1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 changes relative to sbcl-1.0.31
3 * bug fix: SAVE-LISP-AND-DIE option :SAVE-RUNTIME-OPTIONS did not work
4 correctly when starting from an executable core without saved runtime
5 options (reported by Faré Rideau, thanks to Zach Beane)
6 * bug fix: (SETF SLOT-VALUE) signalled a warning which should have been
7 an optimization note instead. (reported by Martin Cracauer)
8 * bug fix: WITH-SLOTS did not work with THE forms. (thanks to David Tolpin)
9 * improvement: better error signalling for bogus parameter specializer names
10 in DEFMETHOD forms (reported by Pluijzer)
12 changes in sbcl-1.0.31 relative to sbcl-1.0.30:
13 * improvement: stack allocation is should now be possible in all nested
14 inlining cases: failure to stack allocate when equivalent code is manually
15 open coded is now considered a bug.
16 * improvements related to Unicode:
17 ** the Unicode character database has been upgraded to the
18 Unicode 5.1 standard, giving names and properties to a number of new
19 characters, and providing a few extra characters with case
21 ** the system now recognizes and produces names for Unicode Hangul
23 ** the EBCDIC-US external-format is now supported for octet operations
24 (as well as for stream operations).
25 * new feature: experimental :EMIT-CFASL parameter to COMPILE-FILE can
26 be used to output toplevel compile-time effects into a separate .CFASL
28 * optimization: COERCE to VECTOR, STRING, SIMPLE-STRING and recognizable
29 one-dimenstional subtypes of ARRAY is upto 70% faster when the coercion is
31 * optimization: TRUNCATE on known single- and double-floats is upto 25%
33 * optimization: division of floating point numbers by constants uses
34 multiplication by reciprocal when an exact reciprocal exists.
35 * optimization: multiplication of single- and double-floats floats by
36 constant two has been optimized.
37 * optimization: ARRAY-IN-BOUNDS-P is resolved at compile-time when
38 sufficient type information is available. (thanks to Leslie Polzer)
39 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) with constant slot names on
40 known structure objects are as efficient as defstruct generated accessors.
41 * optimization: unused vector creation can now be optimized away.
42 * improvement: ASDF systems can now depends on SB-INTROSPECT.
43 * improvement: a STYLE-WARNING is signalled when a generic function
44 clobbers an earlier FTYPE proclamation.
45 * improvement: the compiler is able to track the effective type of
46 generic function across method addition and removal even in the
47 absence of an explicit DEFGENERIC.
48 * improvement: DESCRIBE now reports on symbols naming undefined
49 but assumed or declared function as well.
50 * improvement: recompilation of systems using SB-GROVEL now works
51 (thanks to Leslie Polzer)
52 * improvements to SB-CLTL2 (thanks to Larry D'Anna):
53 ** functions DECLARATION-INFORMATION, PARSE-MACRO, and ENCLOSE have been
55 ** AUGMENT-ENVIRONMENT and DEFINE-DECLARATION have been implemented.
56 ** DECLARATION-INFORMATION now supports declaration name DECLARATION as
57 well as user defined declaration names.
58 ** VARIABLE-INFORMATION is now aware of alien variables.
59 * improvement: improved address space layout on OpenBSD (thanks to Josh
61 * improvement: pretty-printing of various Lisp forms has been improved
62 (thanks to Tobias Rittweiler)
63 * bug fix: calls to DECODE-FLOAT and INTEGER-DECODE-FLOAT whose value was
64 unused were deleted in safe code. (reported by John Fremlin)
65 * bug fix: a failing AVER compiling certain MAKE-ARRAY forms. (reported
67 * bug fix: some out-of-line array predicates were missing (reported by
69 * bug fix: a failing AVER in CONVERT-MV-CALL has been fixed. (thanks to
71 * bug fix: a failing AVER in %ALLOCATE-CLOSURES conversion has been fixed
72 (thanks to Larry D'Anna)
73 * bug fix: SLEEP supports times over 100 million seconds on long on OpenBSD
74 as well. (reported by Josh Elsasser)
75 * bug fix: DELETE-FILE on streams no longer closes the stream with :ABORT T,
76 leading to possible attempts to delete the same file twice. See docstring
77 on DELETE-FILE for details. (reported by John Fremlin)
78 * bug fix: DELETE-FILE once again deletes the file named by the pathname
79 designator argument, rather than its truename. (reported by Luis
81 * bug fix: the low-level debugger had 32-bit assumptions and was missing
82 information about some array types. (thanks to Luis Oliveira)
83 * bug fix: moderately complex combinations of inline expansions could
84 be miscompiled if the result was declared to be dynamic extent.
85 * bug fix: on x86, SAP-REF of sizes greater than 8 bits with offsets of the
86 form (+ <variable> <integer>) were miscompiled under certain
88 * bug fix: in some cases no compiler note about failure to stack allocate
89 was emitted, even if the objects were in fact heap allocated.
90 * bug fix: minor violation of "otherwise inaccessible" rule for stack
91 allocation could cause objects users might reasonably expect to
92 be heap allocated to be stack allocated.
93 * bug fix: DESCRIBE signalled an error for generic functions under
94 certain circumstances. (thanks to Leslie Polzer)
95 * bug fix: Fixed spelling of an error message.
97 changes in sbcl-1.0.30 relative to sbcl-1.0.29:
98 * minor incompatible change: SB-THREAD:JOIN-THREAD-ERROR-THREAD and
99 SB-THREAD:INTERRUPT-THREAD-ERROR-THREAD have been deprecated in favor
100 of SB-THREAD:THREAD-ERROR-THREAD.
101 * new contrib module: SB-QUEUE provides thread-safe lockless FIFO queues.
102 * new feature: docstrings for local and anonymous functions are no longer
103 discarded. (thanks to Leslie Polzer)
104 * new feature: SB-THREAD:SYMBOL-VALUE-IN-THREAD provides access to symbol
105 values in other threads.
106 * new feature: SB-INTROSPECT:ALLOCATION-INFORMATION provides information
107 about object allocation.
108 * optimization: division of a real float by a complex float is implemented
109 with a specialised code sequence.
110 * optimization: MAKE-INSTANCE with non-constant class-argument but constant
111 keywords is an order of magnitude faster.
112 * optimization: MAKE-INSTANCE with constant keyword arguments is x2-4 faster
113 in the presence of :AROUND or non-standard primary INITIALIZE-INSTANCE
114 methods, and similarly for non-standard metaclass classes as long as there
115 are no methods additional on MAKE-INSTANCE.
116 * optimization: more efficient type-checks for FIXNUMs when the value
117 is known to be a signed word on x86 and x86-64.
118 * optimization: compiler now optimizes (EXPT -1 INTEGER), (EXPT -1.0 INTEGER),
119 and (EXPT -1.0d0 INTEGER) into an ODDP test. (thanks to Stas Boukarev and
121 * optimization: compiler is smarter about delegating argument type checks to
123 * optimization: several character functions are now compiled somewhat more
124 efficiently. (reported by Lynn Quam)
125 * optimization: the compiler now derives simple types for LOAD-VALUE-FORMs.
126 * improvement: less unsafe constant folding in floating point arithmetic,
127 especially for mixed complex/real -float operations.
128 * optimization: constant double and single floats are stored in native
129 unboxed format on x86[-64].
130 * optimization: smarter code for arithmetic operations with constant floats,
131 complex floats, or integers on x86[-64].
132 * optimization: smarter code for conjugate/multiplication of float complexes
133 and abs/negate of floats on x86-64.
134 * optimization: more efficient complex float and real float operations on
136 * improvement: complex float division is slightly more stable.
137 * improvement: DESCRIBE output has been reworked to be easier to read and
138 contains more pertinent information.
139 * improvement: failure to provide requested stack allocation compiler notes
140 provided in all cases (requested stack allocation not happening without a
141 note being issued is now considered a bug.)
142 * bug fix: SB-POSIX exports the documented types and functions
143 FILE-DESCRIPTOR and FILENAME, and also the corresponding -DESCRIPTOR
144 types. (reported by "abhi")
145 * bug fix: on 64 bit platforms FILL worked incorrectly on arrays with
146 upgraded element type (COMPLEX SINGLE-FLOAT), regression from 1.0.28.55.
147 (thanks to Paul Khuong)
148 * bug fix: looping around HANDLER-CASE could silently consume stack space
149 on each iteration. (reported by "foobar")
150 * bug fix: better error signalling when calls to functions seeking elements
151 from lists (eg. ADJOIN) are compiled with both :TEST and :TEST-NOT.
152 (reported by Tobias Rittweiler)
153 * bug fix: regressions in DIRECTORY from 1.0.28.61: pattern matching of
154 directory components now works as it used to. (various prolems reported by
155 Michael Becker, Gabriel Dos Reis, Cyrus Harmon, and Harald Hanche-Olsen)
156 * bug fix: :PTY option in RUN-PROGRAM was broken with stream arguments.
157 (reported by Elliot Slaughter, thanks to Stas Boukarev)
158 * bug fix: bogus undefined variable warnings from fopcompiled references to
159 global variables. (thanks to Lars Rune Nøstdal)
160 * bug fix: foreign function names should now appear in backtraces on
161 FC6 as well. (reported by Tomasz Skutnik and Tobias Rautenkranz)
162 * bug fix: SETF compiler macro documentation strings are not discarded
164 * bug fix: GENTEMP is now unaffected by pretty printer dispatch table.
165 (thanks to Alex Plotnick)
166 * bug fix: SLEEP accepts large integer arguments, truncating them to
167 SIGNED-WORD on the assumption that sleeping for 68 years is sufficient
168 for anyone. (reported by Leslie Polzer, thanks to Stas Boukarev)
169 * bug fix: compiler notes for expensive slot type checks could be emitted
170 at runtime MAKE-INSTANCE calls. (reported by Samium Gromoff)
172 changes in sbcl-1.0.29 relative to 1.0.28:
173 * IMPORTANT: bug database has moved from the BUGS file to Launchpad
174 https://bugs.launchpad.net/sbcl
175 Bugs can be reported directly there, or by sending email to
176 sbcl-bugs@lists.sourceforge.net
177 (no subscription required.)
178 * minor incompatible change: under weak type checking policy integer
179 types are weakened less aggressively.
180 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
181 allowed to return, which causes SBCL to quit with exit status 0. Previously
182 if the function returned with a small integer return value, that value
183 was accidentally reused as the exit status.
184 * new feature: SB-EXT:DEFINE-HASH-TABLE-TEST allows defining new arguments
185 to MAKE-HASH-TABLE :TEST, and MAKE-HASH-TABLE has been extended with
186 :HASH-FUNCTION argument. Refer to user manual for details.
187 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
189 * new feature: SB-EXT:GET-TIME-OF-DAY provides access to seconds and
190 microseconds since the Unix epoch on all platforms.
191 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
192 allows the compiler to safely elide boundedness checks for special
194 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
195 the symbol, prohibits both lexical and dynamic binding. This is mainly an
196 efficiency measure for threaded platforms, but also valueable in
198 * new feature: UNC pathnames are now understood by the system on Windows.
199 * optimization: the compiler uses a specialized version of FILL when the
200 element type is know in more cases, making eg. (UNSIGNED-BYTE 8) case
202 * optimization: accesses to potentially non-simple arrays where element type
203 is known are 50% faster.
204 * optimization: compiler now generates faster array typechecking code.
205 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
207 * optimization: multidimensional array accesses in the absence of type
208 information regarding array rank are approximately 10% faster due to
209 open coding of ARRAY-RANK.
210 * optimization: result of (FILL (MAKE-ARRAY ...) ...) and (REPLACE
211 (MAKE-ARRAY ...) ...) can be stack allocated if the result of MAKE-ARRAY
213 * optimization: result of call to VECTOR can now be stack allocated.
214 * optimization: MAKE-ARRAY with :INITIAL-CONTENTS is now vastly faster
215 as long as the resulting array is one-dimensional and has a known
216 element type. In particular, :INITIAL-CONTENTS (LIST ...) where the
217 length of the list matches the known length of the vector does not
218 allocate the list as an intermediate step. Ditto for VECTOR and simple
220 * optimization: MAKE-ARRAY can now stack allocate in the presence of
221 :INITIAL-CONTENTS and :INITIAL-ELEMENT as long as the result has a
222 known element type, and is known to be simple and one dimensional.
223 * improvement: SBCL now emits a compiler note where stack allocation was
224 requested but could not be provided (not in all cases, unfortunately)
225 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
226 * improvement: pretty-printing loop has been implemented properly. (thanks
227 to Tobias Rittweiler)
228 * documentation: CLOS slot typechecing policy has been documented.
229 * bug fix: FILE-AUTHOR no longer signals an error on Windows.
230 * bug fix: SB-SPROF could be foiled by foreign code not have a frame
231 pointer, leading to memory faults. (thanks to Bart Botta)
232 * bug fix: better floating point exception handling on x86/OpenBSD.
233 (thanks to Josh Elsasser)
234 * bug fix: exit status from QUIT when called under --script was lost
235 (reported by Hubert Kauker)
236 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
237 same implementation of FILL to initialize the array, even if a more
238 efficient one was available (reported by Stas Boukarev, thanks to
240 * bug fix: potential miscompilation of array stack allocation on x86 and
241 x86-64. (reported by Time Tossavainen)
242 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
243 result in their subforms being treated as top level forms. (reported by
245 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
246 than a machine register could leave garbage in the high bits of the
247 result register (bug 316325).
248 * bug fix: disable address space randomization Linux/x86-64 as well,
249 not just x86-64. (reported by Ken Olum)
250 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
251 generate incorrect code.
252 * bug fix: #201; type inference for CONS and ARRAY types could derive
253 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
254 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
255 could not be assigned to or bound with PROGV. (reported by Lorenz
257 * bug fix: the value of CL:- in the inspector was the previous expression
258 evaluated rather than the expression being evaluated.
259 * bug fix: constants can no longer be locally declared special.
260 * bug fix: signals delivered to threads started from foreign land (read:
261 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
262 thread by blocking all signals and resignalling.
263 * bug fix: SHARED-INITIALIZE initialized unbound :ALLOCATION :CLASS slots
264 from :INITFORM, if any.
266 changes in sbcl-1.0.28 relative to 1.0.27:
267 * a number of bugs in cross-compilation have been fixed, with the ultimate
268 result that building under (at least) clisp should be much more reliable.
269 * minor incompatible changes: echo-streams now propagate unread-char to the
270 underlying input stream, and no longer permit unreading more than one
272 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
273 frames, allowing for instance more reliable backtraces.
274 * improvement: the debugger REPL can now reference lexical variables
275 by name directly for code compiled with (DEBUG 3).
276 * improvement: errors from malformed declarations now have better source
277 paths associated with them. (thanks to Tobias Rittweiler)
278 * optimization: faster local calls on x86/x86-64
279 * bug fix: some error messages for out-of-bound array indexes confused the
280 index and the bound. (thanks to Stas Boukarev)
281 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
283 * bug fix: running regressions tests in shells without OSTYPE set now works.
284 (reported by Harald Hanche-Olsen)
285 * bug fix: more robust static space exhaustion signalling from
286 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
287 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
288 docstring away instead of storing it under names such as (LAMBDA (X)).
289 (reported by Leslie Polzer)
290 * bug fix: timers could go off in the wrong order, be delayed indefinitely
291 (thanks to Ole Arndt for the patch)
292 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
293 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
294 potentially-invalid effective methods in its cache.
295 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
296 instances as well (thanks to Paul Khuong)
297 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
298 SB-POSIX:WAITPID and their C equivalents.
299 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
301 changes in sbcl-1.0.27 relative to 1.0.26:
302 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
303 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
304 * improvement: the system either recovers from stack exhaustion or dies
305 properly as opposed to leaving the user uncertain of whether the handler
306 trampled on some random memory next to the stack or having to rely on
307 --lose-on-corruption (which is still a good idea to use in production
308 because stack exhaustion can happen in signal handlers which will likely
310 * bug fix: fix gc related interrupt handling bug on ppc (regression from
311 1.0.25.37, reported by Harald Hanche-Olsen)
312 * bug fix: work around signal delivery bug in darwin (regression from
313 1.0.25.44, reported by Sidney Markowitz)
314 * bug fix: fix ERROR leaking memory (reported by David Thompson)
316 changes in sbcl-1.0.26 relative to 1.0.25:
317 * incompatible change: an interruption (be it a function passed to
318 INTERRUPT-THREAD or a timer function) runs in an environment where
319 interrupts can be enabled. The interruption can use
320 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
321 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
322 potentially running out of stack. Keep in mind that in the absence
323 of WITHOUT-INTERRUPTS some potentially blocking operation such as
324 acquiring a lock can enable interrupts.
325 * incompatible change: GC-OFF and GC-ON are removed, as they were
326 always unsafe. Use WITHOUT-GCING instead.
327 * new feature: runtime option --disable-ldb
328 * new feature: runtime option --lose-on-corruption to die at the
329 slightest hint of possibly non-recoverable errors: running out of
330 memory, stack, alien stack, binding stack, encountering a memory
331 fault, etc. In the absence of --lose-on-corruption a warning is
333 * enhancement: detect binding stack exhaustion
334 * enhancement: detect alien stack exhaustion on x86/x86-64
335 * improvement: generally more stable and reliable interrupt handling
336 * improvement: there is a per thread interruption queue,
337 interruptions are executed in order of arrival
338 * improvement: a repeating timer reschedules itself when the it has
339 finished, but expiration times are spaced equally. If an
340 expiration time is in the past it will trigger after a short grace
341 period that may give a chance to other things to run.
342 * optimization: slightly faster gc on multithreaded builds
343 * optimization: faster WITHOUT-GCING
344 * bug fix: when JOIN-THREAD signals an error, do it when not holding
345 important locks so that the debugger/handler doesn't produce
346 recursive errors or deadlock.
347 * bug fix: real-time signals are not used anymore, so no more
348 hanging when the system wide real-time signal queue gets full.
349 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
351 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
352 * bug fix: fix random memory faults related to interrupts on alpha
353 * bug fix: fix random memory faults related to interrupts on sparc
354 * bug fix: fix deadlocks related to starting threads
355 * bug fix: fix deadlines on locks on futex platforms
356 * bug fix: restore errno in signal handlers
357 * bug fix: fix deadlocks related to hash tables
358 * bug fix: fix deadlocks in pcl
360 changes in sbcl-1.0.25 relative to 1.0.24:
361 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
362 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
363 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
364 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
365 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
366 better name and does not return values so stale on multiprocessor systems.
367 Also, HOLDING-MUTEX-P was added for about the only sane usage of
369 * improvement: unithread builds keep track of MUTEX-VALUE.
370 * improvement: reading from a TWO-WAY-STREAM does not touch the output
371 stream anymore making it thread safe to have a concurrent reader and
372 a writer, for instance, in a pipe.
373 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
374 arguments, making compiler-macros for SETF-functions able to inspect
375 their constant arguments.
376 * improvement: COMPILE-FILE reports times with millisecond accuracy
377 (thanks to Luis Oliveira)
378 * optimization: CHAR-CODE type derivation has been improved, making
379 TYPEP elimination on subtypes of CHARACTER work better. (reported
380 by Tobias Rittweiler, patch by Paul Khuong)
381 * bug fix: setting alien structure fields of type struct by value now
382 computes the right offset for the memory copy.
383 * bug fix: compilation problem involving inlined calls to aliens with
384 result type VOID. (reported by Ken Olum)
385 * bug fix: #235a; sequential inline expasion in different policies no
386 longer reuses the functional from the previous expansion site.
387 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
388 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
389 * bug fix: named ENUMs in multiply-referenced alien record types no longer
390 cause an implied type redefinition. Regression from 1.0.21.29.
391 * bug fix: sign-extension of small signed return values now works with C
392 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
393 * improvements to the Windows port:
394 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
395 Westfold, thanks to Rudi Schlatte)
397 changes in sbcl-1.0.24 relative to 1.0.23:
398 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
399 vector of a multidimensional SIMPLE-ARRAY.
400 * new feature: the system now signals a continuable error if standard
401 readtable modification is attempted. (thanks to Tobias Rittweiler)
402 * new feature: DIRECTORY has been extended with a non-standard keyword
403 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
404 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
405 has been robustified and documented better. (thanks to Robert Goldman)
406 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
407 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
408 * tradeoff: constant FORMAT control strings are now compiled unless
409 SPACE > SPEED (previously only when SPEED > SPACE.)
410 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
412 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
414 * bug fix: lexical type declarations are now correctly reported by
415 SB-CLTL2. (reported by Larry D'Anna)
416 * bug fix: STRING-TO-OCTETS did not handle :START properly when
417 using UTF-8 as external format. (thanks to Luis Oliveira)
418 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
419 are signalled correctly. (thanks to Stas Boukarev)
420 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
421 designator. (thanks to Tobias Rittweiler)
422 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
423 designator, and returns T instead of the function. (thanks to
425 * bug fix: direct superclasses of STANDARD-CLASS and
426 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
427 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
428 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
429 when no class with the named slot yet exists no longer causes a
430 compile-time style-warning.
431 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
432 in safe code. (reported by Didier Verna)
433 * bug fix: #430; nested structure constructors can stack allocate.
434 * bug fix: on some 64-bit platforms dynamic space size was truncated
435 to #xffff0000 bytes. (reported by Benjamin Lambert)
436 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
437 REPL unusable. (reported by Daniel Herring)
438 * bug fix: return values of READ-SEQUENCE did not take :START into
439 account on file streams, regressions since 1.0.12.22. (reported by
440 Thomas Russ, patch by Paul Khuong)
441 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
442 method specializer used to confuse permuation vector optimization.
443 * bug fix: system inserted bogus implicit type declarations for local
444 special variables in DEFMETHOD bodies.
445 * bug fix: #354; duplicated frames in backtraces due to
446 non-tail-call-optimized XEPs to functions with return type NIL
448 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
449 initializes structure object slots according to DEFSTRUCT initforms,
450 and DEFSTRUCT forms :INCLUDEind structure classes defined using
451 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
452 (reported by Bruno Haible and Stephen Wilson)
453 * bug fix: #395; fill-pointer output streams used now support
454 element-type BASE-CHAR as well.
455 * bug fix: compiler error when attempting to derive return value of
456 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
458 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
460 * bug fix: Mac OS X binaries should now be portable between Leopard
463 changes in sbcl-1.0.23 relative to 1.0.22:
464 * enhancement: when disassembling method functions, disassembly
465 for the associated fast function is also produced.
466 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
468 * optimization: printing with *PRINT-PRETTY* true is now more
469 efficient as long as the object being printed doesn't require
470 special handling by the pretty printer.
471 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
472 now interact correctly with type declarations.
473 * partial bug fix: PCL detects infinite recursion during wrapper
474 validation. (thanks to Attila Lendvai)
475 * bug fix: #426; nested function calls are inlined properly.
476 Previously if FOO was an inline function, in calls of the form
477 (FOO (FOO ...)) the outer call was not inlined.
478 * bug fix: long long arguments passed in registers to alien
479 functions on PowerPC Linux and NetBSD are now handled correctly
480 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
482 changes in sbcl-1.0.22 relative to 1.0.21:
483 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
484 for the shared object in the current directory, but passes the native
485 namestring of the designated pathname to the operation system's shared
486 object loading function as-it.
487 * minor incompatible change: --disable-debugger toplevel option now takes
488 effect before processing of initialization files and --eval or --load
490 * new feature: new commandline argument: --script, which supports
491 shebang lines. See documentation for details. (based on work by
493 * new feature: SAVE-LISP-AND-DIE can save current values of
494 --dynamic-space-size and --control-stack-size in the executable core,
495 causing it to skip normal runtime option processing. See documentation
496 for details. (thanks to Zach Beane)
497 * enhancement: inoccous calls to EVAL or generic functions dispatching
498 on subclasses of eg. STREAM no longer cause compiler notes to appear.
