DESCRIBE now works on generic functions.
5327 * The printer now prints less-screwed-up representations of closures
5328 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
5329 * A private symbol is used instead of the :EMPTY keyword previously
5330 used to mark empty slots in hash tables. Thus
5331 (DEFVAR *HT* (MAKE-HASH-TABLE))
5332 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
5333 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
5334 now does what ANSI says that it should. (You can still get
5335 similar noncompliant behavior if bang on the hash table
5336 implementation with all the symbols you get back from
5337 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
5338 This breaks binary compatibility, since tests for equality to
5339 :EMPTY are wired into things like the macroexpansion of
5340 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
5342 * There's now a minimal placeholder implementation for CL:STEP,
5343 as required by ANSI.
5344 * An obscure bug in the interaction of the normal compiler, the byte
5345 compiler, inlining, and structure predicates has been patched
5346 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
5347 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
5349 * Missing ordinary arguments in a macro call are now detected even
5350 when the macro lambda list contains &KEY or &REST.
5351 * The debugger no longer complains about encountering the top of the
5352 stack when you type "FRAME 0" to explicitly instruct it to go to
5353 the top of the stack. And it now prints the frame you request even
5354 if it's the current frame (instead of saying "You are here.").
5355 * As specified by ANSI, the system now always prints keywords
5356 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
5358 * The default initial SIZE of HASH-TABLEs is now smaller.
5359 * Type information from CLOS class dispatch is now propagated
5360 into DEFMETHOD bodies, so that e.g.
5361 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5363 is now basically equivalent to
5364 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5365 (DECLARE (TYPE SINGLE-FLOAT X))
5367 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
5368 operation, without having to do run-time type dispatch.
5369 * The macroexpansion of DEFMETHOD has been tweaked so that it has
5370 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
5371 * Since I don't seem to be making big file reorganizations very often
5372 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
5373 is overflowing my ability to conveniently back them up), I've finally
5374 checked the system into CVS. (The CVS repository is on my home system,
5375 not at SourceForge -- putting it on SourceForge might come later.)
5376 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
5377 high-level GC-NOTIFY-FOO functions send their output. (There's
5378 still very little control of where low-level verbose GC functions
5379 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
5380 less than it used to -- the GC-NOTIFY-FOO functions are now under
5381 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
5382 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
5383 in only one place in the source code, and propagates it automatically
5384 everywhere that it's needed. Thus e.g. when I bump the version from
5385 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
5386 * The C source files now include boilerplate legalese and documentation
5387 at the head of each file (just as the Lisp source files already did).
5388 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
5389 to his page will be replaced with a link to his new CLiki service.
5391 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
5393 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
5394 * The build process now looks for GNU make under the default name "gmake",
5395 instead of "make" as it used to. If GNU make is not available as "gmake"
5396 on your system, you can change this default behavior by setting the
5397 GNUMAKE environment variable.
5398 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
5399 can build without error under CMU CL.
5401 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
5403 * There is now a partial SBCL user manual (with some new text and some
5404 text cribbed from the CMU CL manual).
5405 * The beginnings of a profiler have been added (starting with the
5406 CMU CL profiler and simplifying and cleaning up). Eventually the
5407 main interface should be through the TRACE macro, but for now,
5408 it's still accessed through vaguely CMU-CL-style functions and macros
5409 exported from the package SB-PROFILE.
5410 * Some problems left over from porting CMU CL to the new
5411 cross-compilation bootstrap process have been cleaned up:
5412 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
5413 instead of SB!XC:DEFMACRO, compounded by an oversight on my
5414 part when getting rid of the compiler *BACKEND* stuff.)
5415 ** The value of *NULL-TYPE* was screwed up, because it was
5416 being initialized before the type system knew the final
5417 definition of the 'NULL type. This screwed up several key
5418 optimizations in the compiler, causing inefficiency in all sorts
5419 of places. (I found it because I wanted to understand why
5420 GET-INTERNAL-RUN-TIME was consing.)
5421 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
5422 PROCLAIM FTYPE information. Unfortunately this broke binary
5423 compatibility again, since now the forms output by DEFGENERIC
5424 to refer to functions which didn't exist in 0.6.3.