499 * enhancement: the system no longer resignals errors from --load and
500 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
501 associated with the original error to be lost. (thanks to Ariel
503 * enhancement: :DONT-SAVE keyword argument has been added to
504 LOAD-SHARED-OBJECT for controlling interaction with
505 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
506 the effects of an earlier LOAD-SHARED-OBJECT call.
507 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
508 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
509 provided. (reported by Cedric St-Jean)
510 * bug fix: circularity handling in the reader did not treat raw
511 structure slots correctly. (reported by Cedric St-Jean)
512 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
513 file descriptors when there were none.
514 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
515 was broken. (reporteed by Willem Broekema)
516 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
517 pathnames without a directory.
518 * bug fix: COMPILE-FILE did not warn about undefined variable
519 references at toplevel, and LOAD of the resulting fasl did
521 * bug fix: functions with non-required arguments used to end up with
522 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
523 * bug fix: redefining a function with non-required arguments didn't
524 update the system's knowledge about its call signature properly.
525 * bug fix: fixed #431; incompatible alien record type redefinitions
526 are detected and handled. (reported by Neil Haven)
527 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
528 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
530 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
531 after alien stack frames.
532 * bug fix: strings passed to mk{s,d}temp lacked null termination
534 changes in sbcl-1.0.21 relative to 1.0.20:
535 * new feature: the compiler is able to track the effective type of a
536 generic function across method addition and removal.
537 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
538 appropriately typed structure slots without locking.
539 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
540 information like those gathered by TIME using a programming-friendly
542 * new feature: TIME reports time taken even if the form performs a
543 non-local transfer of control.
544 * enhancement: reduced conservativism on GENCGC platforms: on
545 average 45% less pages pinned (measured from SBCL self build).
546 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
547 mutate constant symbols or violate declaimed type of the symbol.
548 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
549 sleeping on the semaphore in heavily contested semaphores.
550 * bug fix: semaphores and condition variables were not interrupt
552 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
553 owned by other threads anymore.
554 * bug fix: FIND on lists called KEY outside the specified
555 subsequence. (reported by budden)
556 * bug fix: LOG doesn't use single-float intermediate results when
557 given mixed integer and double-float arguments, leading to better
558 precision. (reported by Bob Felts)
559 * bug fix: LOG with base zero returned values of inconsistent type.
561 changes in sbcl-1.0.20 relative to 1.0.19:
562 * minor incompatible change: OPTIMIZE qualities
563 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
564 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
565 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
566 * documentation: some slot access efficiency guidelines have been
567 added to the user manual.
568 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
569 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
570 as ASSOC and MEMEBER.
571 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
572 to more efficient EQ-comparison versions more often.
573 * optimization: enhanced derivation of DOLIST iteration variable type
575 * optimization: constant folding of simple (LIST ...) forms as DOLIST
577 * optimization: runtime lookup of function definitions can be
578 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
579 * optimization: compiler is able to derive the return type of
580 (AREF (THE STRING X) Y) as being CHARACTER.
581 * optimization: CLRHASH on empty hash-tables no longer does pointless
582 work. (thanks to Alec Berryman)
583 * optimization: REPLACE deftransforms don't punt when :START1 or
585 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
586 no longer emits a code deletion note. (reported by Andrew
588 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
589 functions with an odd number of &OPTIONAL arguments, a &REST
590 argument, and one or more &KEY arguments at the call site.
591 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
592 arguments appeared at call sites as well.
593 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
594 READ-LINE &co can no longer read from them afterwards. (reported
596 * bug fix: fixed #427: unused local aliens no longer cause compiler
597 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
599 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
600 alien-stack leakage. (reported by Andy Hefner)
601 * bug fix: PROGV signals an error when an attempt to violate declared
602 type of a variable or bind a constant is made.
603 * bug fix: SET signals an error when an attempt to violate declared
604 type of a variable is made.
605 * bug fix: restart computation during the execution of a restart
606 test function no longer causes infinite recursion. (reported by
608 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
609 signals an error. (thanks to Pierre Mai)
610 * bug fix: EXPORT left symbol unexported in conflict situations.
611 (thanks to Michael Weber)
612 * bug fix: correctly handle name conflicts involving CL:NIL.
613 (thanks to Michael Weber)
614 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
615 conflicts arising from USEing package with conflicting symbols
616 correctly. (thanks to Michael Weber)
618 changes in sbcl-1.0.19 relative to 1.0.18:
619 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
620 warnings that go otherwise unhandled will be muffled if they are
621 of the type that's the value of this variable.
622 * optimization: stack allocation is slightly more efficient on x86
624 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
625 where a raw slot always is initialized using the initform whose
626 type is not know sufficiently well a compile-time are now compiled
627 correctly. (reported by John Morrison)
628 * bug fix: compiler no longer makes erronous assumptions in the
629 presence of non-foldable SATISFIES types.
630 * bug fix: stack analysis missed cleanups of dynamic-extent
631 arguments in non-let-converted calls to local functions.
632 * improvements to the Windows port:
633 ** adjusted address spaces for building on both Win32 and
634 Win64. (thanks for John Connors)
635 * fixed some bugs revealed by Paul Dietz' test suite:
636 ** interval arithmetic during type derivation used inexact integer
637 to single-float coercions.
638 ** arithmetic operations involving large integers and single
639 floats give the same results in compiled and interpreted code.
640 ** deriving the result type of COERCE no longer signals an error
641 if the derived type of the second argument is a MEMBER type
642 containing invalid type specifiers.
643 ** ADJOIN with constant NIL as second argument works correctly.
644 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
646 changes in sbcl-1.0.18 relative to 1.0.17:
647 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
648 profiles only the current thread.
649 * minor incompatible change: changes to SYMBOL-VALUE of constants
650 defined with DEFCONSTANT now signal an error.
651 * enhancement: SB-SPROF now has support for wallclock profiling,
652 and is also able to profile specific threads. REPORT output
653 has also additional sorting options.
654 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
656 * optimization: structure allocation has been improved
657 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
658 ** out of line constructors are ~10% faster.
659 ** inline constructors are ~15% faster.
660 ** inline constructors are capable of dynamic extent allocation
661 (generally on x86 and x86-64, in some cases on other platforms
663 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
665 * optimization: file compiler is now able to coalesce non-circular
666 lists, non-base strings, and bit-vectors. Additionally, constants
667 are never referenced through SYMBOL-VALUE at runtime.
668 * optimization: code defining methods on PRINT-OBJECT (and other
669 generic functions in the COMMON-LISP package) now loads faster.
670 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
672 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
673 fasl is now deleted. (reported by Attila Lendvai)
674 * bug fix: READ-LINE always returned NIL for the last line in files.
675 (reported by Yoshinori Tahara)
676 * bug fix: more accurate disassembly annotations of foreign function
677 calls. (thanks to Andy Hefner)
678 * bug fix: trimming non-simple strings and non-string string
679 designators when the there is nothing to trim works properly.
680 (thanks to James Knight)
681 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
682 (thanks to Travis Cross)
683 * fixed some bugs revealed by Paul Dietz' test suite:
684 ** NIL is a valid function name (regression at 1.0.13.38)
685 ** FILL on lists was missing its return value (regression at 1.0.12.27)
686 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
687 fill pointers properly (regression at 1.0.12.23)
688 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
689 displacement indices properly (regression at 1.0.12.23)
691 changes in sbcl-1.0.17 relative to 1.0.16:
692 * temporary regression: user code can no longer allocate closure
693 variable storage on stack, due to bug 419 without explicitly
694 requesting it. Please consult sbcl-devel for advice if you need to
695 use this feature in the meanwhile.
696 * new feature: runtime argument --control-stack-size can be used to
697 adjust thread default control stack size.
698 * enhancement: improved TIME output
699 ** all times are reported using the measured accuracy (milliseconds
700 for real and GC times, microseconds for everything else.)
701 ** processor cycle counts on x86 and x86-64.
702 ** interpreted forms are counted for both evaluator modes.
703 ** number of lambdas converted by the compiler is reported.
704 ** CPU percentage report (computed from real and total run time.)
705 ** more comprehensive run time reporting, using a condenced format
706 ** interperted form, lambda, and page fault counts are omitted
708 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
710 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
711 * optimization: two argument forms of LAST are upto ~10% faster
712 in normal SPEED policies.
713 * optimization: NCONC no longer needs to heap cons its &REST list
714 in normal SPEED policies.
715 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
717 * bug fix: UNION and NUNION work with :TEST-NOT once more,
718 regression since 1.0.9.1. (thanks to Eric Marsden)
719 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
720 since 1.0.15.36. (thanks to Paul Khuong)
721 * bug fix: LAST when always returned the whole list when given a bignum
722 as the second argument.
723 * bug fix: dynamic extent allocation of nested lists and vectors
724 could leak to otherwise accessible parts.
725 * bug fix: invalid optimization of heap-allocated alien variable
727 * bug fix: fasl header checking is less vulnerable to different
728 platform word lengths.
729 * bug fix: more correct assembler syntax for GNU binutils
730 2.18.50.0.4 support. (thanks to Marijn Schouten)
731 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
733 * bug fix: the fopcompiler can handle LOCALLY forms (with no
734 declarations) successfully. (reported by Attila Lendvai)
736 changes in sbcl-1.0.16 relative to 1.0.15:
737 * minor incompatible change: revert the changes to sb-posix's error
738 signaling added in 1.0.14.
739 * minor incompatible change: change PROBE-FILE back to returning
740 NIL whenever we can't get a truename, as was the case before 1.0.14.
741 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
742 inherits from ERROR instead of just CONDITION.
743 * new feature: SB-INTROSPECT can provide source locations for instances
744 as well. (thanks to Tobias Rittweiler)
745 * optimization: binding special variables now generates smaller code
746 on threaded platforms.
747 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
748 and cases where no :TEST is given but the compiler can infer that
749 the element to search is of type (OR FIXNUM (NOT NUMBER)).
750 * optimization: better LOGNOT on fixnums.
751 * optimization: modular arithmetic for a particular requested width
752 is implemented using a tagged representation unless a better
753 representation is available.
754 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
755 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
756 for wellformedness and returns a specialized vector. (reported by
757 Francois-Rene Rideau)
758 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
759 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
760 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
761 unscheduling timer at the same time as another timer fires.
762 * bug fix: don't reschedule timers for dead threads.
763 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
764 * bug fix: copying output from RUN-PROGRAM to a stream signalled
765 bogus errors if select() was interrupted.
766 * enhancement: add support for fcntl's struct flock to SB-POSIX.
768 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
769 * enhancement: cleaner backtraces for interactive interrupts, as
770 well as other cases where the interesting frames used to be
771 obscured by interrupt handling frames.
772 * enhancement: untracing a whole package using (UNTRACE "FOO") is
773 now supported, and tracing a whole package using (TRACE "FOO") now
774 traces SETF-functions as well.
775 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
776 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
777 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
778 * unreadably printed representation of hash-tables now includes
780 * bug fix: partially fixed #188: type propagation from assignments
781 is now more efficient.
782 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
783 'DOUBLE-FLOAT) are not flushable.
784 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
785 full address of the object, and none of the tag bits.
786 * bug fix: readably printing hash-tables now respects other printer
787 control variables. (reported by Cedric St-Jean)
788 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
789 keyword with MAKE-HASH-TABLE.
790 * bug fix: export SB-POSIX:MKSTEMP.
791 * bug fix: SORT was not interrupt safe.
792 * bug fix: XREF accounts for the last node of each basic-block as
794 * bug fix: MAKE-INSTANCE optimizations interacted badly with
795 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
796 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
797 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
798 create obsolete instances in the case of redefinition or
799 obsoletion of a superclass. (thanks to Andy Hefner)
800 * bug fix: Support for the Alpha architecture has been revived; it had
801 suffered somewhat from lack of maintenance since sbcl-1.0.
802 * improvements to the Windows port:
803 ** The system detects the codepage to use at startup, instead of
804 using the value from the saved core. (thanks to Kei Suzuki)
806 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
807 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
808 (see documentation for details.)
809 * revived support for OpenBSD (contributed by Josh Elsasser)
810 * partially fixed bug #108: ROOM no longer suffers from occasional
811 (AVER (SAP= CURRENT END)) failures.
812 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
813 bodies no longer cause a WARNING to be signalled. (reported by
815 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
816 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
817 the bug in TRUNCATE.) (reported by Michael Weber)
818 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
819 no samples. (reported by Andy Hefner)
820 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
821 no longer appear as (NIL ...) frames in backtraces.
822 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
823 is now actually exported from SB-EXT as documented. (reported by
825 * bug fix: sb-aclrepl now correctly understands how to inspect
826 single-floats on 64-bit platforms where single-floats are not boxed.
827 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
828 yet been finalized. (reported by Levente Meszaros)
829 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
830 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
831 SBCL-specific optimize qualities.
833 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
834 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
835 an executable in the search path, and does so in the child
836 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
837 been removed; it can be found in the CVS history, for somebody who
838 needs that search behavior (see the manual).
839 * minor incompatible change: compiler policy re. weakening type
840 checks has changed: now type checks are weakened only if SAFETY < 2
842 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
843 unparsing of directory pathnames as files. Analogously,
844 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
845 filename to parse into a directory pathname.
846 * enhancement: implicit generic function creation now signals a
847 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
848 users can bind handlers for and muffle around calls to LOAD.
849 * enhancement: RUN-PROGRAM allows unicode arguments and environments
850 to be used (using the default stream external format), and allows
851 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
852 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
853 strings and vectors whose element-type or simplicity is not fully
854 known at compile-time.
855 * optimization: STRING-TRIM and related functions no longer allocate
856 a new string when no trimming needs to be performed. These functions
857 are also faster than before when the input string has been declared
859 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
860 * optimization: READ-LINE is significantly faster for files containing
862 * optimization: non-open coded uses of character comparison operators
863 (e.g. char=) no longer cons when called with more than one parameter
864 on platforms supporting dynamic-extent allocation.
865 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
866 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
867 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
869 * bug fix: some sequence functions elided bounds checking when
871 * bug fix: too liberal weakening of union-type checks when SPEED >
873 * bug fix: more bogus fixnum declarations in ROOM implementation
876 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
877 * new feature: MAKE-HASH-TABLE now experimentally accepts a
878 :SYNCHRONIZED argument, which makes the hash-table safe for
879 concurrent accesses (but not iteration.) See also:
880 SB-EXT:WITH-LOCKED-HASH-TABLE, and
881 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
882 * optimization: CONCATENATE on strings is an order of magnitude faster
883 in code compiled with (> SPEED SPACE).
884 * optimization: SUBSEQ is ~50% faster on lists.
885 * bug fix: bug 417 fixed -- source location reporting is now more robust.
886 * bug fix: SUBSEQ on a list will now correctly signal an error if if
887 END is smaller then START.
888 * bug fix: SB-PROFILE will no longer report extra consing for nested
889 calls to profiled functions.
890 * bug fix: ROOM implementation had bogus fixnum declarations which
891 could cause type-errors when calling ROOM on large images.
892 * bug fix: if file compilation is aborted, the partial fasl is now
893 deleted, and COMPILE-FILE returns NIL as the primary value.
894 * bug fix: number of thread safety issues relating to SBCL's internal
895 hash-table usage have been fixed.
896 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
897 be returned from its body when the values were being returned
898 using unknown-values return convection and the W-P-O was wrapped
899 inside an UNWIND-PROTECT.
900 * bug fix: sb-posix should now compile again under Windows, enabling
903 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
904 * incompatible change: hash-table accessor functions are no longer
905 automatically protected by locks. Concurrent accesses on the same hash-table
906 from multiple threads can give inconsistent results or even corrupt the
907 hash-table completely. Multi-threaded applications should do their own
908 locking at the correct granularity. In the current implementation it is
909 still safe to have multiple readers access the same table, but it's not
910 guaranteed that this property will be maintained in future releases.
911 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
912 and will signal an error at runtime.
913 * enhancement: SB-THREAD package now exports a semaphore interface.
914 * enhancement: CONS can now stack-allocate on x86 and
915 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
917 * enhancement: nested lists can now be stack allocated on
918 platforms providing stack allocation support.
919 * enhancement: dynamic-extent support has been extended to support
920 cases where there are multiple possible sources for the stack
922 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
923 if the mutex is uncontested on Linux.
924 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
925 as specified by the CLHS. (thanks to Tobias Rittweiler)
926 * bug fix: NaN comparison now works on x86-64.
927 * bug fix: CLOSE :ABORT T on a stream with pending output now
929 * bug fix: instances of non-standard metaclasses using standard
930 instance structure protocol sometimes missed the slot type checks
932 * bug fix: known functions can be inlined in high-debug code.
934 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
935 * minor incompatible change: the MSI installer on Windows no longer
936 associates .lisp and .fasl files with the installed SBCL.
937 * minor incompatible change: :UNIX is no longer present in *FEATURES*
938 on Windows. (thanks to Luis Oliviera)
939 * new feature: SB-CLTL2 contrib module now implements
940 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
941 * optimization: scavenging weak pointers is now more efficient,
942 requiring O(1) instead of O(N) per weak pointer to identify
943 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
944 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
945 method bodies using constant slot-names when the first argument is
946 a specializer parameter for the method.
947 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
948 names, when the class of the instance is a direct instance
949 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
950 as slow as the constant slot-name case.)
951 * optimization: member type construction is now O(N) instead
953 * optimization: UNION and NUNION are now O(N+M) for large
954 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
956 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
957 now more readable in environments like Slime which display it.
958 (thanks to Tobias C. Rittweiler)
959 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
960 was able to infer the correct argument type for the object on which
961 the CAS operation was being performed.
962 * bug fix: copy propagation interfered with parallel assignment
963 semantics in local calls. (reported by Paul Khuong)
964 * bug fix: the signed modular fixnum shift compiled to wrong code on
965 x86 and x86-64. (spotted by a slight modification to some of
968 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
969 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
970 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
971 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
973 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
974 and generic function now print the object identity for anonymous
975 instances. (thanks to Attila Lendvai)
976 * enhancement: as an extension to MOP, SBCL now supports SETF of
977 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
978 (thanks to Attila Lendvai)
979 * enhancement: sb-cover has an alternate annotation mode which puts
980 the form annotations at the car of the form, instead of annotating
981 it completely. The benefit of this mode is that it shows better
982 which forms were instrumented by the compiler.
983 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
984 by printer control variables. (reported by Dan Corkill)
985 * bug fix: system leaked memory when delayed output was performed by
986 the OS in smaller chunks then expected. (thanks to David Smith)
987 * bug fix: system leaked memory when file streams were not closed
989 * bug fix: large objects written to slow streams that were modified
990 after the write could end up with the modified state written to
991 the underlying file descriptor.
992 * bug fix: multiple threads operating in parallel on the same stream
993 could cause buffer-overflows.
994 * bug fix: source location information is stored correctly
995 (broken since 1.0.6). This bug would generally show up as the
996 Slime debugger highlighting the wrong form when the sldb-show-source
998 * bug fix: Branch forms are again annotated as branches in the sb-cover
1000 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
1001 on x86-64. (reported by Gregory Vanuxem)
1002 * bug fix: EQUALP could return wrong results for structures with raw
1003 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
1004 word). (reported by Vjacheslav Fyodorov)
1005 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
1008 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
1009 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
1010 atomic compare-and-swap operations on threaded platforms.
1011 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
1012 allows assigning a global minimum value to optimization qualities
1013 (overriding proclamations and declarations).
1014 * enhancement: closed over variables can be stack-allocated on x86
1016 * performance bug fix: GETHASH and (SETF GETHASH) are once again
1018 * optimization: slot definition lookup is now O(1). This speeds up
1019 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
1020 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
1021 * optimization: ASSOC and MEMBER can now be open-coded for all
1022 combinations of keyword arguments when second argument is constant
1023 and SPEED >= SPACE. In other cases a specialized version is
1025 * bug fix: using obsoleted structure instances with TYPEP and
1026 generic functions now signals a sensible error.
1027 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
1028 (reported by Kristoffer Kvello)
1029 * bug fix: backtrace construction is now more careful when making
1030 lisp-objects from pointers on the stack, to avoid creating bogus
1031 objects that can be seen by the GC.
1032 * bug fix: defaulting of values in contexts expecting more than 7
1033 variables now works on x86-64. (reported by Christopher Laux)
1034 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
1036 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
1037 as the property-list of a symbol.
1038 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
1039 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
1040 situations requested, are once again file-compileable. (reported
1043 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
1044 * MOP improvement: support for user-defined subclasses of
1045 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
1046 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
1047 the expansion of DEFMETHOD, is responsible for generating a form
1048 which creates a list of specializers when evaluated. Additional
1049 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
1050 debugging and introspective support.
1051 * minor incompatible change: the (unsupported) spinlock interface
1052 has changed: free spinlock now has the value NIL, and a held spinlock
1053 has the owning thread as its value.
1054 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
1055 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
1057 * enhancement: name of a socket-stream is now "a socket" instead of
1058 "a constant string".
1059 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
1060 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
1061 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
1062 for generic functions. (Thanks to Tobias C. Rittweiler)
1063 * optimization: bignum printing speed has been improved by 20-40%
1064 (depending on the bignum size.)
1065 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
1067 * bug fix: the cache used by the CLOS to store precomputed effective
1068 methods, slot offsets, and constant return values is now thread and
1070 * bug fix: generic function dispatch function updating is now thread
1071 and interrupt safe (in the sense that the known issues have been
1073 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
1074 * bug fix: interning EQL-specializers is now thread and interrupt safe.
1075 * bug fix: asdf systems with dependencies to the SB-POSIX or
1076 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
1077 * bug fix: interrupt safety of applicable method computation has been
1080 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
1081 * new contrib: sb-cover, an experimental code coverage tool, is included
1082 as a contrib module.
1083 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
1084 significantly faster.
1085 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1086 produce more efficient code on x86-64 if the compiler can determine
1087 that (+ INDEX <constant>) does not require a bounds check and FOO
1088 has an element type at least 8 bits wide.
1089 * enhancement: a new, experimental synchronous timeout facility is
1090 provided. Refer to SB-SYS:WITH-DEADLINE for details.
1091 * enhancement: when a symbol name conflict error arises, the
1092 conflicting symbols are always printed with a package prefix.
1093 (thanks to Kevin Reid)
1094 * enhancement: stepping is now once again supported on the SPARC and
1095 MIPS platforms. (It is also now more likely to work on CheneyGC
1097 * enhancement: sb-sprof can now also track and report accurate call
1099 * bug fixes: the treatment of non-standard subclasses of
1100 SB-MOP:SPECIALIZER is more correct.
1101 * incompatible change: PURIFY no longer copies the data from the
1102 dynamic space into the static and read-only spaces on platforms
1103 that use the generational garbage collector
1104 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
1106 * bug fix: GC race condition occasionally resulting in crashes with
1107 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
1109 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
1111 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
1112 a STYLE-WARNING to be signalled (regression from 1.0.4.)
1113 * bug fix: an asynchronous interrupt could previously leave the
1114 system running with GC inhibited.
1115 * bug fix: a DECLARE form evaluated at top-level now causes an error
1116 rather than silently (or verbosely) returning NIL.
1117 * bug fix: trying to dissassemble functions compiled at high DEBUG could
1118 result in a error being signalled due to source form lookup errors.
1119 (reported by Peter Graves)
1121 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
1122 * incompatible change: removed writer methods for host-ent-name,
1123 host-ent-addresses -- changing the values did not update the DNS
1125 * minor incompatible change: changed experimental JOIN-THREAD interface
1126 * documentation: the manual now lists reader and writer methods
1127 in class slot documentation sections. (thanks to Richard M Kreuter)
1128 * documentation: unwinding from asyncronous events has been
1129 documented as unsafe.
1130 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
1131 in multithreaded application code.
1132 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
1133 platforms. (thanks to James Anderson for the optimization hint)
1134 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
1136 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
1137 variants no longer cons.
1138 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
1139 their NOT- variants no longer cons.
1140 * optimization: EQUAL hash tables no longer use SXHASH for objects
1141 of all data types, but instead use an EQL hash for types for which
1142 EQUAL is the same as EQL.
1143 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
1144 are significantly faster.
1145 * optimization: new STRING-OUTPUT-STREAM implementation is much
1146 faster for large outputs and conses 30% less on average.
1147 * enhancement: XREF information is now collected for references made
1148 to global variables using SYMBOL-VALUE and a constant argument.
1149 * enhancement: SIGINT now causes a specific condition
1150 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
1151 * bug fix: FILE-POSITION used to signal an error in some cases where
1152 ANSI requires it to return NIL.
1153 * bug fix: ADJUST-ARRAY is now interrupt-safe.
1154 * bug fix: adding and removing fd-handlers is now interrupt-safe.
1155 * bug fix: inlined calls to C now ensure 16byte stack alignment on
1157 * bug fix: bad type declaration in the CLOS implementation has
1158 been fixed. (reported by James Anderson)
1159 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
1160 (thanks to Lutz Euler)
1161 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
1162 required by the CLHS. (reported by Samium Gromoff)
1163 * bug fix: dead unbound variable references now signal an error.
1164 (reported by Marco Monteiro)
1165 * bug fix: / with an unused value was being deleted in safe code.
1166 (thanks to Marco Monteiro and Kevin Reid)
1167 * bug fix: number of characters that can be written onto a single
1168 line in a file is unlimited.
1169 * bug fix: some GC deadlocks caused by asynchronous interrupts have
1170 been fixed by inhibiting interrupts for when GC is disabled.
1171 * bug fix: some interrupt safety issues with GETHASH, (SETF
1172 GETHASH), CLRHASH and REMHASH have been fixed.
1173 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
1174 specifier no longer causes infinite recursion.
1175 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
1176 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
1177 * bug fix: an &environment argument with the correct variable information
1178 is passed to macros that are expanded during byte compilation.
1179 (reported by Samium Gromoff)
1180 * bug fix: a logic error was causing the hash function for bignums to
1181 have excessive amounts of collisions. (reported by Faré Rideau)
1182 * bug fix: modifying the contents of an array could change the return
1183 value of SXHASH on that array, which is only allowed for strings
1184 and bit vectors. (bug introduced in 0.9.16)
1185 * improvement: the x86-64/darwin port now passes all tests (except
1186 for the debugger tests) but should still be considered
1187 experimental until this is fixed.
1188 * improvement: a style-warning is signaled for CASE (etc) clauses with
1189 duplicate keys. (thanks to Kevin Reid)
1190 * improvement: macos/x86 and macos/x86-64 now use mach exception
1191 handlers for dealing with illegal instructions (for trapping and
1192 error handling) and memory protection violations (for GC).
1194 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
1195 * new platform: experimental support for x86-64/darwin (MacOS).
1196 * incompatible change: the thread-safe (on most platforms) getaddrinfo
1197 and getnameinfo sockets functions are used instead of gethostbyaddr
1198 and gethostbyname, on platforms where the newer functions are available.
1199 As a result, the ALIASES field of HOST-ENT will always be NIL on these
1201 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
1202 don't print the startup banner, but behave as if --noinform was passed
1203 as a command line argument. (thanks to Kevin Reid)
1204 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1205 * optimization: code using alien values with undeclared types is much faster.
1206 * optimization: the compiler is now able to open code SEARCH in more cases.
1207 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1208 * bug fix: using standardized COMMON-LISP special variables as loop
1209 variables no longer signals bogus package lock violations. (reported
1211 * bug fix: declaring local loop variables to be of a range-limited type
1212 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1213 (reported by Andras Simon)
1214 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1215 bugs remain on x86-64.)
1216 * bug fix: the #= and ## reader macros now interact reasonably with
1217 funcallable instances.
1218 * bug fix: type-checks for function arguments were compiled using the
1219 compiler policy settings of the wrong lexical environment (in 1.0.2
1221 * bug fix: SHADOW accepts characters as string designators, as required
1222 by the spec (thanks to Eric Marsden)
1223 * bug fix: fixed GC safety issues when foreign functions are called with
1224 non-base strings as arguments
1225 * bug fix: more consistent error output for fatal-compiler-warnings, like
1227 * bug fix: functions with proper names were showing up as NIL in
1228 backtraces (reported by Edi Weitz, regression in 1.0.2)
1230 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1231 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1232 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1233 produce more efficient code on the x86 if the compiler can determine
1234 that (+ INDEX <constant>) does not require a bounds check and FOO
1235 has an element type at least 8 bits wide.
1236 * bug fix: references to undefined variables are handled the same way
1237 in toplevel forms as in the normal compiler
1238 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1240 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1241 (thanks to Jon Buller)
1242 * improvement: faster compilation times for complex functions
1243 * improvement: added readlink support to SB-POSIX (thanks to Richard
1246 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1247 * improvement: experimental support for mach exception handling on
1248 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1250 * improvement: support for GBK external format.
1251 (thanks to Chun Tian (binghe))
1252 * improvement: the debugger now displays variables that have been closed
1253 over, in code compiled with (DEBUG 2) or higher
1254 * improvement: support for executable cores on NetBSD (thanks to
1256 * new feature: added a RESTART-FRAME debugger command
1257 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1258 be used to provide an implementation for FILE-POSITION on Gray streams
1259 (thanks to Eric Marsden)
1260 * new feature: add syslog, openlog and closelog support to SB-POSIX
1261 (thanks to Richard Kreuter)
1262 * optimization: the function call overhead in code compiled with
1263 a high DEBUG optimization setting is significantly
1264 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1265 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1266 (thanks to Tony Martinez)
1267 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1268 evaluated once (reported by Kevin Reid)
1269 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1270 works again (thanks to Kilian Sprotte)
1271 * bug fix: an error is signaled for tagbodies with duplicate tags
1272 (thanks to Stephen Wilson)
1273 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1274 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1275 stack frames from alien callbacks.
1276 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1277 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1278 65k pages (thanks to David Woodhouse)
1279 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1281 changes in sbcl-1.0.1 relative to sbcl-1.0:
1282 * new platform: FreeBSD/x86-64, including support for threading.
1283 * new feature: the compiler stores cross-referencing information
1284 about function calls (who-calls), macroexpansion (who-macroexpands)
1285 and special variables (who-binds, who-sets, who-references) for code
1286 compiled with (< SPACE 3). This information is available through the
1287 sb-introspect contrib.
1288 * new feature: users may subclass SEQUENCE, and have instances of
1289 these classes interoperate with standard Common Lisp functions if
1290 a number of methods are defined. (This feature is experimental
1291 and the interface subject to change based on feedback from SBCL
1292 users and the general community)
1293 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1294 x86/x86-64, rather than the previous fixed depth of 8
1295 * improvement: another pthread back-end of mutex "pthread-futex".
1296 * bug fix: non-ascii command-line arguments are processed correctly
1297 (thanks to Yaroslav Kavenchuk)
1298 * bug fix: non-required arguments were not passed correctly when a method
1299 defined using DEFMETHOD was called from a mop-generated method using
1300 CALL-NEXT-METHOD (reported by Pascal Costanza)
1301 * bug fix: recursion is now permitted in accessors through
1302 SLOT-UNBOUND. (reported by Pascal Costanza)
1303 * bug fix: an error was signaled at startup if the HOME environment
1304 variable was defined, but had an empty value (reported by Peter Van Eynde)
1305 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1306 signal a PROGRAM-ERROR, not a BUG.
1307 * bug fix: SB-POSIX:READDIR works when built with large file support.
1308 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1310 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1311 for better automatic indentation support. (thanks to Matt Pillsbury)
1312 * optimization: loading generic functions no longer takes O(n^2) time,
1313 proportional to the amount of methods in the generic function
1314 (reported by Todd Sabin and Jeremy Brown)
1315 * optimization: the FIND and POSITION family of sequence functions
1316 are significantly faster on arrays whose element types have been
1318 * improvements to the Windows port:
1319 ** Intermittent heap corruption problems have been fixed. (thanks
1320 to Alastair Bridgewater)
1321 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1323 ** Lisp is able to unwind foreign exception frames from alien
1324 callbacks. (thanks to Alastair Bridgewater)
1326 changes in sbcl-1.0 relative to sbcl-0.9.18:
1327 * improvement: experimental support for threading on FreeBSD/x86.
1328 (thanks to NIIMI Satoshi)
1329 * improvement: runtime option --dynamic-space-size can be used
1330 to set the size of the dynamic space reserved on startup.
1331 * improvement: floating point modes in effect are now saved in
1332 core, and restored on startup.
1333 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1334 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1335 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1336 threads are not running after *SAVE-HOOKS* have run.
1337 * improvement: writes to CLOS instance slots are type-checked in code
1338 compiled with (SAFETY 3)
1339 * improvement: floating-point exception handling on FreeBSD (thanks to
1341 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1342 (thanks to Zach Beane)
1343 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1345 * improvement: added support for the Shift-JIS external format.
1346 (contributed by NIIMI Satoshi)
1347 * improvement: callbacks are supported on Linux/PPC. (thanks to
1349 * bug fix: compiler bug triggered by a (non-standard) VALUES
1350 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1351 * bug fix: file compiler no longer confuses validated and already
1352 dumped structurres. (reported by Kaersten Poeck)
1353 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1354 fill-pointer signals a type-error as required. (thanks to
1356 * bug fix: disassemly of funcallable instances works.
1357 * bug fix: single stepping on PPC.
1358 * bug fix: fix thread-safety problems in the type system (generally
1359 manifesting as nonsensical errors like "STRING is a bad type specifier
1360 for sequences" or "The value 1 is not of type FIXNUM")
1361 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1362 (reported by Josip Gracin)
1363 * bug fix: an error is signaled for attempts to displace arrays with
1364 incompatible element types (thanks to Mario Mommer)
1365 * bug fix: more correct handling of wide characters in debug info
1366 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1367 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1368 and don't cause extra consing
1369 * optimization: MAP and MAP-INTO are significantly faster on vectors
1370 whose elements types have been declared.
1371 * Improvements to SB-SPROF:
1372 ** Support for allocation profiling
1373 ** Reduced profiling overhead, especially for long profiling runs
1374 * Improvements to the Windows port:
1375 ** floating point exceptions are now reported correctly.
1376 ** stack exhaustion detection works partially.
1377 ** more accurate GET-INTERNAL-REAL-TIME.
1378 ** better RUN-PROGRAM behaviour in case of errors in the child
1380 ** PROBE-FILE now simplifies pathnames correctly.
1381 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1382 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1383 1st 1970 (although time timezone data will be unreliable).
1384 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1386 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1387 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1388 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1390 * bug fix: two potential GC deadlocks affecting threaded builds.
1391 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1392 returns T (reported by Anton Kazennikov)
1393 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1394 with non-variable places
1395 * bug fix: remove a race condition in the setting of
1396 funcallable-instance functions, this should make threaded CLOS
1397 code more stable against memory faults.
1398 * bug fix: corruption of specials when unbinding is interrupted by an
1399 asynchronous unwind (reported by Hannu Koivisto)
1400 * improvement: the debugger will now also display local variables that
1401 are only used once, for code compiled with a DEBUG optimization quality
1404 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1405 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1406 * incompatible change: External-format support for FFI calls. The
1407 SB-ALIEN:C-STRING no longer implies an ASCII
1408 external-format. Instead the string is subject to external-format
1409 conversion. Additionally return values of type SB-ALIEN:C-STRING
1410 are by default of type (SIMPLE-ARRAY CHARACTER), not
1411 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1412 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1413 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1414 * incompatible change: SB-EXT package no longer contains the
1415 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1416 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1418 * incompatible change: the single-stepper is no longer available
1419 on Alpha, Mips and Sparc platforms.
1420 * minor incompatible change: the direct superclasses of
1421 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1422 not (STANDARD-OBJECT FUNCTION). This makes the
1423 class-precedence-lists of GENERIC-FUNCTION and
1424 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1426 * minor incompatible change: the default stream external format on
1427 non-unicode SBCL is no longer detected from the locale, but is always
1429 * new feature: Add a version of evaluator that uses an interpreter instead
1430 of the compiler. EVAL still uses the compiler by default, to switch it
1431 to use the interpreter, set the value of the variable
1432 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1433 * minor incompatible change: the single-stepper REPL has been merged
1434 with the normal debugger (see the "Stepping" heading of the debugger help
1435 for more details). The debugger command STEP will no longer switch
1436 to the single-stepper REPL.
1437 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1438 for a type now works.
1439 * bug fix: loading of default sysinit file works. (thanks to Leonid
1441 * bug fix: better detection of circularities in the file-compiler.
1442 (reported by Marco Monteiro)
1443 * bug fix: the CL pathname functions now work with files that have
1444 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1445 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1446 systems with Unix98 pty semantics.
1447 * bug fix: ASDF-INSTALL will now work with bsd tar.
1448 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1450 * bug fix: timers expiring in dead threads no longer cause a
1451 type-error (reported by Paul "Nonny Mouse").
1452 * bug fix: thanks to more lightweight single-stepper instrumentation,
1453 code compiled with (DEBUG 3) will compile and execute significantly faster,
1454 and will have more accurate type-inferencing than before
1455 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1456 whose bindings are modified
1457 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1458 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1459 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1460 CL:SOFTWARE-VERSION with proper version
1462 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1463 * feature: implemented the READER-METHOD-CLASS and
1464 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1465 as specified by AMOP.
1466 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1468 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1469 (thanks to Lutz Euler)
1470 * optimization: hashing of general arrays and vectors has been
1471 improved. (reported by Any Fingerhut)
1472 * enhancement: SB-INTROSPECT is now able to find definitions of
1473 profiled functions. (thanks to Troels Henriksen)
1474 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1475 as well, allowing compiler-macros for SETF-functions to expand.
1476 * enhancement: step-instrumentation no longer wraps known
1477 single-value functions with multiple-value context, allowing
1478 better type inference.
1479 * fixed bug #337: use of MAKE-METHOD in method combination now works
1480 even in the presence of user-defined method classes. (reported by
1481 Bruno Haible and Pascal Costanza)
1482 * fixed bug #339(c): if there are applicable methods not part of any
1483 long-form method-combination group, call INVALID-METHOD-ERROR.
1484 (reported by Bruno Haible)
1485 * fixed bug #361: the :FUNCTION initarg in the protocol for
1486 initialization of methods can now be used to override
1487 internally-produced optimized functions. (reported by Bruno
1489 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1490 system-provided lambda expression no longer cause warnings about
1491 unbound #:|pv-table| symbols.
1492 * bug fix: improved the handling of type declarations and the
1493 detection of violations for keyword arguments with non-constant
1495 * bug fix: potentially erronous calls to PATHNAME and
1496 MERGE-PATHNAMES were being flushed in some cases.
1497 (reported by Richard Kreuter)
1498 * bug fix: compiled calls to TYPEP were mishandling obsolete
1499 instances. (reported by James Bielman and Attila Lendvai)
1500 * bug fix: format strings with ~> without matching ~< no longer
1501 trigger an AVER, but signal an understandable error instead.
1502 (reported by Antonio Martinez)
1503 * bug fix: specifying an output-file without a pathname-type for
1504 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1505 of input-file instead of "fasl". (reported by Robert Dodier)
1506 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1507 (reported by James Y Knight).
1508 * bug fix: compiler-macros-function did not consider the environment
1509 argument for shadowing by local functions.
1510 * bug fix: compiler-macros expansion was inhibited by local INLINE
1512 * bug fix: inline expansions of known functions were subject to
1513 step-instrumentation in high debug policies, leading to problems
1514 with type-inference.
1515 * bug fix: compiler failed to differentiate between different CONS
1516 types in some cases.
1517 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1518 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1519 * bug fix: #368: incorrect use of expressed vs. upgraded array
1521 * bug fix: #306a: more precise unions of array types.
1522 * thread-safety improvements:
1523 ** CONDITION-WAIT could return early on Linux, if the thread was
1524 interrupted and subsequently continued with SIGCONT.
1525 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1527 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1528 * added support for the ucs-2 external format. (contributed by Ivan
1530 * minor incompatible change: pretty printing of objects of type
1531 (cons symbol) is, in the default pprint-dispatch-table, now
1532 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1534 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1535 called later in a class's lifetime, possibly as late as when the
1536 first instance of the class is created. Previously,
1537 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1538 class became finalizeable.
1539 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1540 for multibyte external-format streams. (thanks to "vbzoli")
1541 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1542 with :ALLOCATION :CLASS inherited from superclasses of the
1544 * fixed bug: anonymous classes can now be created using the :NAME
1545 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1546 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1547 * fixed bug: core-files saved with :EXECUTABLE T can again be
1548 executed when SBCL_HOME isn't set. (reported by James Knight)
1549 * fixed bug: toplevel LOCALLY forms with declarations could
1550 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1551 * fixed bug: printing from several different threads using different
1552 values of *print-case* could cause invalid output, due to
1553 some internal special variables of the printer not being bound
1554 thread-locally (reported by Max Mikhanosha)
1555 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1556 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1557 (reported by Pascal Costanza)
1558 * fixed bug: FINALIZE-INHERITANCE is called from
1559 REINITIALIZE-INSTANCE on classes when the class has previously
1560 been finalized, as required by AMOP.
1561 * minor code generation optimizations:
1562 ** better register allocation in CLOS dispatching functions
1563 ** overflow detection when coercing signed bytes to fixnums on x86-64
1564 ** is now implemented with one IMUL instruction instead of three shifts
1565 ** more efficient bit-vector access on x86 and x86-64
1566 ** more efficient access to raw structure slots on x86 and x86-64
1567 * fixed some bugs revealed by Paul Dietz' test suite:
1568 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1569 return its argument.
1571 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1572 * feature: thread support on Solaris/x86, and experimental thread support
1574 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1576 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1577 default method for SLOT-UNBOUND.
1578 * minor incompatible change: prevent the user from specializing the
1579 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1580 somewhat counter to the intent of the protocol, I (CSR) think, and
1581 additionally it just doesn't work in SBCL as currently
1582 implemented, thanks to optimizations (that are always valid for
1583 the other three SLOT-VALUEish functions, but not for the setter).
1584 * minor incompatible change: the :SB-LDB feature is now enabled by
1585 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1586 the low-level debugger.
1587 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1588 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1589 down to an SB-SYS:FD-STREAM, allowing direct communication in
1590 more cases. Temporary files and pipes are still used as a fallback
1592 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1593 on the non-reentrant C functions localtime(3) and gmtime(3).
1594 * bug fix: global optimization policy was not visible in LOCALLY and
1596 * bug fix: class objects can be used as specializers in methods.
1597 (reported by Pascal Costanza)
1598 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1599 with a type error. (reported by blitz_ on #lisp)
1600 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1601 failed with a type error. (reported by Pascal Bourguignon)
1602 * bug fix: merging pathnames against defaults with :DIRECTORY
1603 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1604 (reported by James Y Knight)
1605 * bug fix: saving large (>2GB) cores on x86-64 now works
1606 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1607 constant index and a (simple-array (signed-byte 32)) array
1608 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1609 error (patch by Robert J. Macomber)
1610 * feature: TIME now displays the amount of run-time spent in GC
1611 * bug fix: The debugger now does a better job of respecting
1612 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1613 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1614 workaround for bug 403.)
1615 * bug fix: floating point exception handling now works on FreeBSD
1616 (thanks to NIIMI Satoshi)
1617 * fixed some bugs revealed by Paul Dietz' test suite:
1618 ** MISC.641: LET-conversion were not supposed to work in late
1620 * improvements to the Win32/x86 port:
1621 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1622 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1624 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1625 * new feature: source path information is generated for macro-expansion
1626 errors for use in IDE's like Slime (thanks to Helmut Eller)
1627 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1628 * bug fix: compiler does not loop forever on an invalid type in
1630 * improvement: compilation of most CLOS applications is significantly
1632 * optimization: added a limited bytecode compiler for simple toplevel
1633 forms, speeding up compilation and FASL loading
1634 * bug fix: the statistical profiler now properly distinguishes anonymous
1637 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1638 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1639 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1640 into subclasses of the finalized class. Now user calls to
1641 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1642 system return before any subclasses are finalized.
1643 * minor incompatible change: The reader no longer ignores errors
1644 regarding non-existent packages in #+ and #- feature tests.
1645 * new feature: command line options --no-sysinit, --no-userinit to
1646 inhibit loading the corresponding init files
1647 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1648 for sending data through UDP sockets (thanks to François-René Rideau)
1649 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1650 error is signalled instead (thanks to François-René Rideau)
1651 * minor incompatible change: Error signalling behaviour of lexical
1652 operations violating package locks has changed slightly. Refer to
1653 documentation on package locks for details.
1654 * enhancement: EVAL can process IF-expressions without resorting to the
1656 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1657 constant-folded properly due to missing out-of-line functions.
1658 (reported by Mika Pihlajamäki)
1659 * bug fix: LISTEN sometimes returned T even in cases where no data was
1660 immediately available from the stream
1661 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1662 were reversed. (reported by Levente Mészáros)
1663 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1664 when given a vector displaced to an adjustable array. (reported
1666 * bug fix: derivation of float boundaries from numbers outside the
1667 appropriate float range (reported by John Wiseman)
1668 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1670 * bug fix: precomputing cache entries for generic functions with
1671 some subclasses of specializers as yet invalid does not attempt to
1672 fill a cache line with a negative offset. (reported by Levente
1674 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1675 allow condition class objects as arguments to DOCUMENTATION and
1676 (SETF DOCUMENTATION); only find and set documentation for
1677 structure names for the STRUCTURE doc-type. (suggested by Gary
1679 * improvements to the Win32/x86 port:
1680 ** (user-homedir-pathname) and default initialization file
1681 locations now know about the user's "Documents and Settings"
1682 directory (thanks to Yaroslav Kavenchuk)
1683 ** run-program is implemented (thanks to Mike Thomas)
1684 ** sockets support (thanks to Timothy Ritchey)
1685 ** better backtrace support (thanks to Alastair Bridgewater)
1686 ** sb-grovel supported
1687 ** asdf-install and sb-posix work somewhat
1688 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1689 * improvements to the Solaris/x86 port:
1690 ** works on Solaris 11/Solaris Express
1691 ** floating-point exception handling support
1692 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1693 * fixed some bugs revealed by Paul Dietz' test suite:
1694 ** REMOVE-METHOD returns its generic function argument even when
1695 the method is not one of the generic functions' methods.
1696 ** objects declared as MEMBER types can be admissible arguments to
1697 structure accessors.
1698 ** printing characters should simply be printed by the FORMAT ~:C
1700 ** compiler failure when compiling functions with hairy constant
1701 defaults for optional parameters.
1702 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1703 function, which is already optimized.
1705 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1706 * new platform: experimental support for SBCL x86/Darwin, including
1707 MacOS X 10.4.5 on Intel.
1708 * new feature: Unicode character names are now known to the system
1709 (through CHAR-NAME and NAME-CHAR).
1710 * new feature: the filesystem location of SBCL's core file is
1711 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1712 * minor incompatible change: the contrib modules SB-POSIX and
1713 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1714 this change is to make it easier to distribute
1715 location-independent binaries.