5425 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
5426 can use the new (as of 0.6.3) transform for SOME into MAP into
5428 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
5429 values to (INTEGER 0 1000000), so that the time code will no longer
5430 occasionally get blown up by Linux returning 1000000 microseconds
5431 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
5432 its output conform to the ANSI spec. (Alas, this makes its output
5433 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
5434 * A full call to MAP NIL with a single sequence argument no longer conses.
5435 * fixes to problems pointed out by Martin Atzmueller:
5436 * The manual page no longer talks about multiprocessing as though
5437 it were currently supported.
5438 * The ILISP support patches have been removed from the distribution,
5439 because as of version 5.10.1, ILISP now supports SBCL without us
5440 having to maintain patches.
5441 * added a modified version of Raymond Toy's recent CMU CL patch for
5442 EQUALP comparison of HASH-TABLE
5444 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
5446 * The system still can't cross-compile itself with
5447 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
5448 entails), but at least it can compile more of itself that way
5449 than it used to be able to, and various buglets which were uncovered
5450 by trying to cross-compile itself that way have now been fixed.
5451 * This release breaks binary compatibility again. This time
5452 at least I've incremented the FASL file format version to 2, so that the
5453 problem can be detected reliably instead of just causing weird errors.
5454 * various new style warnings:
5455 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
5456 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
5457 ** using the lexical binding of a variable named in the *FOO* style
5458 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
5459 as specified by ANSI.
5460 * *RANDOM-STATE* is no longer automatically initialized from
5461 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
5462 default behavior of the system is to repeat its behavior every time
5463 it's run. If you'd like to change this behavior, you can always
5464 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
5465 old convention there was no comparably easy way to get the system to
5466 repeat its behavior every time it was run.
5467 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
5468 FTYPE declarations has been removed, in favor of their ANSI
5469 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
5470 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
5471 the types of their sequence arguments can be determined at compile time.
5472 This is done through a new open code expansion for MAP which eliminates
5473 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
5474 argument types can be determined at compile time.
5475 * The optimizer now transforms COERCE into an identity operation when it
5476 can prove that the coerced object is already of the correct type. (This
5477 can be a win for machine generated code, including the output of other
5478 optimization transforms, such as the MAP transform above.)
5479 * Credit information has been moved from source file headers into CREDITS.
5480 * Source file headers have been made more standard.
5481 * The CASE macro now compiles without complaining even when it has
5484 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
5486 * (Note that the way that the PCL macroexpansions were rewritten
5487 to accommodate the change in DEFGENERIC below breaks binary
5488 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
5489 not run under sbcl-0.6.2. Once we get out of alpha releases,
5490 i.e. hit release 1.0.0, we'll probably try to maintain binary
5491 compatibility between maintenance releases, e.g. between sbcl-1.4.3
5492 and sbcl-1.4.4. Until then, however, it might be fairly common
5493 for maintenance releases to break binary compatibility.)
5494 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
5496 * The reporting of unhandled signals has been changed to print some
5497 explanatory text as well as the report form. (Previously only
5498 the report form was printed.)
5499 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
5500 it defines, so that the compiler no longer issues undefined function
5501 warnings for compiled-but-not-yet-loaded generic functions.
5502 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
5503 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
5504 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
5505 section of the ANSI spec are supported.
5506 * The "" nickname for the "KEYWORD" package has been removed.
5507 The reader still handles symbol tokens which begin with a package marker
5508 as keywords, but it doesn't expose its mechanism for doing so in the
5509 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
5510 * The system now issues STYLE-WARNINGs for contradictory TYPE
5511 proclamations. (Warnings for contradictory FTYPE proclamations would
5512 be nice too, but those can't be done usefully unless the type system
5513 is made smarter about FUNCTION types.)
5514 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
5515 systematized, so that "*target-*.lisp is supposed to exist only on the
5516 target and imply that there's a related file which exists on the
5517 host, and *host-*.lisp is supposed to exist only on the host and imply
5518 that there's a related file which exists on the target. This involves a
5519 lot of renaming. Hopefully the acute confusion caused by the renaming
5520 will be justified by the reduction in chronic confusion..
5521 ** runtime-type.lisp -> early-target-type.lisp
5522 ** target-type.lisp -> late-target-type.lisp
5523 ** early-host-format.lisp -> early-format.lisp
5524 ** late-host-format.lisp -> late-format.lisp
5525 ** host-error.lisp -> misc-error.lisp
5526 ** early-error.lisp -> early-target-error.lisp
5527 ** late-error.lisp -> late-target-error.lisp
5528 ** host-defboot.lisp -> early-defboot.lisp
5529 ** code/misc.lisp -> code/target-misc.lisp
5530 ** code/host-misc.lisp -> code/misc.lisp
5531 ** code/numbers.lisp -> code/target-numbers.lisp
5532 ** code/early-numbers.lisp -> numbers.lisp
5533 ** early-host-type.lisp -> early-type.lisp
5534 ** late-host-type.lisp -> late-type.lisp
5535 ** host-typep.lisp -> typep.lisp
5536 ** load.lisp -> target-load.lisp
5537 ** host-load.lisp -> load.lisp
5538 ** host-disassem.lisp -> disassem.lisp
5539 ** host-insts.lisp -> insts.lisp
5540 ** byte-comp.lisp -> target-byte-comp.lisp
5541 ** host-byte-comp.lisp -> byte-comp.lisp
5542 ** host-signal.lisp -> signal.lisp
5543 ** host-defstruct.lisp -> defstruct.lisp
5544 ** late-target-type.lisp -> deftypes-for-target.lisp
5545 Furthermore, several other previously target-only files foo.lisp (e.g.
5546 hash-table.lisp and random.lisp) have been split into a target-and-host
5547 foo.lisp file and a target-only target-foo.lisp file, with their key type
5548 definitions in the target-and-host part, so that the cross-compiler will
5549 know more about target types.
5550 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5551 gone away. In their place are various *BACKEND-FOO* variables
5552 corresponding to the slots of the old structure.
5553 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5554 into the target SBCL has been fixed.
5555 * The chill.lisp system for loading cold code into a running SBCL
5557 * Support for the CMU CL "scavenger hook" extension has been removed.
5558 (It was undocumented and unused in the CMU CL sources that SBCL was
5559 derived from, and stale in sbcl-0.6.1.)
5560 * Various errors in the cross-compiler type system were detected
5561 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5562 (enabling various consistency checks). Many of them were fixed,
5563 but some hard problems remain, so the compiler is back to
5564 running without *TYPE-SYSTEM-INITIALIZED* for now.
5565 * As part of the cross-compiler type system cleanup, I implemented
5566 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5567 * I have started adding UNCROSS calls throughout the type system
5568 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5569 on cross-compiler input files will be able to go away, and various
5571 * CONSTANTP now returns true for quoted forms (as explicitly required
5574 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5576 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5577 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5578 (MAKE-LIST -1). In the longer term, it ought to become true
5579 that declarations are assertions even at SAFETY 1. For now, it's not
5580 quite true even at SAFETY 3, but it's at least more nearly true..
5581 (Note that this change seems to increases the size of the system by
5582 O(5%) and to decrease the speed of the compiler by 20% or more.)
5583 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5584 for the problem of printing dozens of lines of distracting information
5585 about low-level system machinery as part of the top stack frame
5586 on entry to the debugger when an undefined function was called.
5587 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5588 is not reset to COMMON-LISP-USER.
5589 * Compilation of stuff related to dyncount.lisp has been made conditional
5590 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5591 smaller. The various dyncount-related symbols have been moved into
5592 a new "SB-DYNCOUNT" package.
5593 * tty-inspect.lisp has been renamed to inspect.lisp.
5594 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5595 feature has gone away. (When we eventually port to other flavors of
5596 libc and/or Unix, we'll try to make the differences between flavors
5597 invisible at the user level.)
5598 * Various other *FEATURES* tags, and/or their associated conditionals,
5599 have been removed if obsolescent, or given better documentation, or
5600 sometimes given more-mnemonic names.