1716 * bug fix: as implied by AMOP, standardized classes no longer have
1717 slots named by external symbols of public packages. (reported by
1719 * optimization: calling functions via constant symbols -- as in
1720 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1721 via the function object as in (FUNCALL #'FOO).
1722 * optimization: on x86 and x86-64, the calling convention for the
1723 case of dealing with an unknown number of values has been altered
1724 to be friendlier to the prediction heuristics implemented,
1725 particularly with reference to CALL and RET pairing. (thanks to
1726 Alastair Bridgewater)
1727 * optimization: on x86, the code for access to thread-local symbol
1728 values and binding/unbinding of thread-local symbols is smaller.
1729 (thanks to Alastair Bridgewater)
1730 * enhancement: CONSTANTP is now able to determine constantness of
1731 more complex forms, including calls to constant-foldable standardized
1732 functions and some special forms beyond QUOTE.
1733 * fixed bug: occasional GC crashes on Solaris/x86
1734 * optimization: x86-64 supports stack allocation of results of simple
1735 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1736 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1737 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1738 (thanks to James Knight)
1739 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1740 by the standard (thanks to Stephen Compall)
1742 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1743 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1744 be used for bundling the runtime and the core file into one
1745 executable binary. This feature is not currently supported on all SBCL
1746 platforms. (thanks to James Bielman and NIIMI Satoshi)
1747 * new feature: a generational or ephemeral garbage collector is now
1748 the default on the PowerPC platform (both Linux and Darwin). The
1749 old Cheney (stop and copy) collector is a build-time option.
1750 (thanks to Cyrus Harmon, after Raymond Toy)
1751 * minor incompatible change: the method by which SBCL finds its
1752 contributed modules has changed; it no longer relies on symbolic
1753 links from an $SBCL_HOME/systems directory, but searches directly
1754 in the subdirectories of $SBCL_HOME.
1755 * enhancement: the dynamic heap size on the Linux/PPC platform is
1756 markedly larger, even using the older Cheney garbage collector.
1757 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1758 certain complicated string types. (reported by Gary King)
1759 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1760 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1761 * fixed bug: DOCUMENTATION on structures no longer signals an
1762 error if no structure type of the right name exists. (reported by
1764 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1765 applicable non-standard primary or :AROUND method on
1766 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1767 no longer cause unbound variable errors. (reported by Kevin Reid)
1768 * fixed bug: metacircle resolution in cases where methods have slots
1769 added before the slots from STANDARD-METHOD. (reported by Jean
1771 * fixed bug: the Power PC assembly code for calling into lisp
1772 sometimes computed its return address wrongly (depending on the
1773 15th bit of the address, fixed at link time). This used to
1774 manifest itself in a segmentation violation while building PCL.
1775 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1776 many others over the years)
1777 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1778 the stack pointer (treated as a signed integer) is negative.
1779 (thanks to Peter van Eynde)
1781 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1782 * new platform: experimental support for the Windows operating
1783 system has been added. (thanks to Alastair Bridgewater)
1784 * fixed several bugs in and robustified the PPC FFI (including
1785 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1786 * bug fix: allow non-simple string symbol names (reported by Paul
1788 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1790 * bug fix: saving a core corrupted callbacks on x86/x86-64
1791 * bug fix: closed a loophole in metacircularity detection and
1792 grounding in the PCL implementation of CLOS.
1793 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1794 (reported by tomppa on #lisp)
1795 * bug fix: clear the direction flag on Lisp -> C transitions, as
1796 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1797 (reported by Andrej Grozin and Hendrik Maryns)
1798 * optimization: major improvements to GC efficiency on GENCGC platforms
1799 * optimization: faster implementation of EQUAL
1800 * optimization: emit more efficient opcodes for some common
1801 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1802 * optimization: several other minor code-generation improvements on x86-64
1803 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1805 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1806 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1807 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1808 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1809 by ANSI as a generic function, and for consistency (SETF
1810 GENERIC-FUNCTION-NAME) is treated likewise.
1811 * fixed bug #233.b: make constraint propagation notice when a variable
1812 value is changed after it is referenced but before it is used
1813 * fixed bug #296: no more arbitrary behaviour from filenames with
1814 odd characters as --load arguments. (reported by Adam Warner)
1815 * fixed bug #390: :CHARACTER-SET pathname components now work as
1816 expected. (reported by Tim Daly Jr)
1817 * fixed bug #391: complicated :TYPE intersections in slot
1818 definitions no longer cause an error in PCL internals.
1819 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1820 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1821 returning the number of octets which would be written to the
1822 file-stream. (thanks to Robert J. Macomber)
1823 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1824 arguments to the internal stream functions. (thanks to David
1826 * bug fix: stack exhaustion is now handled on the x86/SunOS
1827 platform. (thanks to Daisuke Homma)
1828 * bug fix: more accurate ROOM results on GENCGC platforms
1829 * optimization: improved type inference for arithmetic-for
1830 index variables in LOOP
1831 * optimization: faster floating-point SQRT on x86-64
1832 * fixed some bugs revealed by Paul Dietz' test suite:
1833 ** DOCUMENTATION returns NIL instead of "" for method combinations
1834 that don't have a docstring
1836 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1837 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1838 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1839 therefore are not customizeable by user code (as seems to be at
1840 least permitted and maybe required by AMOP). As a consolation,
1841 however, the SBCL implementation of these functions now calls
1842 REINITIALIZE-INSTANCE as specified by AMOP.
1843 * bug fix: slot-definition documentation is propagated as per ANSI
1844 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1845 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1846 argument, as do eslotd initialization methods. (from Pascal
1847 Costanza's "Closer" project)
1848 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1849 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1851 * bug fix: it is now possible to have more than one subclass of
1852 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1853 (reported by Bruno Haible, Pascal Costanza and others)
1854 * bug fix: the dependent update protocol now works for generic
1855 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1856 and Pascal Costanza)
1857 * bug fix: condition-class instances corresponding to
1858 DEFINE-CONDITION forms are now created eagerly. (reported by
1859 Kalle Olavi Niemitalo on comp.lang.lisp)
1860 * bug fix: floating point printing is more accurate in some
1861 circumstances. (thanks to Simon Alexander)
1862 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1863 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1864 * bug fix: callbacks on OS X now preserve stack-alignment.
1865 (thanks to Cyrus Harmon)
1866 * enhancement: the x86-64 disassembler is much better at
1867 disassembling SSE instructions. (thanks to Lutz Euler)
1868 * enhancement: CHECK-TYPE on a variable now allows type-inference
1869 (thanks to Vincent Arkesteijn)
1870 * optimization: improved performance of EUC-JP external format.
1871 (thanks to NIIMI Satoshi)
1872 * optimization: performance improvements to IO on file streams of
1873 :ELEMENT-TYPE CHARACTER
1874 * optimization: much faster memory allocation on x86-64
1875 * optimization: faster garbage collections (latency and throughput)
1877 * optimization: faster bignum subtraction and division on x86 and x86-64
1878 (thanks to Lutz Euler)
1879 * optimization: more accurate type-inference for loops that use a
1880 floating point index variable or a negative step.
1882 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1883 * bug fix: add a workaround to SBCL looping infinitely at startup on
1884 Linux kernels with apparently buggy implementations of personality().
1885 (thanks to Svein Ove Aas)
1886 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1887 (thanks to David Lichteblau)
1888 * bug fix: local bindings shadow global symbol macros for
1889 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1890 * optimization: non-open coded uses of numeric comparison operators
1891 (e.g. >) no longer cons when called with more than one parameter
1892 on platforms supporting dynamic-extent allocation.
1893 * enhancement: saving cores with foreign code loaded is now supported
1894 on MIPS/Linux in addition to the previously supported platforms.
1895 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1896 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1897 * bug fix: division by zero in sb-sprof when no samples were collected
1898 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1899 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1900 (thanks to Alexander Kjeldaas)
1901 * new feature: ignoring the return values of destructive functions
1902 like NREVERSE signals a compile-time style-warning.
1903 (thanks to Kevin Reid)
1904 * new feature: the alignment of alien structure fields can be
1905 explicitly specified. (thanks to Cyrus Harmon)
1906 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1907 and dump core on SIGQUIT
1909 ** incompatible change: threads do not inherit values of specials
1910 from their parents (see manual)
1911 ** bug fix: threads stacks belonging to dead threads are freed by the
1912 next exiting thread, no need to gc to collect thread stacks anymore
1913 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1914 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1915 ** bug fix: dynamic variable and thread start related gc lossage
1916 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1918 * fixed some bugs revealed by Paul Dietz' test suite:
1919 ** SUBTYPEP is slightly more accurate on heinously complicated
1920 CONS types where some of the members have uncertain (in the
1921 NIL, NIL sense) type relationships to each other.
1923 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1924 no longer purified unless :PURIFY T is explicitly specified.
1925 ** Non-purified cores are significantly smaller than before
1927 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1928 * new feature: timers based on Zach Beane's excellent timer package
1929 * added support for the following external formats: koi8-u,
1930 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1931 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1932 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1933 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1934 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1935 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1936 * incompatible change: a threaded SBCL will no longer revert to
1937 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1938 * bug fix: interrupts are disabled until startup is complete; no
1939 more sigsegvs when receiving a signal to soon
1940 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1942 * bug fix: add a workaround for the memory randomization features in
1943 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1944 workaround will only be in effect on systems with the proc filesystem
1946 * bug fix: printing objects of type HASH-TABLE signals a
1947 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1949 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1950 have been running for over 50 days. (reported by Gilbert Baumann)
1951 * bug fix: the logic for getting names of functions gets less
1952 confused when confronded with alternate-metaclass
1953 funcallable-instances. (reported by Cyrus Harmon)
1954 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1955 consistent, even on internal alternate-metaclass objects.
1956 * bug fix: flush closure information collected by physical
1957 environment analysis prepass before the main pass. (bug reported
1959 * bug fix: compiler pack phase does not modify a hashtable, which is
1960 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1961 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1962 may be expanded. (reported by Alexander <malishev> on c.l.l)
1963 * bug fix: no more style-warnings when compiling code using LAST
1964 under high-SPEED low-SPACE settings. (reported by David Wallin
1966 * fixed some incompatibilities between SBCL's MOP and the MOP
1968 ** the METAOBJECT class is now implemented;
1969 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1970 STANDARD-OBJECT, as required;
1971 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1972 now compatible (as required by VALIDATE-SUPERCLASS); there
1973 remains a consistency requirement over the presence or absence
1974 of the FUNCTION class in the superclass list of finalized
1975 classes; see the manual for more details;
1976 ** the :AROUND method for COMPUTE-SLOTS on
1977 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1978 requested slot ordering.
1980 ** bug fix: parent thread now can be gc'ed even with a live
1982 ** bug fix: binding a special with PROGV to no value is not
1983 the same as that symbol not having been bound (thanks to
1985 ** bug fix: binding specials is thread safe (thanks to
1987 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1988 and SB-SPROF do not die with 'no handler for signal XX in
1989 interrupt_handle_now(..)' anymore
1990 ** bug fix: WITH-TIMEOUT works with multiple running threads
1991 * fixed some bugs revealed by Paul Dietz' test suite:
1992 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1993 the :method-class keyword argument.
1995 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1996 * new port: the Solaris operating system on x86 processors is now
1997 mostly supported, though some rough edges in the environment
1998 remain. (thanks to Daisuke Homma)
1999 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
2000 space for its heap. (thanks to Thiemo Seufer)
2001 * optimization: SBCL on MIPS platforms now supports dynamic-extent
2002 closures. (thanks to Thiemo Seufer)
2003 * minor incompatible change: eof selects abort in the debugger.
2004 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
2005 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
2007 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
2008 deprecated, and will go away in a future revision of SBCL.
2009 * minor incompatible change: GC-ON and GC-OFF are no longer
2010 implemented with a counter, it does not matter how many times gc
2011 is switched on or off
2012 * bug fix: discriminating functions for generic function classes
2013 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
2014 make invalid assumptions about method precedence order. (reported
2016 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
2017 platforms. (thanks to Thiemo Seufer)
2018 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
2019 * bug fix: correct pathname printing with printer escaping is on.
2020 (thanks to Kevin Reid)
2021 * bug fix: complex VOP definitions in "user-space" no longer trigger
2022 package locks. (reported by Zach Beane)
2023 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
2024 for funcallable-instance functions, and is no different from
2026 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
2027 characters in the current readtable. (reported by Nicholas Neuss)
2028 * bug fix: SUBTYPEP on various CONS types returns more a more
2029 accurate acknowledgment of its certainty.
2030 * optimizations: REMOVE-DUPLICATES now runs in linear time on
2031 lists in some cases. This partially fixes bug 384.
2032 * flush all standard streams before prompting in the REPL and the
2034 * bug fix: signal handling and triggering gc do not conflict
2035 directly anymore, in particular a high frequency sb-sprof does
2036 not prevent gc from running
2037 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
2038 approximation for timezone and DST information between the
2039 universal time epoch and the smallest negative 32-bit time_t.
2040 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
2041 year 1899 so long as the encoded time is non-negative.
2042 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
2043 the correct answer for non-constant rotate amounts when performing
2044 an inline 32-bit rotation.
2046 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
2047 there is only one thread in the session
2048 ** bug fix: memory leak for streams created in one thread and
2049 written to in another
2050 ** bug fix: lockup when compiled with gcc4
2051 ** bug fix: race that allows the gc to be triggered when gc is
2053 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
2054 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
2056 ** bug fix: finished threads are gc'ed properly
2057 * fixed some bugs revealed by Paul Dietz' test suite:
2058 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
2059 has arguments with a different set of applicable methods from
2060 the orignal arguments.
2061 ** The type error thrown by MAP now has a correct expected-type
2063 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
2064 name a compiled function.
2065 ** Three MISC tests where a large bignum was improperly coerced to
2066 a float that couldn't represent that bignum during type
2067 derivation were fixed.
2068 ** SUBTYPEP can now handle types involving the negation of a
2069 list-form FUNCTION type.
2070 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
2071 as (COMPLEX (AND RATIO (NOT FIXNUM))).
2072 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
2074 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
2075 * New feature: Experimental support for bivalent streams: streams
2076 opened with :element-type :default now allow character and binary
2077 (unsigned-byte 8) I/O
2078 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
2079 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
2081 * bug fix: no more highly sporadic "couldn't check whether ~S is
2082 readable" when reading a stream and an interrupt hits in the middle
2083 of a select system call
2084 * compiler better recognizes complex arrays (reported by Eduardo
2086 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
2087 for variables on SBCL built with linkage-tables. (reported by Luis
2089 * various error reporting improvements.
2090 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
2091 (Thanks to Thiemo Seufer)
2092 * bug fix: nested reader invokations work correctly
2093 * bug fix: it is possible to have simultaneous references to foreign
2094 code and foreign data with the same name.
2096 ** added x86-64 support
2097 ** incompatible change: the threading api now works with thread
2098 objects instead of thread ids
2099 ** bug fix: threads are protected from signals and interruption when
2100 starting up or going down
2101 ** bug fix: a race where an exiting thread could lose its stack to gc
2102 ** bug fix: don't halt on infinite error in threads if possible
2103 ** fixed numerous gc deadlocks introduced in the pthread merge
2104 ** bug fix: fixed thread safety issues in read and print
2105 ** bug fix: debugger doesn't hang on session lock if interrupted at
2106 an inappropriate moment
2107 ** bug fix: run-program is now thread safe(r)
2108 ** bug fix: inner with-recursive-lock no longer releases the mutex
2109 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
2110 * fixed some bugs revealed by Paul Dietz' test suite:
2111 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
2112 EXPECTED-TYPE slots filled.
2113 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
2115 ** MAKE-CONDITION accepts classes as type-designators.
2116 ** COMPILE may never return NIL.
2117 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
2118 range before calling Unix time functions
2120 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
2121 * numerous signal handling fixes to increase stability
2122 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
2123 * minor incompatible change: we now correctly canonize default
2124 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
2125 as per the MOP, rather than the historical (INITARG INITFUNCTION
2126 INITFORM). (reported by Bruno Haible)
2127 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
2128 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
2129 for more information.
2130 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
2131 pathname is a directory pathname.
2132 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
2133 * dynamic space size on PPC has been increased to 768Mb. (thanks to
2135 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
2136 :METACLASS argument in addition to a class name. (reported by
2137 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
2138 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
2139 (reported by Bruno Haible; patch from Gerd Moellmann)
2140 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
2142 * bug fix: more cleanups to the floating point exception handling on
2143 x86-64 (thanks to James Knight)
2144 * bug fix: the compiler does not try to constant fold calls of
2145 COERCE and BIT-* functions when they return freshly constructed
2146 objects. (reported by Nikodemus Siivola and Paul Dietz)
2147 * optimization: improved the MIPS versions of generic (in the
2148 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
2149 * optimization: direct conversion of (unsigned-byte 32) to floats on
2150 the PowerPC platform.
2151 * optimization: structure instances with raw slots now use less
2152 memory, and probably show better memory locality. (thanks to
2154 * optimization: DYNAMIC-EXTENT declarations for lists and closures
2155 are treated as requests for stack allocation on the x86-64,
2156 Alpha-32, PPC and SPARC platforms.
2157 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
2158 the wrong file descriptor; implementation of SOCKET-OPEN-P.
2159 (thanks to Tony Martinez)
2161 ** gcing a dead thread can no longer lead to lockups
2162 ** threads block signals until they are set up properly
2163 ** errno is no longer shared by threads
2164 ** interrupt-thread restores the eflags register on x86
2165 ** fixed some lockups due to gc/thread interaction
2166 * fixed some bugs revealed by Paul Dietz' test suite:
2167 ** invalid dotted lists no longer raise a read error when
2168 *READ-SUPPRESS* is T
2169 ** signal an error if a symbol that names a declaration is used
2170 as the name of a type, or vice versa
2171 ** allow using the (declare (typespec var*)) abbreviation for
2172 (declare (type typespec var*)) with all type specifiers
2173 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
2174 ** always return NIL from PPRINT-POP when OBJECT is NIL
2175 ** don't signal errors when pretty-printing malformed LABELS,
2176 FLET or MACROLET forms
2177 ** declarations in a DOLIST body are also in scope for the
2179 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
2181 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
2184 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2185 * fixed cross-compiler leakages that prevented building a 32-bit
2186 target with a 64-bit host compiler.
2187 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
2188 opened with :IF-EXISTS :OVERWRITE.
2189 * fixed bug 281: error for an invalid qualifier in a short-form method
2190 combination method is not signalled until the faulty method is called.
2191 * bug fix: iteration variable type inferrer failed to deal with open
2192 intervals. (reported by Alan Shields)
2193 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
2194 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
2195 or more runtime options were provided to the sbcl binary.
2196 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
2197 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
2198 * optimization: there's now a fast-path for fixnum arguments in the
2199 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
2200 * optimization: the code generated on x86-64 is more compact thanks
2201 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
2202 * bug fix: floating-point exception handling now partly works on
2203 x86-64. (Thanks to James Knight)
2204 * improvement to the MIPS backend from Thiemo Seufer: C-style
2205 64-bit long long arguments and return values to alien functions
2207 * fixed some bugs revealed by Paul Dietz' test suite:
2208 ** the type-error signalled from WARN has a filled-in DATUM slot.
2209 ** the type-error required when a stream is not associated with
2210 a file has the stream as its datum.
2211 ** type-errors on single-floats on x86-64 no longer have
2212 :INVALID-OBJECT as the datum
2213 ** the type-errors signalled for invalid function names now have
2214 a correct expected type
2215 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2216 for structures defined with a :TYPE. Documentation strings for
2217 typed structures are no longer immediately discarded
2218 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2219 broadcast streams and synonym streams. FILE-LENGTH now also works
2220 on broadcast streams.
2222 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2223 * incompatible change: the --noprogrammer option, deprecated since
2224 version 0.7.5, has been removed. Please use the equivalent
2225 --disable-debugger option instead.
2226 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2228 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2229 inoperational for a while now) has been completely removed.
2230 * null lexical environments are now printed as #<NULL-LEXENV>,
2231 significantly reducing the amount of clutter in typical backtraces.
2232 * documentation on weak pointers, finalization, and after GC hooks
2233 has been added to the manual.
2234 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2235 as other specialized array types, is much faster. SUBSEQ and
2236 COPY-SEQ on such arrays have also been sped up.
2237 * optimization: EQL is now more efficient when at least other argument
2238 is known to be of type (OR FIXNUM (NOT NUMBER)).
2239 * fixed bug: the runtime is now less vulnerable to changes in the
2240 size of the SBCL object on OS X, and virtual memory is reserved for
2241 all spaces that need to be at a fixed address.
2242 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2243 * fixed bug: finalizers and after GC hooks that cause consing are now
2245 * fixed bug: compiler error messages and summaries are now printed to
2246 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2247 * fixed inference of the upper bound of an iteration variable.
2248 (reported by Rajat Datta).
2249 * fixed bug 211e: calling local functions with duplicated constant
2250 keyword argument no longer causes a bogus style warning about an
2252 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2253 ftype declaration to be disregarded. (reported by Dave Roberts)
2254 * fixed bug 373: caused by erronous compilation of references to alien
2255 variables in the runtime on ppc/darwin.
2256 * fixed bug 376: CONJUGATE type deriver.
2257 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2258 (reported by Baughn on #lisp)
2259 * a cross-compiler bug on non-x86 platforms has been identified and
2260 fixed. (thanks to Bruno Haible)
2261 * improvements to the MIPS runtime code for increased stability.
2262 (thanks to Thiemo Seufer)
2263 * increased the maximimum compact environment size to allow
2264 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2265 * partial workaround for bug 135: don't name the function we're
2266 calling for hairy cases of EVAL, so as not to accumulate one environment
2267 entry per call to EVAL. (reported by Kevin Reid)
2268 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2269 * optimization: type testing for non-vector arrays should be faster.
2270 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2272 * bug fix: setting 31st element of a bit vector to zero did not work
2274 * bug fix: redefining a class definition which failed due to a
2275 previous accessor / function clash now works (but see BUGS entry
2276 #380 for more problems in this area). (thanks to Zach Beane)
2277 * the long form of DEFINE-METHOD-COMBINATION disables method group
2278 checking when given a single method group with pattern *. (thanks
2280 * on x86 compiler supports stack allocation of results of simple
2281 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2282 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2283 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2284 * fixed some bugs related to Unicode integration:
2285 ** the restarts for recovering from input and output encoding
2286 errors only appear when there is in fact such an error to
2288 * fixed some bugs revealed by Paul Dietz' test suite:
2289 ** MISC.549 and similar: late transformation of unsafe type
2290 assertions into derived types caused unexpected code
2292 ** SCALE-FLOAT type deriver is less wrong.
2293 ** type derivers for EXP, LOG and similar functions compute result
2294 types for complex arguments better.
2295 ** (MISC.563) CONJUGATE type deriver works for very restricted
2297 ** out-of-line type testers for character strings are available.
2298 ** EQUAL compiler transform understands specialness of objects
2300 ** accessing double-floats stored on the stack now works on x86-64.
2301 ** debugger internals could sometimes create invalid lispobjs,
2302 resulting in GC crashes.
2303 ** MISC.548: type check weakening can convert required type into
2305 ** initialization forms for bindings are not in scope of free special
2308 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2309 * incompatible change: thread support for non-NPTL systems has
2310 been removed - locking is buggy and unreliable. A threaded
2311 SBCL build will now warn at startup and refuse to create
2312 new threads, unless futex support is detected in the kernel
2313 * incompatible change: the top level REPL now has only an ABORT
2314 restart associated with it, not TOPLEVEL and ABORT as it used to.
2315 TOP and TOPLEVEL are now available as debugger commands for
2316 returning to the top level.
2317 * incompatible change: forms evaluated in the REPL now use the
2318 global optimization policy.
2319 * incompatible change: user- and system-initialization files are
2320 no longer processed with LOAD, but by READ and EVAL; hence the
2321 global optimization policy, startup package, readtable, etc,
2323 * The .fasl file format number has been incremented because of
2324 various incompatible changes.
2325 * internal entry point details and argument counts no longer appear
2326 in backtraces unless explicitly requested by setting
2327 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2328 * built-in and standard functions no longer have names like "top
2329 level local call to FOO".
2330 * fixed bug 32: functions defined in non-null lexical environments
2331 now have more legible printed representation
2332 * fixed bug 33: functions defined in non-null lexical environemnts
2333 are now more amenable to inspection by INSPECT.
2334 * workaround for bug 354: XEPs no longer appear in backtraces unless
2335 explicitly requested.
2336 * fixed bug: receiving the signal which results from attempting to
2337 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2338 SIGBUS on other *BSDs) on architectures where the C stack is also
2339 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2340 to debugger confusion or wild instability. (reported by Cheuksan
2342 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2343 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2344 (reported by Lutz Euler)
2345 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2346 compiled code. (reported by Paul Dietz)
2347 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2348 specialized parameter to a method to something that is not TYPEP
2349 the specializer is now possible.
2350 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2351 face of package deletion.
2352 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2353 pathname. (reported by David Tolpin on c.l.l)
2354 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2355 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2356 STRING-STREAM and FILE-STREAM, respectively.
2357 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2358 than previously; in addition, its test suite is now run on build.
2359 (thanks to Luke Gorrie)
2360 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2361 * optimization: added a immediate representation for single-floats
2363 * fixed some bugs revealed by Paul Dietz' test suite:
2364 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2365 ** attempting to create a package with a colliding nickname causes
2366 correctable errors to be signalled.
2367 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2368 ** several x86-64 backend bugs related to sign-extension of immediate
2371 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2372 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2373 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2375 * fixed loading of multiply forward-referenced layouts.
2376 (thanks to Cheuksan Wang)
2377 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2378 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2379 (see COMPILE-FILE documentation for details of :PRINT option)
2380 * fixed bugs 19 and 317: fixed-format floating point printing is
2381 more accurate. This also fixes a bug reported by Adam Warner
2382 related to the ~@F format directive.
2383 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2385 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2386 dispatch table if the from-char is a dispatch macro character.
2387 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2388 by the wordsize no longer ignore the last word. (reported by Lutz
2390 * fixed bug in type checking of dynamic-extent variables. (reported
2392 * optimization: sequence traversal functions use their freedom to
2393 coerce function designators to functions.
2394 * optimization: code with many calls to CLOS methods specialized on
2395 CLOS classes has had redundant type checks removed.
2396 * optimization: added declarations to speed up operations that access
2397 the internal character database (for example STRING-UPCASE)
2398 * optimization: comparison operations between floats and sufficiently small
2399 fixnums no longer create extra rationals
2400 * fixed some bugs related to Unicode integration:
2401 ** portions of multibyte characters at the end of buffers for
2402 character-based file input are correctly transferred to the
2403 start of the buffer at the next read.
2404 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2405 passing it through to OPEN.
2406 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2407 argument given to internal calls to OPEN.
2408 ** the built-in comment readers (introduced by character sequences
2409 ";" and "#|") are more forgiving to encoding errors; they will
2410 STYLE-WARN and then attempt to resync the stream at a character
2411 boundary. (thanks to Teemu Kalvas)
2412 * fixed some bugs in the x86-64 port:
2413 ** Negative short int return values from c-calls are sign-extended
2415 ** The stack is aligned to 16-bytes for c-calls, as required by
2416 the ABI. (thanks to Cheuksan Wang)
2417 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2418 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2419 for enums. (thanks to Vincent Arkesteijn)
2420 ** Multiple small optimizations and bugfixes for floating point
2422 * fixed some bugs revealed by Paul Dietz' test suite:
2423 ** Space, Tab, Linefeed, Return and Page have the invalid
2424 secondary constituent character trait.
2425 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2427 ** WITH-INPUT-FROM-STRING should only update the index place on
2429 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2430 works more reliably.
2431 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2432 with one-dimensional arrays with fill pointers.
2433 ** TYPE-OF failed on a complex with an integer realpart and a
2435 ** compiler failure during type inference for the code of form
2436 (IF (EQL X (THE ...)) ...) (MISC.535).
2438 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2439 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2440 platform. The port supports 61-bit fixnums, large memory spaces
2441 and reloading shared object files.
2442 * enhancement: saving cores with foreign code loaded is now
2443 supported on ppc/Darwin in addition to the previously supported
2445 * enhancement: the statistical profiler now walks deeper into the
2446 call stack for more meaningful call-graphs and accrued time
2447 reports (x86/x86-64 only). It also now reports time spent in
2449 * enhancement: it is now possible to trace most individual methods
2450 of a generic function in addition to tracing the generic function
2452 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2454 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2455 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2456 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2457 produces an error. (thanks to Vincent Arkesteijn)
2458 * bug fix: NAMESTRING on pathnames with :WILD components in their
2459 directories works correctly. (thanks to Artem V. Andreev)
2460 * fixed bug 125: compiler preserves identity of closures. (reported
2462 * bug fixed: functions with &REST arguments sometimes failed with
2463 "Undefined function" when compiled with (DEBUG 3). (reported by
2465 * bug fix: overflow during compiling of setting element of a bit
2466 vector with constant index and value. (reported by Timmy Douglas)
2467 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2468 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2469 * on x86 compiler partially supports stack allocation of dynamic-extent
2471 * GO and RETURN-FROM do not check the extent of their exit points
2472 when compiled with SAFETY 0.
2473 * fixed some bugs related to Unicode integration:
2474 ** encoding and decoding errors are now much more robustly
2475 handled; it should now be possible to recover even from invalid
2476 input or output to the terminal. (thanks to Teemu Kalvas)
2477 ** provided a first cut at implementing STRING-TO-OCTETS and
2478 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2479 ** altered the SB-MD5 contributed module slightly, changing the
2480 interface just enough for it to be supportable for builds where
2481 lisp characters are not eight bits.
2482 * fixed some bugs revealed by Paul Dietz' test suite:
2483 ** the FORMATTER-generated functions for ~V[ conditionals require
2484 the correct number of arguments.
2485 ** READ-FROM-STRING returns the mandated second value when applied
2486 to displaced strings.
2487 ** the #\Rubout and #\Backspace characters are treated as invalid
2488 constituent characters by the tokenizer.
2490 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2491 * new feature: reloading changed shared object files with
2492 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2493 * new feature: references to foreign variables and functions
2494 can now be compiled and loaded before the corresponding shared
2495 object file is loaded, as long as the foreign definitions are
2496 available at runtime.
2497 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2498 supported. (thanks to Dan Debertin)
2499 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2500 just keywords. (thanks to Vincent Arkesteijn)
2501 * fixed bug #331: structure-class instances corresponding to
2502 DEFSTRUCT forms are now created eagerly.
2503 * fixed bug #345: backtraces from calls to undefined functions work
2504 on x86 as well. Related bug #61 is now also partially fixed on x86
2505 (backtraces from throws to unknown catch tags.)
2506 * bug fix: lambda-list parsing is now stricter vrt. order and number
2507 of lambda-list keywords.
2508 * bug fix: as specified by AMOP, an error is signalled if a
2509 class-option appears multiple times in a DEFCLASS form. (reported
2511 * bug fix: RANDOM can be compiled when the compiler derives the type
2512 of its numeric argument as a disjoint set of small integers.
2513 (reported by Paul Dietz)
2514 * bug fix: starting a core saved with shared objects loaded when
2515 those objects are not available no longer causes threaded SBCL to
2516 hang. (reported by Sean Ross)
2517 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2519 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2520 argument, which must always be NIL. (reported by Kalle Niemitalo)
2521 * bug fix: printing 1.0d+23 no longer results in an error.
2522 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2523 * bug fix: structure slot setters preserve evaluation order. (thanks
2525 * fixed some bugs related to Unicode integration:
2526 ** RUN-PROGRAM can allow its child to take input from a Lisp
2527 stream. (reported by Stefan Scholl)
2528 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2529 locales. (reported by Ken Causey)
2530 * fixed some bugs revealed by Paul Dietz' test suite:
2531 ** INCF, DECF and REMF evaluate their place form as specified in
2533 ** FORMATTER expands ~{ ~} iteration directives with V or #
2534 parameters correctly.
2535 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2536 consequent uses no arguments correctly.
2537 ** the system has a partial understanding of the (COMPLEX RATIO)
2539 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2542 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2543 * new feature: a build-time option (controlled by the :SB-UNICODE
2544 keyword feature, enabled by default) for building the system with
2545 support for the entire 21-bit character space defined by the
2547 * new feature: the system now has rudimentary external-format
2548 support; the primary user-visible change at this time is that
2549 characters with the high bit set (such as lower-case-e-acute) will
2550 print correctly to a terminal in a UTF-8 environment.
2551 * minor incompatible change: BASE-CHAR no longer names a class;
2552 however, CHARACTER continues to do so, as required by ANSI.
2553 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2554 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2555 has been moved to the SB-EXT package (temporarily re-exported from
2557 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2558 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2560 * fixed bug #351: better error-handlind and reporting for malformed
2562 * fixed bug #350: bignum-printing is now more memory-efficient,
2563 allowing printing of very large bignums, eg. (expt 2 10000000).
2564 (reported by Bruno Haible)
2565 * fixed bug #302: better primitive-type selection for intersection
2567 * fixed bug #308: non-graphic characters now all have names, as
2568 required. (reported by Bruno Haible)
2569 * bug fix: redefining a class with different superclasses now correctly
2570 removes it from the direct-subclasses of its previous superclasses.
2571 (reported by David Morse)
2572 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2573 new value now works. (reported by Bruno Haible)
2574 * bug fix: correct canonicalization of multiple non-standard slot
2575 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2576 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2577 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2578 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2579 now exists, an signals an error.
2580 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2581 during macroexpansion for non-iteration variables as well. (reported
2582 by Bruno Haible for CMUCL)
2583 * bug fix: Cyclic structures and unprintable objects in compiler
2584 messages no longer cause errors. (reported by Bruno Haible)
2585 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2586 and READ-FROM-STRING all now return a primary value of NIL if
2587 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2588 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2589 (reported by Bruno Haible for CMUCL)
2590 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2591 do not designate a condition. (reported by Bruno Haible for
2593 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2594 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2595 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2596 permitting resolution in favour of any of the conflicting symbols.
2597 (reported by Bruno Haible for CMUCL)
2598 * bug fix: EQUAL compiler optimizations is less aggressive on
2599 strings which can potentially compare true despite having distinct
2600 specialized array element types.
2601 * bug fix: unit enumerations can be defined without dividing by
2602 zero. (thanks to Vincent Arkesteijn)
2603 * FORMAT compile-time argument count checking has been enhanced.
2604 (report from Bruno Haible for CMUCL)
2605 * a partial workaround for the bug 262: the compiler does not try to
2606 inline-expand a local function doing RETURN-FROM from a deleted
2607 BLOCK. (thanks to Peter Denno for the bug report and to David
2608 Wragg for the simple test case)
2609 * fixed some bugs revealed by Paul Dietz' test suite:
2610 ** RENAME-PACKAGE allows all package designators as new package
2612 ** constraint propagation and lambda variable substitution are
2613 more cautious in dealing with partially deleted code.
2614 ** compiler sometimes lost reoptimization passes.
2615 ** CERROR, given a condition as condition designator, passes its
2616 remaining arguments to the continue format control without
2618 ** Case-altering FORMAT directives work correctly on non-ASCII
2620 ** The REAL type specifier handles bounds outside the floating
2621 point ranges without signalling FLOATING-POINT-OVERFLOW.
2622 ** Functions with IR1-transformations can create intercomponent
2623 references to global functions.
2624 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2626 ** FORMAT ~R treats a nil value for its first parameter correctly.
2628 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2629 * enhancement: saving cores with foreign code loaded is now
2630 supported on x86/NetBSD and sparc/Linux in addition to the previously
2631 supported platforms.
2632 * bug fix: on some platforms repeated installations caused multiple
2633 copies of HTML documentation to be installed -- should not happen
2634 any more. (reported by Stefan Scholl)
2635 * bug fix: parsing self-recursive alien record types multiple times
2636 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2637 original patch by Helmut Eller for CMUCL)
2638 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2639 (thanks to Richard Kreuter)
2640 * bug fix: defining classes whose accessors are methods on existing
2641 generic functions in other (locked) packages no longer signals
2642 bogus package lock violations. (reported by François-René Rideau)
2643 * bug fix: special variables as DEFMETHOD parameters no longer have
2644 associated bogus type declarations. (reported by David Wragg and
2646 * bug fix: read-write consistency on streams of element-type
2647 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2648 * bug fix: redefiniton of the only method of a generic function with
2649 no DEFGENERIC no longer emits a full WARNING. In addition,
2650 redefinition of generic functions with no DEFGENERIC to an
2651 incompatible lambda list now signals an error. (thanks to Zach
2653 * bug fix: DEFGENERIC now works even when there's a function of the
2654 same name in an enclosing lexical environment. (thanks to Zach
2656 * fixed compiler failure, caused by instrumenting code during
2657 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2658 * optimization: added loop analysis and improved register allocation
2659 * fixed some bugs revealed by Paul Dietz' test suite:
2660 ** POSITION on displaced vectors with non-zero displacement
2661 returns the right answer.
2662 ** (SIMPLE-STRING) is a valid type specifier for sequence
2664 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2666 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2667 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2669 ** The justification version of the FORMAT ~< directive treats
2670 non-zero minpad parameter correctly.
2672 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2673 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2674 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2675 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2676 the supported interface.
2677 * new feature: Single-stepping of code compiled with DEBUG 2 or
2678 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2679 * new feature: saving cores with foreign code loaded is now
2680 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2681 Timothy Moore's work for CMUCL)
2682 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2683 parameters to * instead of NIL if no initform is supplied.
2684 (reported by Johan Bockgaard)
2685 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2686 correctly when FUNCALL appears as the car of the form. Note:
2687 despite this FUNCALL forms are not currently subject to
2688 compiler-macro expansion. (port of Raymond Toy's fix for the
2689 same from CMUCL, reported by Johan Bockgaard)
2690 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2691 (thanks for Teemu Kalvas)
2692 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2693 case instead of terminating the iteration (thanks for Julian
2694 Squires, Sean Champ and Raymond Toy)
2695 * bug fix: incorrect expansion of defgeneric that caused a style
2696 warning. (thanks for Zach Beane)
2697 * bug fix: slot accessor effective method computation works properly
2698 for classes with multiple non-standard applicable methods on
2699 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2700 * on x86 compiler supports stack allocation of results of LIST and
2701 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2702 CMUCL implementation by Gerd Moellmann)
2703 * fixed some bugs revealed by Paul Dietz' test suite:
2704 ** FORMAT strings with both the ~<~:;~> form of the justification
2705 directive and pretty-printing directives cause an error.
2707 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2708 * incompatible change: the internal functions
2709 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2710 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2711 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2712 instead of the old functions.
2713 * new feature: on platforms where "dladdr" is available foreign
2714 function names now appear in backtraces. (based on Helmut Eller's
2716 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2717 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2719 * documentation: ASDF manual is now distributed with SBCL.
2720 * bug fix: SBCL can now load its contributed modules with REQUIRE
2721 even if the system-provided entries have been removed from
2722 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2724 * bug fix: backtraces involving undefined functions or assembly
2725 routines are more informative. (thanks to Brian Downing)
2726 * bug fix: mutually referent alien structures now work correctly.
2727 (reported by Rick Taube)
2728 * bug fix: structures defined by WITH-ALIEN can be referred to
2729 within other definitions in the same WITH-ALIEN.
2730 * bug fix: division operators (MOD, TRUNCATE and the like) with
2731 constant zero divisors and integer dividends no longer generate
2733 * bug fix: provide default methods for INPUT-STREAM-P and
2734 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2735 * bug fix: improve the detection and resolution of MOP metacycles
2736 (where effective method computation depends on the generic function
2737 having its effective method computed). (reported by Bruno Haible)
2738 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2739 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2740 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2741 * bug fix: pathnames with relative directory components can be
2742 represented relative to default pathnames.
2743 * optimization: in taking the GCD of bignums, reduce the two bignums
2744 to approximately the same size (using Euclid's algorithm) before
2745 applying the more sophisticated binary GCD. (thanks to Juho
2747 * optimization: COUNT on bitvectors now operates word-at-a-time.
2748 * optimization: ASH with a positive, but not necessarily constant,
2749 (leftwards) shift, when in a modular context, is compiled to a
2751 * fixed some bugs revealed by Paul Dietz' test suite:
2752 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2753 if the corresponding argument is NIL.
2754 ** FORMAT directives accepting a minpad parameter treat negative
2756 ** literal commas as character parameters to FORMAT directives are
2758 ** literal spaces directly after ~<Newline> directives within a
2759 format-logical-block (~:< ~@:>) do not induce :FILL-style
2760 conditional newlines.
2761 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2762 as specified: it no longer includes conditional newlines).
2763 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2765 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2766 :PER-LINE-PREFIX argument does not evaluate to a string.
2768 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2769 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2770 the manual for details; for now, package locks can be disabled by
2771 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2772 if no major problems are found then it is likely that they will be
2773 compiled in unconditionally.
2774 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2775 unsupported operators on all platforms. To load a shared library
2776 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2777 object file, link it into a shared library outside of SBCL and
2778 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2780 * bug workaround: ROOM T and the GC were not working together
2781 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2782 reporting facility aren't preserved. That reporting has been
2783 disabled, so now until and unless someone figures out how to make it
2784 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2785 * minor incompatible change: as threatened around sbcl-0.8.0, the
2786 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2787 The symbols are also exported from SB-PCL for backwards
2788 compatibility, but more so than before SB-PCL should be treated as
2789 an implementation-internal package.
2790 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2792 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2793 It is known as of this release not to work on the Alpha, however.
2794 * fixed bug #167: errors signalled due to illegal syntax in method
2795 bodies are now more legible.
2796 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2797 designators and can hence be used with TYPEP.
2798 * fixed bug #333: CHECK-TYPE now ensures that the type error
2799 signalled, if any, has the right object to be accessed by
2800 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2801 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2802 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2804 * fixed bug #334: programmatic addition of slots using specialized
2805 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2806 and :ALLOCATION :CLASS slots.
2807 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2808 float exponents. (rereported by Peter Seibel)
2809 * fixed a bug: #\Space (and other whitespace characters) are no
2810 longer considered to be macro characters in standard syntax by
2811 GET-MACRO-CHARACTER.
2812 * fixed bug: initialization of condition class metaobjects no longer
2813 causes an instance of the condition to be created. (reported by
2815 * fixed bug: it is now possible to have slots such that
2816 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2817 description is neither :INSTANCE nor :CLASS.
2818 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2819 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2820 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2821 non-standard SLOT-DEFINITION-ALLOCATION values.
2822 * fixed bugs: various race conditions handling exiting threads.
2823 CL-PPCRE's thread torture test now passes.
2824 * fixed arguably-a-bug: GC time proportional to number of threads in
2825 system even when most of them are idle
2826 * optimization: improved performance of BIT and SBIT on bit-vectors.
2827 * .fasl file incompatibility: The fasl file version number has
2828 been incremented because of changes associated with package locks.
2830 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2831 * minor incompatible change: the system no longer provides
2832 optimization hints (conditions which are TYPEP
2833 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2834 mode; these hints are still emitted when higher SPEED optimization
2836 * new contrib module: a sampling profiler (profiling by statistical
2837 sampling, rather than by instrumenting functions) is available as
2838 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2839 * the behaviour of the standard function ED is now customizeable by
2840 third parties through a hook variable: see ED's documentation
2841 string for information on the protocol.
2842 * the compiler no longer emits efficiency notes for (FUNCALL X)
2843 when the type of X is uncertain under default optimization
2845 * fixed bug 276: mutating a binding of a specialized parameter to a
2846 method to something that is not TYPEP the specializer is now
2848 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2849 exhibit the required behaviour on the x86 platform. (thanks to
2850 Peter van Eynde, Eric Marsden and Bruno Haible)
2851 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2852 even for difficult arguments. (reported by Peter Graves)
2853 * fixed bug 141a: the backquote printer now descends quoted
2855 * fixed another bug in backquote printing: no more destructive
2856 modification of the form's list structure. (reported by Brian
2858 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2859 move between its address being taken and the call to
2860 interrupt_thread, fixing a crashing race condition.
2861 * the SB-POSIX contrib implementation has been adjusted so that it
2862 no longer exhibits ridiculously poor performance when constructing
2863 instances corresponding to C structs.
2865 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2866 * minor incompatible change: the sb-grovel contrib now treats C
2867 structures as alien (in the sense of SB-ALIEN) objects rather than
2868 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2869 has implications for memory management of client code
2870 (sb-grovel-returned objects must now be manually managed) and for
2871 type safety (alien objects now have full types).
2872 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2873 used to control emission of compiler diagnostics, rather than the
2874 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2875 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2876 quality should be considered deprecated.
2877 * install.sh now installs the user manual as well
2878 * (not quite a new documentable feature, but worth considering in
2879 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2880 beginnings of a semantically meaningful condition hierarchy is
2881 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2883 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2884 designator as the defaults argument.
2885 * fixed bug: Displaced arrays whose displaced-to array has become
2886 too small now cause ARRAY-DIMENSION to signal an error, providing
2887 for safer bounds-checking. (reported by Bruno Haible)
2888 * fixed bug: DEFCLASS slot definitions with identical :READER and
2889 :WRITER names now signal a reasonable error. (reported by Thomas
2891 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2893 * fixed bug: Invalid slot specification errors now print correctly.
2894 (thanks to Zach Beane)
2895 * fixed bug 320: Shared to local slot value transfers in class
2896 redefinitions now happen corrently from superclasses as
2897 well. (reported by Bruno Haible)
2898 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2900 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2901 improper lists correctly. (reported by Bruno Haible)
2902 * fixed bug 313: source-transform for <fun-name> was erroneously
2903 applied to a call of a value of a variable with name <fun-name>.
2904 (reported by Antonio Menezes Leitao)
2905 * fixed bug 307: The obsolete instance protocol ensures that
2906 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2907 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2908 remove dead unknown-values globs from the middle of the stack.
2909 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2910 easier to resume long computations after using *BREAK-ON-SIGNALS*
2911 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2912 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2913 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2914 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2915 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2916 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2918 * fixed a (fairly theoretical) bug in string printing: if
2919 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2920 does not have array-element-type equal to the most general string
2922 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2923 function name. (thanks to Nikodemus Siivola)
2924 * fixed bug: compile-time format string checker failed on
2925 non-closed ~{. (reported by Thomas F Burdick)
2926 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2927 don't warn on legal loop code involving more than one aggregate
2928 boolean. (thanks to Nikodemus Siivola)
2929 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2930 in keyword position to MAKE-PACKAGE was not being reported as
2932 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2933 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2935 * fixed bugs 280 and 312: the checking for multiple definitions in a
2936 file is less likely to become confused by uses of inline
2938 * fixed bug: the #S reader macro performs the keyword coercion
2939 specified for slot names. (reported by Kalle Niemitalo)
2940 * fixed bug: lambda lists may contain symbols whose names start with
2941 & but are not lambda-list-keywords; their occurrence triggers a
2943 * fixed bug 321: define-method-combination argument lambda lists do
2944 not cause invalid code to be generated when &optional or &aux
2945 variables have default values. (reported by Bruno Haible)
2946 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2947 have CLOS classes; this appears also to have fixed reported
2948 instability in stack exhaustion detection.