5602 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5604 * tidied up "make.sh" script
5605 * tidied up system directory structure
5606 * better "clean.sh" behavior
5607 * added doc/FOR-CMUCL-DEVELOPERS
5608 * many many small tweaks to output format, e.g. removing possibly-confusing
5609 trailing #\. character in DESCRIBE-INSTANCE
5610 * (EQUALP #\A 'A) no longer signals an error.
5611 * new hashing code, including EQUALP hashing
5612 * tidied up Lisp initialization and toplevel
5613 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5614 * command line argument processing
5615 * added POSIX-GETENV function to deal with Unix-ish environment variables
5616 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5617 terminating SBCL on EOF
5618 * non-verbose GC by default
5619 * There is no more "sbcl" shell script; the sbcl file is now the C
5620 runtime executable (just like CMU CL).
5621 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5623 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5624 symbol and package information as Lisp native symbol and package objects
5625 instead of strings naming symbols and strings naming packages. This way,
5626 whenever packages are renamed (as in warm init), debug information is
5627 transformed along with everything else.
5628 * tweaked the optimization policy declarations which control the building
5629 of SBCL itself. Now, among other things, the system no longer saves
5630 source location debugging information. (This helps two problems at once
5631 by reducing SBCL size and by keeping SBCL from trying to look for its
5632 sources -- which may not exist -- when reporting errors.)
5633 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5634 debugging and testing purposes
5635 * cleaned up printing, making the printer call PRINT-OBJECT for
5636 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5637 methods, giving nearly-ANSI behavior
5638 * converted almost all special variables to use *FOO* naming convention
5639 * deleted PARSE-TIME functionality, since it can be done portably
5640 * moved some files out of cold init into warm init
5641 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5643 * regularized formatting of source files
5644 * added an install.sh script
5645 * fixed ridiculous memory usage of cross-compiler by making
5646 compiler/alloc.lisp not try to do pooling unless it can hook
5647 itself into the GC of the cross-compilation host. Now the system
5648 builds nicely on my old laptop.
5649 * added :SB-ALLOC in target-features.lisp-expr
5650 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5651 was not implemented)
5652 * re-did condition handling and note reporting in the compiler. Notes
5653 are no longer handled by signalling conditions. Style warnings
5654 and warnings are handled more correctly and reported in such a way
5655 that it's easy to find one or the other in your output (so that you
5656 can e.g. figure out which of many problems caused COMPILE-FILE to
5658 * changed the severity of several compiler warnings from full WARNING
5659 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5660 compiler note reporting so that it doesn't use the condition system
5661 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5662 COMPILE-FILE command)
5663 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5664 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5665 no longer supported, since their implementation was deeply intertwingled
5666 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5667 * removed bogus "support" for compiler macros named (SETF FOO), and
5668 removed the compiler macro for SETF INFO (but only after making a fool
5669 of myself on the cmucl-imp mailing list by posting a bogus patch for
5670 DEFINE-COMPILER-MACRO..)
5671 * Compiled files containing forms which have side effects on the Lisp
5672 reader (such as DEFPACKAGE forms) are now handled more correctly.
5673 (Compiler queuing of top level lambdas has been suppressed by setting
5674 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5675 * deleted various currently-unused source files, e.g. gengc.lisp. They
5676 may be added back at some point e.g. when porting to other architectures,
5677 but until they are it's distracting to distribute them and to try to
5679 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5680 there were so many of them they're just distractions, and UNCROSS is
5681 known to be able to handle the current sources
5682 * moved PROFILE functionality into TRACE, so that it will be clear
5683 how the wrapping and unwrapping of functions when you profile them
5684 interacts with the wrapping and unwrapping of functions when you
5685 trace them. (Actually, the functionality isn't there yet, but at least
5686 the interface specification is there. Hopefully, the functionality will
5687 arrive with some future maintenance release.)
5688 * removed host-oops.lisp
5689 * changed signature of QUIT function to allow UNIX-CODE argument
5690 * fixed READ-SEQUENCE bug
5691 * tweaked verbose GC output so that it looks more like the progress
5692 output that ANSI specifies for functions like LOAD
5693 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5694 * added <http://sbcl.sourceforge.com> to the banner information printed by