2949 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2950 MUFFLE-WARNING when no associated restart is present is now
2952 * optimization: rearranged the expansion of various defining macros
2953 so that each expands into only one top-level form in a
2954 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2956 * optimization: used a previously little-used slot in symbols to
2957 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2959 * fixed some bugs revealed by Paul Dietz' test suite:
2960 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2962 ** Fixed an optimization bug related to inheritance of initforms
2963 from local to shared slots.
2964 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2965 ** CAST optimizer forgot to flush argument derived type.
2966 ** print/read consistency on floats is now orders of magnitude
2967 more likely. (thanks also to Bruno Haible for a similar report
2969 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2970 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2971 for objects of type REAL. Make it so.
2972 ** SXHASH is specified (infelicitously) to respect similarity,
2973 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2974 it so. (thanks to Markus Ziegler)
2975 ** on the Alpha, the compiler succeeds in compiling functions
2976 returning a known number of arguments greater than 63.
2977 ** fixed handling of invalid NIL arguments in keyword position
2979 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2981 ** provided a readably-printable representation for RANDOM-STATE
2983 ** ensured that pathnames, where they have a namestring, always
2984 print using #P"..." syntax.
2986 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2987 * Support for the forthcoming 2.0 version of the NetBSD kernel
2988 running on x86 hardware has been added. (thanks to Perry
2989 E. Metzger most immediately, and others for their past work)
2990 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2991 current release version; SBCL's OpenBSD support had been broken
2992 since about the time of OpenBSD's switch to ELF binary format.)
2993 * [placeholder for DX summary]
2994 ** user code with &REST lists declared dynamic-extent, under high
2995 speed or space and low safety and debug optimization policy.
2996 * The manual has been converted to Texinfo format and the debugger
2997 chapter from the cmucl manual has been added.
2998 * A facility has been added to extract documentation strings from
2999 sbcl and store them as Texinfo-formatted snippets for inclusion in
3000 the manual (via Texinfo's @include directive)
3001 * bug fix: compiler emitted division in optimized DEREF. (thanks for
3002 the test case to Dave Roberts)
3003 * bug fix: multidimensional simple arrays loaded from FASLs had fill
3004 pointers. (reported by Sean Ross)
3005 * bug fix: PROFILE output is printed nicely even for large numerical
3006 values. (thanks to Zach Beane)
3007 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
3008 greater than 32 handle EOF correctly.
3009 * bug fix: on X86 an immediate argument of the IMUL instruction is
3010 correctly printed in disassembly. (thanks to Lutz Euler)
3011 * bug fix: class slots in redefined classes preserve their old
3012 values. (thanks to Bruno Haible and Nikodemus Siivola)
3013 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
3014 succeeds. (reported by Marco Baringer)
3015 * fixed some bugs revealed by Paul Dietz' test suite:
3016 ** READ-SEQUENCE now works on ECHO-STREAMs.
3017 ** RATIONALIZE works more according to its specification. (thanks
3020 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
3021 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
3022 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
3023 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
3024 should matter to you only if you rebind the printer control
3025 variables and then find you want different bindings in the
3026 debugger than in the ordinary execution of your program.)
3027 * The runtime build system has been tweaked to support building
3028 (on SPARC/SunOS) using a C compiler which invokes Sun's own
3029 assembler and linker. (thanks to Nikodemus Siivola)
3030 * Unbound, undefined, undeclared variables now trigger full
3031 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
3032 more often programmer error than deliberate exploitation of undefined
3034 * optimization: the hash algorithm for strings has changed to one
3035 that is less vulnerable to spurious collisions. (thanks to Juho
3037 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
3038 needless bounds checking. (thanks to Juho Snellman)
3039 * optimization: implemented multiplication as a modular
3040 (UNSIGNED-BYTE 32) operation on the PPC backend.
3041 * fixed some bugs revealed by Paul Dietz' test suite:
3042 ** ADJUST-ARRAY now copies the datum in a zero rank array if
3044 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
3045 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
3046 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
3048 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
3050 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
3052 ** ECHO-STREAMs no longer attempt to echo the end of file value to
3053 their output stream on EOF from read.
3054 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
3055 have been read to end-of-file.
3056 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
3058 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
3059 description of determination of which consecutive characters
3061 ** Printing the "Space" character with escaping on now yields "#\\ ",
3062 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
3063 ** Reading floating-point numbers with *READ-BASE* set to a number
3064 less than 10 works correctly.
3065 ** Reading floating-point numbers with *READ-BASE* set to a number
3066 more than 10 works correctly.
3067 ** Printing with *PRINT-READABLY* targets the standard readtable, not
3068 the readtable currently in effect.
3070 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
3071 * minor incompatible change: parsing of namestrings on a physical
3072 (Unix) host has changed; numbers after the final #\. in a
3073 namestring are no longer interpreted as a version field. This is
3074 intented to be largely invisible to the user, except that the
3075 meaning of the namestring "*.*.*" has changed: it now refers to a
3076 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
3077 should usually be replaced by
3078 (make-pathname :name :wild :type :wild :version :wild)
3079 with the added benefit that this is more likely to be portable.
3080 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
3081 option to OPEN now signals an error if the file being opened
3082 exists; this may have an impact on existing code.
3083 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
3084 signals received, on the PowerPC platforms (both Linux and
3085 Darwin). (thanks to Pierre Mai for pointing out the location of
3087 * several fixes on OS X: The system now builds and runs cleanly on
3088 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
3089 (thanks to Brian Mastenbrook)
3090 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
3091 second-resolution: integer multiples of 1/3600 between -24 and 24.
3092 (thanks to Vincent Arkesteijn)
3093 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
3094 type when called with 1 argument; PEEK-CHAR checked type of
3095 PEEK-TYPE only after having read first character from a
3096 stream. (reported by Peter Graves)
3097 * bug fix: the garbage collector now has much better locality
3098 behaviour, and in particular no longer treats all memory as being
3099 exhausted when many small objects point to each other in a deeply
3101 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
3102 recognized as being TYPEP their class.
3103 * bug fix: the PUSHNEW documentation string has been corrected.
3104 (thanks to Vincent Arkesteijn)
3105 * bug fix: defaulting of the value for the last of an atypically
3106 large number of multiple values being bound was not being
3107 performed correctly on the Alpha or PPC platforms
3108 * optimization: implemented multiplication as a modular
3109 (UNSIGNED-BYTE 32) operation on the x86 backend.
3110 * optimization: SEARCH on simple-base-strings can now be open-coded.
3111 (see also contrib/compiler-extras.lisp for inspiration for
3112 teaching the compiler about the Boyer-Moore algorithm).
3113 * value, returned by MAX (and MIN) called with several EQUALP, but
3114 not EQL, arguments now does not depend on compiler settings.
3115 * fixed some bugs revealed by Paul Dietz' test suite:
3116 ** in stack analysis liveness information is propagated from
3117 non-local entry points.
3118 ** pathwise CAST removing failed when the CAST node did not start
3120 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
3121 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
3123 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
3124 host is already defined.
3125 ** RENAME-FILE works on streams instead of signalling an internal
3127 ** PEEK-CHAR uses the current readtable when determining whether
3128 or not a character is whitespace.
3129 ** MERGE-PATHNAMES handles the case when the pathname does not
3130 specify a name while the default-pathname specifies a version.
3131 ** Pathnames now stand a better chance of respecting print/read
3133 ** Attempting to use standardized file system operators with a
3134 pathname with invalid :DIRECTORY components signals a
3136 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
3138 ** DIRECTORY on logical pathnames is more correct.
3139 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
3140 signal a TYPE-ERROR if their argument is not a stream.
3141 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
3142 designator argument does not designate a stream.
3143 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
3144 examining the synonym.
3145 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
3147 ** OPEN and WITH-OPEN-STREAM allow opening streams with
3148 element-type larger than ([UN]SIGNED-BYTE 32).
3150 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
3151 * When built with the :SB-FUTEX feature, threaded builds now take
3152 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
3153 for faster/more reliable mutex and condition variable support.
3154 * Incompatible change (but one you probably shouldn't have been using
3155 anyway): the interface and code for arbitrating between multiple
3156 threads in the same user session has been redesigned.
3157 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
3158 error when called without an explicit environment argument.
3159 (thanks to Nikodemus Siivola)
3160 * bug fix: buffered :DIRECTION :IO streams are less likely to become
3161 confused about their position. (thanks to Adam Warner and Gerd
3163 * bug fix: Pretty printing backquoted forms with unquotations in the
3164 argument list position of various code constructs such as LAMBDA
3165 now works correctly. (reported by Paul Dietz)
3166 * bug fix: Pretty printing unquotations no longer loses all
3167 stream position information.
3168 * optimization: performance of string output streams is now less
3169 poor for multiple small sequence writes.
3170 * optimization: performance of CSUBTYPEP in the presence of complex
3171 expressions involving CONS and NOT many times has been improved.
3172 (reported by Paul Dietz)
3173 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
3175 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
3176 streams for the REPL, for the debugger, and so on. (thanks to
3178 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
3179 * fixed some bugs revealed by Paul Dietz' test suite:
3180 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
3181 is ignored if it is too small, rather than propagating through
3182 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
3183 ** extremely complex negations of CONS types were not being
3184 sufficiently canonicalized, leading to inconsistencies in
3186 ** VALUES tranformer lost derived type.
3188 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
3189 * fixed a bootstrapping bug: the build process no longer assumes
3190 that the various BOOLE-related constants have the same value in
3191 host and target lisps. (noted by Paul Dietz' test suite on an
3192 SBCL binary built from CLISP)
3193 * The system can now be dynamically linked on the MIPS platform,
3194 which enables dynamic loading of foreign code from Lisp. (thanks
3195 to Ralf Baechle for discussions on the MIPS ABI)
3196 * The system now records debugging information for its own source
3197 files in a filesystem-position-independent manner, relative to
3198 the "SYS" logical host.
3199 * fixed a compiler bug: MV-LET convertion did not check references
3200 to the "max args" entry point. (reported by Brian Downing)
3201 * tweaked disassembly notes to be less confident about proclaiming
3202 some instruction as an LRA. (thanks to Brian Downing)
3203 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
3204 now each have their own history, command character, and other
3205 characteristics. (thanks to David Lichteblau)
3206 * fixed some bugs revealed by Paul Dietz' test suite:
3207 ** compiler failure in compiling LOGAND expressions including a
3209 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3210 shift greater than 32.
3211 ** FLUSH-DEST did not mark blocks for type check regeneration.
3212 ** HANDLER-CASE failed to accept declarations in handler clauses
3213 in some circumstances.
3215 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3216 * New code in contrib/sb-introspect (still probably not entirely
3217 stable yet) provides some support for smart Lisp development
3218 environments like SLIME.
3219 * The conditions signalled for errors occurring when loading .fasl
3220 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3221 in a way which should help ASDF recover gracefully.
3222 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3223 stuff has been cleaned up. If you code contrib/ stuff, this might
3224 affect you, and you can look at contrib/README, contrib/STANDARDS,
3225 and/or the 0.8.4.27 diff to check.
3226 * In full calls the compiler now does not generate checks for declared
3227 argument types for all arguments.
3228 * various threading fixes
3229 ** and some experimental patches which didn't make it into
3230 the main tree for this release, but which are shipped in
3231 contrib/experimental-thread.patch as a possible fix for some
3232 failures (deadlock, spinning...) in GC-intensive multithreaded
3234 * fixed PPC build problem (source code incompatibility of different
3235 library versions): added offsetof() hackery which attempts to divine
3236 where glibc maintainers put uc_mcontext today
3237 * fixed bug 282: compiler does not trust type assertions while passing
3238 arguments to a full call.
3239 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3240 &OPTIONAL VALUES type parameter.
3241 * fix bug 214: algorithm for noting rejected templates is now more
3242 similar to that of template seletion. (also reported by rydis on
3244 * fixed bug 141b: printing backquoted information readably and prettily
3245 inserts a space where necessary.
3246 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3247 (CHARACTER 10) are now reported as errors, rather than propagated
3248 as unknown types. (reported by piso on #lisp)
3249 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3250 with values NIL and :ERROR. (thanks to Milan Zamazal)
3251 * fixed bug 191c: CLOS now does proper keyword argument checking as
3252 described in CLHS 7.6.5 and 7.6.5.1.
3253 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3254 longer raise an error; further, using a list as a for-as-arithmetic
3255 counter now raises a meaningful error.
3256 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3257 now understood by sequence creation functions such as MAKE-SEQUENCE
3259 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3260 a STRING-INPUT-STREAM.
3261 * compiler enhancement: SIGNUM is now better able to derive the type
3263 * type declarations inside WITH-SLOTS are checked. (reported by
3265 * fixed some bugs revealed by Paul Dietz' test suite:
3266 ** incorrect optimization of TRUNCATE for a positive first
3267 argument and negative second.
3268 ** compiler failure in let-convertion during flushing dead code.
3269 ** compiler failure while deriving type of TRUNCATE on an
3270 interval, containing 0.
3271 ** ASH of a negative bignum by a negative bignum count now returns
3273 ** intersection of CONS types now canonicalizes properly, fixing
3274 inconsistencies in SUBTYPEP.
3276 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3277 * incompatible change: The --disable-debugger command line
3278 option now clobbers the debugger at a more fundamental
3279 level, by redefining #'INVOKE-DEBUGGER instead of by
3280 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3281 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3282 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3283 enter the debugger REPL and then suffer recursive errors
3284 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3285 while under the new system, BREAK in --disable-debugger mode
3286 terminates the system just as an unhandled error would.
3287 * fixed compiler performance when processing loops with a step >1;
3288 * bug fix: DOCUMENTATION now retrieves generic function
3289 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3290 support has been systematized, and now supports the methods
3291 specified by ANSI, along with a default method and a method for
3292 slot documentation. (reported by Nathan Froyd)
3293 * bug fix: effective methods associated with a generic function are
3294 no longer cached over a change of that generic function's method
3295 combination. (reported by Andreas Fuchs)
3296 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3297 names a non-existent directory. (thanks to Andreas Fuchs)
3298 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3299 the PPC platform, lead to stack corruption; this has been fixed.
3300 (reported by Rainer Joswig)
3301 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3302 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3303 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3304 arithmetic in general, is now much more reliable on the PPC
3306 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3307 platform now returns the right answer.
3308 * optimization: restored some effective method precomputation in
3309 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3310 precomputation is now tunable.
3311 * optimization: compiler-internal data structure use has been
3312 reviewed, and changes have been made that should improve the
3313 performance of the compiler by about 20%.
3314 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3315 simple-base-strings and simple-bit-vectors is improved.
3316 * optimization: the optimization of 32-bit logical and arithmetic
3317 functions introduced in version 0.8.3 on the x86 has been
3318 implemented on the mips, ppc and sparc platforms; an
3319 implementation of the same facility, but for 64-bit arithmetic,
3320 has been added for the alpha.
3321 * microoptimization: the compiler is better able to make use of the
3322 x86 LEA instruction for multiplication by constants.
3323 * bug fix: in some situations compiler did not report usage of
3324 generic arithmetic in (SPEED 3) policy.
3325 * bug 145b fix: compiler used wrong type specifier while converting
3326 MEMBER-types to numeric.
3327 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3329 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3330 index argument now works properly on the Alpha platform.
3331 * bug fix: floating point exception treatment on the Alpha platform
3333 * bug fix: FILE-POSITION works much better on string input and
3334 output streams. (thanks to Nikodemus Siivola)
3335 * bug fix: many threading/garbage collection symptoms sorted.
3336 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3337 might be pseudo-atomic.
3338 * internal change: Stopping for GC is now done with signals not
3339 ptrace. GC is now done in whichever thread wanted it, instead of
3341 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3343 * bug fix: result form in DO is not contained in the implicit
3345 * incompatible change: ICR structure is changed; the value part of
3346 CONTINUATION is now called LVAR; corresponding functions are
3347 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3348 * added type deriver for ISQRT (thanks to Robert E. Brown).
3349 * bug fix: better support for loading from the command line when an
3350 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3352 * fixed some bugs revealed by Paul Dietz' test suite:
3353 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3354 ** ROUND and FROUND now give the right answer when given very
3355 small float arguments.
3356 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3358 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3359 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3360 ** LCM should return a non-negative integer.
3361 ** PARSE-INTEGER returned the index of a terminator instead of the
3362 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3363 ** PARSE-INTEGER returned an incorrect index being applied to a
3365 ** LCM with two arguments of 0 returns 0 rather than signalling
3367 ** unsigned addition of a 32-bit constant with the high bit set no
3368 longer causes an internal compiler error on the x86.
3369 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3370 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3371 with negative last argument.
3372 ** byte specifiers with zero size and position no longer cause
3373 an error during type derivation.
3374 ** bignum multiplication on the Alpha platform now returns the
3376 * porting: The system now builds on SuSE AMD64, although it still
3377 generates a 32-bit binary.
3378 * .fasl file incompatibility: The fasl file version number has
3379 been incremented (because of the changes to internal compiler
3380 data structures referred to above).
3382 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3383 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3384 more accurately, on the Darwin kernel running on PowerPC hardware.
3385 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3386 * Compiler code deletion notes now signal a condition of type
3387 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3388 an associated MUFFLE-WARNING restart.
3389 * The compiler now performs limited argument count validation of
3390 constant format strings in FORMAT, and where appropriate in ERROR,
3391 CERROR and WARN. (thanks to Gerd Moellmann)
3392 * New ASDF-INSTALL contrib can be used for automatic download and
3393 installation of third-party Lisp code from CCLAN or other sites
3395 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3396 which forces another thread to execute a function supplied by the
3398 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3399 now accept and act upon their :ELEMENT-TYPE keyword argument.
3400 (reported by Martin Atzmueller, Edi Weitz)
3401 * bug fix: FILE-POSITION now accepts position designators up to
3402 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3403 is the greater. (thanks to Patrik Nordebo)
3404 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3405 Kalle Olavi Niemitalo)
3406 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3407 printable. (reported by Eric Marsden)
3408 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3409 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3410 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3411 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3412 (thanks to Stig Sandoe)
3413 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3414 UNREAD-CHAR now correctly includes the unread character in the
3415 target sequence. (thanks to Gerd Moellmann)
3416 * bug fix in threaded builds: the system can now be suspended and
3417 resumed by shell job control with minimal disruption.
3418 * bug fixes in times and timezones >2038AD
3419 * better handling of "where is GNU make?" problem in build scripts
3420 (thanks to Nikodemus Siivola)
3421 * new optimization: inside a named function any reference to a
3422 function with the same name is considered to be a self-reference;
3423 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3424 optimization quality.
3425 * new optimization on x86: logical functions and + now have
3426 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3427 used when the result is truncated to 32 bits.
3428 * VALUES declaration is partially enabled.
3429 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3430 * fixed some bugs revealed by Paul Dietz' test suite:
3431 ** The system now obeys the constraint imposed by
3432 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3433 types form a lattice under type intersection.
3434 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3435 ** ASSOC now ignores NIL elements in an alist.
3436 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3437 and (1+ MOST-POSITIVE-FIXNUM) answers.
3438 ** The addition of a method with invalid qualifiers to a generic
3439 function does not cause an error to be signalled immediately;
3440 a warning is signalled, and the error is generated only on
3441 calling the generic function.
3442 * changed .fasl file version number, in part to add type codes for
3443 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3444 obscure ANSI requirements
3446 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3447 * fixed bug 148: failure to inline-expand a local function left
3448 garbage, confusing the compiler.
3449 * fixed bugs 3cd: structure slot readers perform type check if the
3450 slot can have an invalid value (i.e. it is either not initialized
3451 or can be written with a less specific slot writer).
3452 * bug fix: the compiler now traps array references to elements off
3453 the end of an array; previously, the bounds checking in some
3454 circumstances could go off-by-one.
3455 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3457 * type declarations for array element types now obey the description
3458 on the CLHS page "Declaration TYPE", as per discussions on
3459 sbcl-help around 2003-05-08. This means that a declaration
3460 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3461 declaration, all references to BAR will be asserted or assumed
3462 (with THE, so dependent on compiler policy) to involve objects of
3463 type FOO. Note that no such declaration is implied in
3464 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3465 * declared types of functions from the "Conditions"
3466 chapter. (reported by Paul Dietz)
3467 * bug fix: CERROR accepts a function as its first argument.
3468 * bug fix: NTH an NTHCDR accept a bignum as index
3469 arguments. (reported by Adam Warner)
3470 * optimization: character compare routines now optimize comparing
3471 against a constant character. (reported by Gilbert Baumann)
3472 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3473 argument now works properly on the MIPS platform.
3474 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3475 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3476 declared to have a functional type, does not cause an error
3478 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3479 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3480 * fixed bug: if last continuation of a deleted block has a
3481 destination, this destination should be deleted too. (reported by
3483 * fixed a bug in the bootstrap process: the host compiler's values
3484 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3485 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3486 test case from Patrik Nordebo)
3487 * improved the ability of the disassembler on the PPC platform to
3488 provide helpful disassembly notes.
3489 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3490 the class in more cases than previously.
3491 * bug fix: FILE-POSITION now understands :START and :END for
3492 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3493 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3494 8) data. (thanks to David Lichteblau)
3495 * bug fix: it is possible to add a method to a generic function
3496 without lambda list.
3497 * bug fix: reader failed to signal END-OF-FILE inside an
3498 object representation. (reported by Nikodemus Siivola)
3499 * fixed some bugs revealed by Paul Dietz' test suite:
3500 ** LAST and [N]BUTLAST should accept a bignum.
3501 ** condition slot accessors are methods.
3502 ** (VECTOR NIL) is a subtype of STRING.
3504 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3505 * minor incompatible change: some nonsensical specialized lambda
3506 lists (used in DEFMETHOD) which were previously ignored now signal
3508 * minor incompatible change: the system is now aware of the types of
3509 variables in the COMMON-LISP package, and will signal errors for
3510 most violations of these type constraints (where previously they
3511 were silently accepted).
3512 * minor incompatible change: COMPILE-FILE now uses the freedom
3513 afforded (ANSI 3.2.2.3) to use derived function types for
3514 functions defined in the same file. This also permits the system
3515 to warn on static type mismatches and function
3516 redefinition. (Currently it does not work with high DEBUG level.)
3517 * minor incompatible change: VALUES declaration is disabled.
3518 * When issuing notes, the compiler now signals a condition of type
3519 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3520 restart for use in user handlers. It is expected that the
3521 COMPILER-NOTE condition will eventually become a condition
3522 supertype to a hierarchy of note types, which will then be
3523 handleable in a similar fashion. However, at the moment, no such
3524 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3525 but it's an implementation detail, not a classification for the
3527 * Changes in type checking closed the following bugs:
3528 ** type checking of unused values (192b, 194d, 203);
3529 ** template selection based on unsafe type assertions (192c, 236);
3530 ** type checking in branches (194bc).
3531 * A short form of VALUES type specifier has ANSI meaning (it has
3532 increased the number of situations when SBCL cannot perform type
3534 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3535 keywords or constants is permissible.
3536 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3537 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3538 classes. (thanks to Antonio Martinez)
3539 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3540 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3541 argument to a pretty-print stream. Instead, it leaves any such
3542 operations to DESCRIBE-OBJECT methods.
3543 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3545 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3546 lambda lists. (thanks to David Lichteblau)
3547 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3548 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3549 (thanks to Gerd Moellmann)
3550 * fixed bug 246: increased compilation speed of long
3551 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3553 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3555 * DEFCONSTANT now throws a condition of type
3556 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3557 constant to a non-EQL value; CONTINUE and ABORT restarts
3558 respectively change and preserve the value.
3559 * fixed bug 63: The code walker, part of the implementation of CLOS,
3560 is now better at handling symbol macros.
3561 * bug fix: There is no longer an internal implementation type named
3562 CL:LENGTH. (reported by Raymond Toy)
3563 * bug fix: In macro-like defining macros/special operators the
3564 implicit block does not enclose the lambda list.
3565 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3566 enclosed in parentheses) are not suitable as type specifiers, and
3567 their use properly signals an error now.
3568 * bug fix: An argument count mismatch for a type specifier in code
3569 being compiled no longer causes an unhandled error at compile
3570 time, but signals a compile-time warning.
3571 * fixed simple vector readable printing
3572 * bug fix: DESCRIBE takes more care over whether the class
3573 precedence list slot of a class is bound before accessing it.
3574 (reported by Markus Krummenacker)
3575 * bug fix: FORMATTER can successfully compile pretty-printer format
3576 strings which use variants of the ~* directive inside.
3577 * bug fix: SEARCH now applies its TEST predicate to the elements of
3578 the arguments in the correct order. (thanks to Wolfhard Buss)
3579 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3580 in type inference. (thanks to Robert E. Brown)
3581 * bug fix: Reading in symbols with an explicit package name of ""
3582 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3583 * fixed some bugs revealed by Paul Dietz' test suite:
3584 ** NIL is now allowed as a structure slot name.
3585 ** Arbitrary numbers, not just REALs, are allowed in certain
3586 circumstances in LOOP for-as-arithmetic clauses.
3587 ** Multiple class redefinitions before slot access no longer
3588 causes a type error.
3589 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3590 association between the name and a class.
3591 ** Generic functions with non-standard method-combination and over
3592 five methods all of which return constants no longer return NIL
3593 after the first few invocations. (thanks to Gerd Moellmann)
3594 ** CALL-NEXT-METHOD with no arguments now passes the original
3595 values of the arguments, even in the presence of assignment.
3596 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3598 ** DISASSEMBLE works with closures and funcallable instances.
3599 ** ADD-METHOD now returns the generic function, not the new method.
3600 ** FIND-METHOD signals an error if the lengths of the specializers
3601 is incompatible with the generic function, even if the ERRORP
3603 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3604 which its argument is a member.
3605 ** DEFCLASS only redefines the class named by its class-name
3606 argument if that name is the proper name of the class;
3607 otherwise, it creates a new class.
3608 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3609 of the UNBOUND-SLOT condition to the name of the slot.
3610 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3611 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3612 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3613 SLOT-MAKUNBOUND in the specified fashion.
3615 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3616 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3617 cross-compilation host. As a consequence, we can now bootstrap our
3618 way up to SBCL starting with a bare gcc toolchain and human-readable
3619 source code (first the source to CLISP, then the source to SBCL).
3620 * A contributed module containing a partial implementation of the
3621 simple-streams interface has been included. (thanks to Rudi
3623 * A contributed module implementing the RFC1321 Message Digest
3624 Algorithm, known as MD5, has been included.
3625 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3626 no longer has any effect, as the code controlled by this feature
3627 has been deleted. (As far as we know, no-one has ever built using
3628 this feature, and its semantics were confused in any case).
3629 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3630 work on conditions (as required by the ANSI specification),
3631 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3632 expected behaviour on conditions. Users should note, however,
3633 that such behaviour is not required by the ANSI specification,
3634 and so use of this behaviour may render their code unportable.
3635 * fixed some bugs revealed by Paul Dietz' test suite:
3636 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3638 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3639 ** SXHASH on condition objects no longer returns NIL.
3640 ** :ALLOCATION :CLASS slots are better treated; their values are
3641 updated on class redefinition, and initforms inherited from
3642 superclasses are applied.
3643 ** REMOVE-METHOD returns its generic function argument even when
3644 no method was removed.
3645 ** SHARED-INITIALIZE now initializes the values of the requested
3646 slots, including those with :ALLOCATION :CLASS.
3647 ** ALLOCATE-INSTANCE now works on structure classes defined via
3648 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3650 ** SLOT-EXISTS-P now works on conditions, as well as structures
3652 ** MAKE-LOAD-FORM now has the required methods on
3653 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3654 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3655 keyword, and now implements the SLOT-NAMES argument.
3656 ** methods with &OPTIONAL arguments no longer allow too many
3657 arguments to be passed in the call without error.
3658 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3659 option is consistent with the required arguments of the generic
3660 function lambda list.
3661 * bug fix: REQUIRE accepts a string designator. (Thanks to
3663 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3664 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3665 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3666 * bug fix: The compiler now checks for duplicated variables in macro
3668 * bug fix: SETQ on globals returns the correct value.
3669 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3670 not a COMPILER-ERROR (followed by some other strange error on
3671 choosing the CONTINUE restart).
3672 * bug fix: make.sh and friends are now more consistent in the way that
3673 they look for GNU "make".
3675 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3676 * experimental native threads support (on x86 Linux >=2.4 only).
3677 This is not compiled in by default: you need to add :SB-THREAD to
3678 the target features. See the "Beyond ANSI" chapter of the manual
3680 * fix for longstanding nonANSIism: The old distinction between
3681 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3682 The return value from CL:FIND-CLASS is now a CLOS class, and
3683 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3684 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3685 * An interface to the MetaObject Protocol, as described in Kiczales,
3686 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3687 MIT Press, 1991, is available from the SB-MOP package.
3688 * incompatible change: the SB-PCL package should now be considered
3689 a private implementation detail, and no longer a semi-private MOP
3691 * minor incompatible change: due to rearrangement for threads, the
3692 control stack and binding stack are now allocated at arbitrary
3693 addresses instead of being hardcoded per-port. Users affected by
3694 this probably have to be doing advanced things with shared
3695 libraries, and will know who they are.
3696 * minor incompatible change: Previously, all --eval forms used were
3697 processed with READ before any of them were processed with EVAL.
3698 Now each --eval form is processed with both READ and EVAL before
3699 the next --eval form is processed. (Thus package operations like
3700 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3701 work as the user might reasonably expect.)
3702 * minor incompatible change: *STANDARD-INPUT* is now only an
3703 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3705 * minor incompatible change: Y-OR-N-P is now character-oriented,
3706 not line oriented. Also, YES-OR-NO-P now works without errors.
3707 (thanks to Antonio Martinez)
3708 * sb-aclrepl module improvements: an integrated inspector, added
3709 repl features, and a bug fix to :trace command.
3710 * Known functions, which cannot be open coded by the backend, are
3711 considered to be able to check types of their arguments. (fixing
3712 a bug report by Nathan J. Froyd)
3713 * fixed a bug in computing method discriminating functions: It is
3714 now possible to define methods specialized on classes which have
3715 forward-referenced superclasses. (thanks to Gerd Moellmann)
3716 * fixed evaluation order in optional entries (reported by Gilbert
3718 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3719 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3720 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3721 required-by-AMOP NAME argument, as well as CLASS and
3722 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3723 * fixed bug 20: DEFMETHOD can define methods using names that are
3724 not the proper names of classes to designate class specializers.
3725 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3726 Unix file descriptors, instead of blowing up. (thanks to Antonio
3728 * Garbage collection refactoring: user-visible change is that a
3729 call to the GC function during WITHOUT-GCING will not do garbage
3730 collection until the end of the WITHOUT-GCING. If you were doing
3731 this you were probably losing anyway.
3732 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3733 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3734 TYPEP the latter but not the former.
3735 * The compiler issues a full WARNING for calls to undefined functions
3736 with names from the CL package.
3737 * MAP-INTO for a vector destination is open coded. (reported by
3738 Brian Downing on c.l.l)
3739 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3740 documentation string.
3741 * fixed some bugs revealed by Paul Dietz' test suite:
3742 ** COPY-ALIST now signals an error if its argument is a dotted
3744 ** Condition slots are now accessed more correctly in the presence
3745 of multiple initargs for a given slot.
3746 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3747 exclude restarts of the same name associated with a different
3749 ** DEFCLASS of forward-referenced classes with another
3750 forward-referenced class in the superclasses list no longer
3752 ** Condition slots are now initialized once each, not multiple
3753 times. (thanks to Gerd Moellmann)
3754 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3756 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3757 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3758 arguments contain duplicated elements.
3759 ** RESTART-CASE understands local macros.
3760 ** RESTART-CASE associates exactly its own restarts with a condition.
3761 ** ENDP in safe mode checks its argument to be of type LIST.
3762 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3763 in question is unbound.
3764 ** Optimized MAKE-INSTANCE functions no longer cause internal
3765 assertion failures in the presence of duplicate initargs.
3766 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3767 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3768 propagated even in the ctor optimized implementation of
3770 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3772 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3773 * a better implementation of SXHASH on (simple) bit vectors,
3774 measured both in execution speed and in distribution of results
3775 over the positive fixnums, has been installed. Likewise, a better
3776 implementation of EQUAL for simple bit vectors is now available.
3777 * fixed CEILING optimization for a divisor of form 2^k.
3778 * fixed bug 240 (emitting extra style warnings "using the lexical
3779 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3780 by Antonio Martinez)
3781 * fixed SXHASH, giving different results for NIL depending on type
3782 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3783 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3784 variable. (found by Rolf Wester)
3785 * SBCL does not ignore type declarations for special
3786 variables. (reported by rif on c.l.l 2003-03-05)
3787 * some bug fixes in contrib/sb-aclrepl/
3788 * fixed some bugs revealed by Paul Dietz' test suite:
3789 ** a bug in the CONS type specifier, whereby the CAR and CDR
3790 types got intertwined, has been fixed;
3791 ** the type system is now able to reason about the interaction
3792 between INTEGER and RATIO types more completely;
3793 ** APPEND, [N]REVERSE and NRECONC check that those their
3794 arguments, which must be proper lists, are really so;
3795 ** An array specialized to be unable to hold elements has been
3796 implemented, as required -- yes, really -- by ANSI;
3797 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3798 on malformed property lists;
3800 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3801 * incompatible packaging change: in line with Unix convention,
3802 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3803 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3804 * REQUIRE and PROVIDE are now optionally capable of doing something
3805 useful. See the documentation string for REQUIRE.
3806 * infrastructure for a managed SBCL contrib system: contributed
3807 modules in this release include:
3808 ** the ASDF system definition facility;
3809 ** an interface to the BSD Sockets API;
3810 ** an ACL-like convenience interface to the repl;
3811 (thanks to Kevin Rosenberg)
3812 ** an implementation of ROTATE-BYTE, with efficient implementation
3814 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3815 gives a closer approximation to the right answer than previously.
3816 (thanks to Raymond Toy)
3817 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3818 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3819 optional environment arguments, as required by ANSI.
3820 * fixed bugs in other functions taking environment objects, allowing
3821 calls with an explicit NIL environment argument to be compiled
3823 * fixed bug 228: primary return values from
3824 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3825 COMPILE or FUNCTION.
3826 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3827 :TYPE will no longer signal errors on innocuous objects.
3828 * fixed bug 231b: SETQ is better at respecting type declarations in
3829 the lexical environment.
3830 * fixed a bug in DEFCLASS: classes named by symbols with no or
3831 unprintable packages can now be defined.
3832 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3833 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3834 * fixed compiler failure related to checking types of functions.
3835 (reported by Robert E. Brown)
3836 * the compiler is now much more consistent in its error-checking
3837 treatment of bounding index arguments to sequence functions: in
3838 (SAFETY 3) code, errors will be signalled in almost all cases if
3839 invalid sequence bounding indices are passed to functions defined
3840 by ANSI to operate on sequences.
3841 * fixed a bug in the build procedure: documentation of SBCL-specific
3842 packages is now preserved and available in the final Lisp image.
3843 * lifted FDEFINITION lookup out of loops in the implementation of
3844 many list operations. (thanks to Robert E. Brown)
3845 * fixed a bug in the reader: the #n# reader macro now works for
3846 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3847 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3848 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3849 E. Brown and Tony Martinez respectively)
3850 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3851 SLOT-BOUNDP: in particular, it is now able to optimize them much
3852 better, and is now not vulnerable to having packages renamed.
3853 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3854 always called when a slot is not present in an instance. (thanks
3856 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3857 longer signal an error on unreading a character following EOF on
3858 the previous constituent stream. (thanks to Tony Martinez)
3859 * fixed some bugs revealed by Paul Dietz' test suite:
3860 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3861 not just nonnegative fixnums;
3862 ** the logical bit-array operators such as BIT-AND now accept an
3863 explicit NIL for their "opt-arg" argument (to indicate a
3864 freshly-consed result bit-array);
3865 ** ELT now signals an error on an invalid sequence index in safe
3867 ** the type system is now cleverer about negations of numeric
3868 types, and consequently understands the BIGNUM and RATIO types
3870 ** the type system is now cleverer about the interaction between
3871 INTEGER and RATIO types: while bugs still remain, many more
3872 cases are accurately computed;
3873 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3874 if it is in the last clause;
3875 ** CONSTANTLY now correctly returns a side-effect-free function in
3877 ** DECLARE is no longer treated as a special-operator; in
3878 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3879 * incremented fasl file version number due to the change in the
3880 DEFSTRUCT-SLOT-DESCRIPTION structure.
3882 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3883 * minor incompatible change: code processed by the "interpreter" or
3884 EVAL now has a compilation optimization policy of (DEBUG 2)
3885 (changed from (DEBUG 1)) to improve debuggability of interactive
3886 development, and to allow the use of the debug RETURN command in
3888 * an experimental implementation of the RETURN command for the
3889 debugger has been included. (thanks to Frederik Kuivinen)
3890 * fixed bug 62: constraints were not propagated into a loop.
3891 * fixed bug in embedded calls of SORT (reported and investigated by
3893 * fixed some bugs revealed by Paul Dietz' test suite:
3894 ** printing and reading of arrays with some dimensions having
3895 length 0 (thanks to Gerd Moellmann);
3896 ** BOA constructor with &AUX argument without a default value does
3897 not cause a type error;
3898 ** CONSTANTP now returns true for all self-evaluating objects.
3900 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3901 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3902 accessors that are related by inheritance, as specified in the
3903 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3904 Valtteri Vuorikoski)
3905 * The compiler is now able to inline functions that were defined in
3906 a complex lexical environment (e.g. inside a MACROLET).
3907 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3909 * Support for the upcoming FreeBSD-5.0 release has been included.
3910 (thanks to Dag-Erling Smorgrav)
3911 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3912 effect when it is not in a toplevel context.
3913 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3914 stand a better chance of being correct. (thanks to Gerd
3916 * fixed bug in COERCE, which now signals an error on coercing a
3917 rational to a bounded real type which excludes the expected
3919 * The compiler is now able to derive types more accurately from the
3920 COERCE and COMPILE functions.
3921 * fixed bug 223: functional binding is considered to be constant
3922 only for symbols in the CL package.
3923 * fixed bug 231: SETQ did not check the type of a variable being set
3924 (reported by Robert E. Brown)
3925 * A new optimization for MAKE-INSTANCE has been included, fixing
3926 various bugs (including relating to :ALLOCATION :CLASS slots and
3927 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3929 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3930 ** As required by ANSI, LOOP now disallows anonymous collection
3931 clauses such as COLLECT I in conjunction with aggregate boolean
3932 clauses such as THEREIS (= I 1);
3933 ** LOOP now signals an error when any variable is reused in the
3934 same loop (including the potentially useful construct analogous
3935 to WITH A = 1 WITH A = (1+ A);
3936 ** IT is only a special loop symbol within the first clause of a
3937 conditional loop clause;
3938 ** LOOP with a typed iteration variable over a hashtable now
3939 signals a type error iff it should.
3940 * fixed some other bugs revealed by Paul Dietz' test suite:
3941 ** FILE-STREAM now names the class previously known as FD-STREAM;
3942 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3943 argument) no longer signals an error;
3944 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3945 of the slot symbol, rather than using the current package
3946 ((:CONC-NAME "") continues to intern the slot's name in the
3948 * incremented fasl file version number, because of the incompatible
3949 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3950 of the new implementation of DEFINE-COMPILER-MACRO.
3952 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3953 * Support for building SBCL for MIPS platforms running in
3954 little-endian mode has now been checked in, and basic
3955 functionality on said platforms verified.
3956 * minor incompatible change: PCL now records the pathname of a file
3957 in which methods and the like are defined, rather than its
3959 * minor incompatible change: TRUENAME now considers the truename of
3960 a file naming a directory to be the pathname with :DIRECTORY
3961 component indicating that directory.
3962 * minor incompatible change: a NAMED clause in the extended form of
3963 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3964 reason for the previous behaviour is unclear.
3965 * more systematization and improvement of CLOS and MOP conformance
3966 in PCL (thanks to Gerd Moellman and Pierre Mai):
3967 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3969 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3970 primary methods with no specializers;
3971 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3973 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3974 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3975 FORWARD-REFERENCED-CLASSes; error reporting on
3976 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3978 ** SXHASH on CLOS instances now uses a slot internal to the
3979 instance to return different numbers on distinct instances,
3980 while preserving the same return value through invocations of
3982 ** DEFMETHOD signals errors when methods with longer incongruent
3983 lambda lists are added to generic functions;
3984 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3985 CLASS, as specified in AMOP;
3986 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3987 on the order returned by the primary method for classes of
3988 class STANDARD-CLASS;
3989 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3990 * fixed some bugs shown by Paul Dietz' test suite:
3991 ** DOLIST puts its body in TAGBODY;
3992 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3994 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3995 value producing form;
3996 ** if more variables are given to PROGV than values, extra
3997 variables are bound and made to have no value;
3998 ** NSUBSTITUTE on list arguments gets the right answer with
4000 ** ELT signals an error of type TYPE-ERROR when the index argument
4001 is not a valid sequence index;
4002 ** LOOP signals (at macroexpansion time) an error of type
4003 PROGRAM-ERROR when duplicate variable names are found;
4004 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
4005 ** FUNCALL of special-operators now cause an error of type
4007 ** PSETQ now works as required in the presence of side-effecting
4008 symbol-macro places;
4009 ** NCONC accepts any object as its last argument;
4010 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
4012 ** loop-for-as-package does not require a package to be explicitely
4014 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
4016 * fixed bug 166: compiler preserves "there is a way to go"
4017 invariant when deleting code.
4018 * fixed bug 172: macro lambda lists with required arguments after
4019 &REST arguments now cause an error to be signalled. (thanks to
4021 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
4022 bug. (thanks to Matthew Danish)
4023 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
4025 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
4026 arguments are lexically rebound. (thanks to Gerd Moellmann and
4028 * fixed bug 194: error messages are now more informative when there
4029 is no primary method applicable in a call to a generic
4030 function. (thanks to Gerd Moellmann)
4031 * fixed bug in command line argument checking (thanks to Julian
4033 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
4034 (thanks to Matthew Danish)
4035 * incremented fasl file version number, because of the
4036 SXHASH-related changes in the layout of CLOS data structures
4038 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
4039 * minor incompatible change: The runtime (the Unix executable named
4040 "sbcl") is now much pickier about the .core files it will load.
4041 Essentially it now requires .core files to descend from the same
4042 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
4043 as the runtime does. (The intent is to prevent the crashes which
4044 can occur, and which can even be reported as mysterious failures,
4045 when people patch the sources or change the build parameters
4046 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
4047 sbcl and .core files.)
4048 * fixed bug: VALUES-LIST is no longer optimized away.
4049 * fixed bug 142: The FFI conversion of C string values to Lisp
4050 string values no longer conses excessively. (thanks to Nathan
4051 Froyd porting Raymond Toy's fix to CMU CL)
4052 * began to systematize and improve MOP conformance in PCL (thanks to
4053 Nathan Froyd, Gerd Moellman and Pierre Mai):
4054 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
4056 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
4057 ** FINALIZE-INHERITANCE is now called on class finalization;
4058 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
4059 argument precedence order.
4060 * fixed bug 202: The compiler no longer fails on functions whose
4061 derived types contradict their declared type.
4062 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
4063 so it can be non-toplevel.
4064 * The fasl file version number has changed (because of the new
4065 implementation of DEFMACRO).
4066 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
4067 safe code, that any length requirement by their type-specifier
4068 argument is valid. The exceptions to this are described in bug
4070 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
4071 functions now check the types of their inputs as required by ANSI.
4072 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
4073 introduced symbol is DECLAREd to be SPECIAL.
4074 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
4075 * fixed bug: :COUNT argument to sequence functions may be negative
4076 * fixed bug: body of DO-SYMBOLS may contain declarations
4077 * fixed bug: PUSHNEW now evaluates its arguments from left to right
4078 (reported by Paul F. Dietz, fixed by Gerd Moellman)
4079 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
4080 symbol macro only once
4081 * fixed printing of call frame when argument list is unavailable
4082 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
4083 * compiler no longer signals WARNING on unknown keyword
4086 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
4087 * A beta-quality port to the mips architecture running Linux,
4088 based on the old CMUCL backend, has been made. It has been tested
4089 on a big-endian kernel, and works sufficiently well to be able to
4090 rebuild itself; it has not been tested in little-endian mode.
4091 * fixed an inconsistency between gencgc.c and purify.c which made
4092 dumping/loading .core files unreliable
4093 * fixed bug 120a: The compiler now deals correctly with IFs where
4094 the consequent is the same as the alternative, instead of
4095 misderiving the return type. (thanks to Alexey Dejneka)
4096 * fixed bug 113: Logical pathnames are now dumpable (the logical
4097 host is resolved at load-time, throwing an error if it is not
4099 * fixed bug 174: FORMAT's error message is slightly clearer when a
4100 non-printing character is used in a format directive.
4101 * fixed several bugs in compiler checking of type declarations, i.e.
4102 violations of the Python "declarations are assertions" principle
4103 (thanks to Alexey Dejneka)
4104 * fixed several bugs in PCL's error checking (thanks to Gerd
4106 * fixed bug: printing of FILE-ERROR (thanks to Antonio
4108 * fixed bug in compilation of functions as first class values
4109 (thanks to Antonio Martinez-Shotton)
4110 * The compiler's handling TYPE-ERRORs which it can prove will
4111 inevitably happen at runtime has been cleaned up and corrected
4112 in several ways. (thanks to Alexey Dejneka)
4113 * improved argument type checking for various basic arithmetic
4114 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
4115 lavished on them in the past that they can be compiled in many
4116 ways in different special cases
4117 * fixed bug 181: compiler checks validity of user supplied type
4119 * cleaned up code flushing in optimization: Function calls which
4120 should signal errors for safety purposes (e.g. which ANSI says
4121 should signal errors when their arguments are of incorrect type)
4122 are no longer optimized away.
4123 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
4124 * incremented fasl file version number, because changes in the
4125 implementation of sequence functions like COERCE caused
4126 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
4127 in old inline expansions) to become undefined. (Actually these
4128 changes were later undone, so we might very well be binary
4129 compatible with 0.7.7 after all, but leaving the version number
4130 incremented seemed like the simplest and most conservative
4133 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
4134 * An alpha-quality port to the parisc architecture running Linux,
4135 based on the old CMUCL backend, has been made. This, even more so
4136 than the other backends, should be considered still a work in
4137 progress; known problems include that the Linux kernel in 64-bit
4138 mode does not propagate the correct sigcontext structure to
4139 userspace, and consequently SBCL on a parisc64 kernel will not
4141 * fixed bug 189: The compiler now respects NOTINLINE declarations for
4142 functions declared in FLET and LABELS. (I.e. "LET conversion" is
4143 suppressed.) Also now that the compiler is looking at declarations
4144 in the environment, it checks optimization declarations as well,
4145 and suppresses inlining when (> DEBUG SPEED).
4146 * More fixes have been made to treatment of floating point exception
4147 treatment and other Unix signals. In particular, floating point
4148 exceptions no longer cause Bus errors on the SPARC/Linux platform.
4149 * The detection and handling of control stack exhaustion (infinite
4150 or very deeply nested recursion) has changed. Stack exhaustion
4151 detection is now done by write-protecting pages at the OS level
4152 and applies at all optimization settings; when found, a
4153 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
4154 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
4155 be caught using IGNORE-ERRORS.
4156 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
4157 correctly when transforming recursive LABELS functions to LETs.
4158 (thanks to Alexey Dejneka)
4159 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
4160 that are names of constants or global variables.
4161 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
4162 alien routines with docstrings.
4163 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
4164 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
4166 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
4167 (thanks to Eric Marsden)
4168 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
4169 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
4170 * Bug fix: The SPARC backend can now compile functions involving
4171 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
4172 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
4173 to a functions expecting a PATHNAME-DESIGNATOR.
4174 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
4175 lambda lists. (thanks to Alexey Dejneka)
4176 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
4177 functions so that EXPORTing the name of the class doesn't cause
4178 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
4179 to Antonio Martinez for reporting this)
4180 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
4181 constant-folded without failing an assertion. (thanks to Einar
4182 Floystad Dorum for reporting this)
4183 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
4184 types are now dealt with more correctly by the compiler.
4185 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
4186 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
4187 behaviour analogous to RENAME-FILE. This puts its behaviour more
4188 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
4190 * The fasl file version number has changed again. (because of the
4191 bug fix involving the names of PCL MAKE-INSTANCE functions)
4193 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
4194 * bug fix: Floating point exceptions are treated much more
4195 consistently on the x86/Linux and PPC/Linux platforms.
4196 * Array initialization with :INITIAL-ELEMENT is now much faster for
4197 cases when the compiler cannot open code the array creation, but
4198 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
4199 array accesses have also seen a speed increase.
4200 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
4201 specified. (This was at the root of some bad interactions between
4202 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
4204 * bug fix: Internal error arguments for undefined functions are now
4205 computed correctly on the PPC/Linux platform.
4206 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4207 Raymond Toy's patch for CMU CL)
4208 * Support for the Solaris 9 operating environment has been included
4209 (thanks to Daniel Merritt)
4210 * A very ugly but hopefully complete draft of the missing FFI chapter
4211 of the manual has been created by reformatting the corresponding
4212 CMU CL manual chapter into (currently very ugly and incoherent)
4213 DocBook and bringing it up to date for SBCL behavior. Thus, the
4214 manual is now essentially complete, at least by my extreme
4215 once-and-only-once standards, whereby it's acceptable to refer to
4216 the doc strings of SB-EXT functions as the primary documentation.
4217 * The fasl file version number has changed again, due to cleanup of
4218 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4219 is no longer a static symbol.)
4221 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4222 * SBCL now builds with OpenMCL (version 0.12) as the
4223 cross-compilation host; also, more progress has been made toward
4224 bootstrapping under CLISP.
4225 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4227 * bug 158 fixed: The compiler can now deal with integer loop
4228 increments different from 1; fixing this turned out also to fix
4230 * bug 169 fixed: no more bogus warnings about using lexical bindings
4231 despite the presence of perfectly good SPECIAL declarations (thanks
4232 to David Lichteblau)
4233 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4234 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4235 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4237 * bug 180 fixed: Method combination specifications no longer ignore
4238 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4239 * bug fix: Structure type predicate functions now check their argument
4240 count as they should.
4241 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4242 correctly. (thanks to Pierre Mai)
4243 * minor incompatible change: The --noprogrammer option is deprecated
4244 in favor of the new --disable-debugger option, which is very similar.
4245 (The major difference is that it takes effect at a slightly different
4246 time at startup, causing handling of errors in --sysinit and
4247 --userinit files will be affected differently.) The
4248 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4249 been added to allow this functionality to be controlled from ordinary
4250 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4251 maintainers, who might want to run non-interactive scripts to
4252 build SBCL cores which will later be used interactively.)
4253 * minor incompatible change: The LOAD function no longer, when given
4254 a wild pathname to load, loads all files matching that pathname.
4255 Instead, an error of type FILE-ERROR is signalled.
4257 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4258 * bug 147 fixed: The compiler preserves its block link/count
4259 invariants more correctly now so that it doesn't crash. (thanks
4261 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4263 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4264 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4265 no-value-for-this-character as NIL (as specified by ANSI).
4266 * HOST-NAMESTRING on physical pathnames now returns a string that is
4267 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4268 (thanks to Christophe Rhodes)
4269 * The Alpha port handles icache flushing more correctly. (thanks to
4271 * More progress has been made toward bootstrapping under CLISP. (thanks
4272 to Christophe Rhodes)
4273 * The fasl file format has changed again, because dynamic loading
4274 on OpenBSD (which has non-ELF object files) motivated some cleanups
4275 in the way that foreign symbols are transformed and passed around.
4276 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4277 is no longer treated as whitespace by the reader, but instead as
4278 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4279 |A
\7fB|, instead of A as it used to.
4281 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4282 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4283 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4284 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4285 this as of 0.7.1.45, but was left out of the previous news section
4286 (thanks to Dan Barlow)
4287 * SBCL now runs on the Solaris operating system on SPARC architectures
4288 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4289 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4290 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4291 * SPARC backend cleanups, allowing builds of cores optimized for V8
4292 and V9 SPARCS, and also emission of code targeted to a particular
4293 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4295 * SBCL is closer to bootstrapping under CLISP, thanks to various
4296 fixes by Christophe Rhodes.
4297 * The fasl file format has changed again, to allow the compiler's
4298 INFO database to support symbol macros.
4299 * The user manual (in doc/) is formatted into HTML more nicely.
4300 (thanks to coreythomas)
4301 * The system is smarter about SUBTYPEP relationships, especially
4302 those involving NOT types (including types such as ATOM which are
4303 represented internally using NOT types). Thus SUBTYPEP is less
4304 likely to return (VALUES NIL NIL) in general, and in particular
4305 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4307 * The fasl file format has changed again, because the internal
4308 representation of types now includes a new slot to support the new
4309 SUBTYPEP-of-NOT-types logic.
4310 * (not a change in the main branch of SBCL, but a related prototype
4311 which can hopefully be merged into the main branch of SBCL in the
4312 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4313 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4314 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4315 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4316 dispatch macro character combinations. (thanks to Alexey Dejneka)
4317 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4318 that superficially look like logical namestrings correctly.
4319 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4321 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4322 default supplied-p) for &optional and &key arguments. (thanks to
4325 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4326 * incompatible change: The compiler is now less aggressive about
4327 tail call optimization, doing it only when (> SPACE DEBUG) or
4328 (> SPEED DEBUG). (This is an incompatible change because there are
4329 programs which relied on the old CMU-CL-style behavior to optimize
4330 away their unbounded recursion which will now die of stack overflow.)
4331 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4332 for non-GENCGC systems has been increased to 20M (since that
4333 seems much closer to the likely performance optimum for modern
4334 systems than the old 4M value was)
4335 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4336 and *DEBUG-PRINT-LEVEL*
4337 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4338 of CMU CL's support for SPARC, and various endianness and other
4339 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4340 * new syntactic sugar for the Unix command line: --load foo.bar is now
4341 an alternate notation for --eval '(load "foo.bar")'.
4343 ** The system now detects stack overflow and handles it gracefully,
4344 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4345 optimization settings. (This is a good thing in general, and
4346 its introduction in this version should be particularly timely
4347 for anyone whose code fails because of suppression of tail
4349 ** The system now hunts for the C variable "environ" in a more
4350 devious way, to avoid segfaults when the C library version differs
4351 between compile time and run time. (thanks to Christophe Rhodes)
4352 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4353 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4354 ** The compiler no longer issues bogus style warnings for undefined
4355 classes in the same source file as the DEFCLASSes which defined
4356 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4358 ** fixes in CONDITION class precedence list for undefined function
4359 errors (thanks to Alexei Dejneka)
4360 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4361 correctly. (thanks to Dan Barlow)
4362 ** portability fixes aiming at bootstrapping under CLISP (thanks
4363 to Dave McDonald and Christophe Rhodes)
4364 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4365 ** fixes in type translation and and type inference (thanks to
4367 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4368 ** various fixes in the new ports (thanks to Dan Barlow)
4369 * several changes related to debugging:
4370 ** suppression of tail recursion, as noted above
4371 ** stack overflow detection, as noted above
4372 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4373 is now the default. (For some time encapsulation has been more
4374 reliable than the breakpoint-based :ENCAPSULATE NIL
4375 implementation, at least on X86 systems; and I just noticed that
4376 encapsulation also seems closer to the spirit of the ANSI
4379 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4381 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4382 up properly again. (There was a packaging bug in 0.7.0 which
4383 left their definitions in SB-SYS::LOAD-FOREIGN and
4384 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4385 vital for most things which interface to C-level interfaces,
4386 like extensions working with sockets or databases or
4387 Perl-compatible regexes or whatever, and the need to fix
4388 this bug is the main reason that 0.7.1 was released so
4390 ** DEFGENERIC is now choosier about the methods it redefines, so that
4391 reLOADing a previously-LOADed file containing DEFGENERICs does
4392 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4393 cycle now works as it should. (thanks to Alexey Dejneka)
4394 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4395 fixed by Christophe Rhodes. (He actually submitted this patch
4396 months ago, and I delayed until after 0.7.0.)
4397 ** Bug 111 (internal compiler confusion about runtime checks on
4398 FUNCTION types) was fixed by Alexey Dejneka.
4399 * Some internal cleanups (getting rid of variables which aren't
4400 needed now that the byte interpreter is gone) caused the fasl
4401 file format number to change again.
4403 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4404 * major incompatible change: The default fasl file extension, i.e. the
4405 default extension for files produced by COMPILE-FILE, has changed
4406 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4408 ** There are many changes in the implementation of the compiler.
4409 SBCL is now essentially a compiler-only implementation of ANSI
4410 Common Lisp. EVAL still "interprets" a few special cases, but
4411 almost all the interesting cases are handled by creating
4412 a LAMBDA expression, calling COMPILE on it, then calling
4413 FUNCALL on the result.
4414 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4415 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4416 Since the code is newer, there might still be some new bugs
4417 (though not as many as before Martin Atzmueller's fixes:-). But
4418 the new code is substantially simpler and clearer, and hopefully
4419 any remaining bugs will be simpler, less fundamental, and more
4420 fixable then the bugs in the old code.
4421 ** The revised compiler is still a little unsteady on its feet.
4423 *** The debugging information it produces (particularly the names
4424 of FUNCTION objects) is sometimes much less useful than what
4425 the old compiler produced.
4426 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4427 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4428 within a MACROLET) has been temporarily weakened.
4429 ** There are new compiler optimizations for various functions:
4430 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4431 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4432 *** the math functions TRUNCATE, FLOOR, and CEILING
4433 *** the function-of-all-trades COERCE
4434 Mostly these should be transparent, but there's one
4435 potentially-annoying problem (bug 117): when the compiler
4436 inline-expands a function and does type analysis on the result,
4437 it can create control paths which have type mismatches, and
4438 when it can't prove that those control paths aren't taken,
4439 it will issue WARNINGs about the type mismatches. This is
4440 a particular problem in practice for the new sequence functions.
4441 It's not clear how this should be fixed, and for now, a
4442 workaround is given in the entry for 117 in the BUGS file.
4443 ** (Because of the interaction between the two previous items --
4444 occasional inlining problems and new inline expansions -- some
4445 of the new sequence function optimizations won't really kick in
4446 completely until debugging information, and then inlining, are
4447 straightened out in some future version.)
4448 * minor incompatible changes:
4449 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4450 DIRECTORY no longer implicitly promotes NIL slots of its
4451 pathname argument to :WILD. In particular, when you ask for the
4452 contents of a directory (which you used to be able to do without
4453 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4454 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4455 ** changes in behavior that ANSI explicitly defines to be
4456 implementation dependent:
4457 *** The new compiler-only implementation still conforms with ANSI,
4458 but acts a little different than before. Besides the obvious
4459 changes in performance tradeoffs (that the cost per form passed
4460 to EVAL has gone up, and the cost per form executed by EVAL
4461 has gone down), the behavior of the system changes a little
4462 because there are no longer any interpreted function objects.
4463 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4464 e.g. doing COMPILE on the output of interactive DEFUN is
4466 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4468 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4469 has changed from (:CL) to NIL.
4470 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4471 e.g. #\Newline and #\Tab, don't have names specified in the
4472 ANSI Common Lisp standard, is now based on their ASCII symbolic
4473 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4474 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4475 are no longer used for output.
4476 ** changes in internal implementation constants:
4477 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4478 4 million. (If your application spends a lot of time GCing and
4479 you have a lot of RAM, you might want to experiment with
4480 increasing it even more.)
4481 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4482 However, almost all old code should still continue to work without
4483 immediate update, as SB-C-CALL is now a (deprecated) nickname
4485 ** Old operator names in the style DEF-FOO are now deprecated in
4486 favor of new corresponding names DEFINE-FOO, for consistency with
4487 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4488 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4489 internal symbols, but a few supported extensions like
4490 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4491 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4492 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4493 etc. as you get deeper into recursive calls to the debugger
4494 command loop, instead of the old "5]", "5]]", "5]]]"
4495 sequence. (I was motivated to do this when squabbles between
4496 ILISP and SBCL left me very deeply nested in the debugger. In the
4497 short term, this change will probably provoke more ILISP/SBCL
4498 squabbles, but hopefully it will be an improvement in the long run.)
4499 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4500 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4501 function, since declaiming return types involving aliens is
4502 (1) annoyingly messy to do by hand and (2) vital to efficient
4503 compilation of code which calls such functions.
4504 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4505 longer reexported by the SB-EXT package. They're solely useful
4506 for alien code, so it seems more logical that you should get
4507 them from the SB-ALIEN package, not in SB-EXT.
4508 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4509 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4510 features. Instead, the code that they used to control is always
4511 built into the system.
4512 * many other bug fixes
4513 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4514 advantage of the new EVAL-WHEN stuff and to clean them up in
4515 general, and they are now more ANSI-compliant in a number of
4516 ways. Martin Atzmueller is responsible for a lot of this.
4517 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4519 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4520 *** correct ERROR type for various file operations
4521 *** some fixes for Lisp streams
4522 *** DEFMETHOD syntax checking
4523 *** changing old weird representation of debug information as
4524 strings (which, among their other deficiencies, don't transform
4525 correctly when you rename packages, and don't change their
4526 print representation when you change things like *PACKAGE*
4527 and *PRINT-LENGTH*) to symbols and lists of symbols
4528 He also made several improvements and fixed several bugs in DESCRIBE.
4529 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4531 *** misbehavior of WRITE-STRING/WRITE-LINE
4532 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4533 and several other LOOP problems as well
4534 *** DIRECTORY when similar filenames are present
4535 *** DEFGENERIC with :METHOD options
4536 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4537 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4538 *** argument ordering in FIND with :TEST option
4539 *** mishandled package designator argument in APROPOS-LIST
4540 *** various problems in the backquote readmacro
4541 *** a bug in APROPOS
4542 *** probably some others that I'm not describing very well here,
4543 since the CVS log documents them by reference to sbcl-devel
4544 messages, and the SourceForge archives aren't working well.:-(
4545 ** Dan Barlow improved the Alpha port (and is making progress on the
4546 PPC port, for those of you who think different).
4547 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4548 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4549 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4550 continues to work on the SPARC port (for those of you in a position
4551 to look down upon our little PC-compatible boxes from a great height).
4552 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4553 and uses that copy, rather than the current dynamic value, when
4554 it's trying to decide whether to truncate output. Thus e.g.
4555 (let ((*print-lines* 50))
4556 (pprint-logical-block (stream nil)
4558 (let ((*print-lines* 8))
4559 (print (aref possiblybigthings i) stream)))))
4560 should now truncate the logical block only at 50 lines, instead of
4561 often truncating it at 8 lines, as it did before.
4562 * The doc/cmucl/ directory, containing old CMU CL documentation
4563 from the time of the fork, is no longer part of the base system.
4564 SourceForge has shut down its anonymous FTP service, and with it
4565 my original plan for distributing the old CMU CL documentation
4566 there. For now, if you need these files you can download an old
4567 SBCL source release and extract them from it.
4568 * The fasl file version number changed again, for dozens of reasons,
4569 some of which are apparent above.
4571 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4572 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4573 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4574 patch, so that DISASSEMBLE output is much nicer.
4575 * The code in the SB-PROFILE package now seems reasonably stable.
4576 I still haven't decided what the final interface should look like
4577 (I'd like PROFILE to interact cleanly with TRACE, since both
4578 facilities use function encapsulation) but if you have a need
4579 for profiling now, you can probably use it successfully with
4580 the current CMU-CL-style interface.
4581 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4582 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4583 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4584 link now returns the dangling link itself, and for similar
4585 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4586 link itself. (In these cases the old code signalled an error and
4587 looped endlessly, respectively.) Thus, DIRECTORY now works even
4588 in the presence of dangling and cyclic symbolic links.
4589 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4590 is now a supported extension again, since the consensus on
4591 sbcl-devel was that it can be useful for ordinary development
4592 work, not just for debugging SBCL itself.
4593 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4594 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4595 that currently-defined functions might be redefined later with
4596 different return types.
4597 * Hash tables can be printed readably, as inspired by CMU CL code
4598 of Eric Marsden and SBCL code of Martin Atzmueller.
4599 * better error handling in CLOS method combination, thanks to
4600 Martin Atzmueller porting Pierre Mai's CMU CL patches
4601 * more overflow fixes for >16Mbyte I/O buffers
4602 * A bug in READ has been fixed, so that now a single Ctrl-D
4603 character suffices to cause end-of-file on character streams.
4604 In particular, now you only need one Ctrl-D at the command
4605 line (not two) to exit SBCL.
4606 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4608 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4609 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4610 does the right thing.
4611 * The implementation of some type tests, especially for CONDITION
4612 types, is now tidier and maybe faster, due to CMU CL code
4613 originally by Douglas Crosher, ported by Martin Atzmueller.
4614 * Some math functions have been fixed, and there are new
4615 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4616 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4617 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4618 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4619 likely to be built into sbcl-0.7.0.)
4620 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4621 CL, ported to SBCL by Martin Atzmueller)
4622 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4623 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4624 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4625 * There's a new slam.sh hack to shorten the edit/compile/debug
4626 cycle for low-level changes to SBCL itself, and a new
4627 :SB-AFTER-XC-CORE target feature to control the generation of
4628 the after-xc.core file needed by slam.sh.
4629 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4630 COMPILE-FILE is no longer supported, so that now every function
4631 gets an entry point, so that block compilation looks a little
4632 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4633 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4634 returns the number of bytes consed since the system started,
4635 rather than the number consed since the first time the function
4636 was called. (The new definition parallels ANSI functions like
4637 CL:GET-INTERNAL-RUN-TIME.)
4638 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4639 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4640 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4641 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4642 * Fasl file version numbers are now independent of the target CPU,
4643 since historically most system changes which required version
4644 number changes have affected all CPUs equally. Similarly,
4645 the byte fasl file version is now equal to the ordinary
4648 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4649 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4650 is no longer recognized. I apologize for this, because it was
4651 listed in SB-EXT as a supported extension, but I found that
4652 its existing behavior was poorly specified, as well as incorrectly
4653 specified, and it looked like too much of a mess to straighten it
4654 out. I have enough on my hands trying to get ANSI stuff to work..
4655 * many patches ported from CMU CL by Martin Atzmueller, with
4656 half a dozen bug fixes in pretty-printing and the debugger, and
4657 half a dozen others elsewhere
4658 * fixed bug 13: Floating point infinities are now supported again.
4659 They might still be a little bit flaky, but thanks to bug reports
4660 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4661 as flaky as they were.
4662 * The --noprogrammer command line option is now supported. (Its
4663 behavior is slightly different in detail from what the old man
4664 page claimed it would do, but it's still appropriate under the
4665 same circumstances that the man page talks about.)
4666 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4667 are now supported, and enabled by default. Thus, the compiler can
4668 handle many floating point and complex operations much less
4669 inefficiently. (Thus e.g. you can implement a complex FFT
4671 * The compiler now detects type mismatches between DECLAIM FTYPE
4672 and DEFUN better, and implements CHECK-TYPE more correctly, and
4673 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4674 thanks to patches from Martin Atzmueller.
4675 * various fixes to make the cross-compiler more portable to
4676 ANSI-conforming-but-different cross-compilation hosts (notably
4677 Lispworks for Windows, following bug reports from Arthur Lemmens)
4678 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4679 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4680 fixed thanks to Pierre Mai's CMU CL patches.
4681 * fixes to keep the system from overflowing internal counters when
4682 it tries to use i/o buffers larger than 16M bytes
4683 * fixed bug 45a: Various internal functions required to support
4684 complex special functions have been merged from CMU CL sources.
4685 (When I was first setting up SBCL, I misunderstood a compile-time
4686 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4687 * improved support for type intersection and union, fixing bug 12
4688 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4689 more obscure bugs as well
4690 * some steps toward byte-compiling non-performance-critical
4691 parts of the system, courtesy of patches from Martin Atzmueller
4692 * Christophe Rhodes has made some debian packages of sbcl at
4693 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4694 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4695 stable, but are nonetheless usable. When he's ready, I'd be happy
4696 to add them to the SourceForge "File Releases" section. (And if
4697 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4698 * new fasl file format version number (because of changes in
4699 internal representation of (OR ..) types to accommodate the new
4700 support for (AND ..) types, among other things)
4702 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4703 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4705 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4707 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4708 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4709 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4710 * DESCRIBE now works on structure objects again.
4711 * Most function call argument type mismatches are now handled as
4712 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4713 know whether the function will be redefined before the call is
4714 executed. (The compiler could flag local calls with full WARNINGs,
4715 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4716 it doesn't keep track of enough information to know whether calls
4717 are local in this sense.)
4718 * Compiler output is now more verbose, with messages truncated
4719 later than before. (There should be some supported way for users
4720 to override the default verbosity, but I haven't decided how to
4721 provide it yet, so this behavior is still controlled by the internal
4722 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4723 src/compiler/ir1util.lisp.)
4724 * Fasl file format version numbers have increased again, because
4725 support for the Gray streams extension changes the layout of the
4726 system's STREAM objects.
4727 * The Gray subclassable streams extension now works, thanks to a
4728 patch from Martin Atzmueller.
4729 * The full LOAD-FOREIGN extension (not just the primitive
4730 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4731 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4732 but like most other programs, it defaults to copying the Unix
4733 environment from the original process instead of starting the
4734 new process in an empty environment.
4735 * Extensions which manipulate the Unix environment now support
4736 an :ENVIRONMENT keyword option which doesn't smash case or
4737 do other bad things. The CMU-CL-style :ENV option is retained
4738 for porting convenience.
4739 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4740 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4742 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4744 * A patch from Martin Atzmueller seems to have solved the SIGINT
4745 problem, and as far as we know, signal-handling now works cleanly.
4746 (If you find any new bugs, please report them!)
4747 * The system no longer defaults Lisp source file names to types
4748 ".l", ".cl", or ".lsp", but only to ".lisp".
4749 * The compiler no longer uses special default file extensions for
4750 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4751 seems to expect a single default extension for all compiled code,
4752 and there's no compelling reason to try to stretch the standard
4753 to allow two different extensions.) Instead, byte-compiled files
4754 default to the same extension as native-compiled files.
4755 * Fasl file format version numbers have increased again, because
4756 a rearrangement of internal implementation packages made some
4757 dumped symbols in old fasl files unreadable in new cores.
4758 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4759 many fewer weird special cases.
4760 * Bug #17 (differing COMPILE-FILE behavior between logical and
4761 physical pathnames) has been fixed, and some related misbehavior too,
4762 thanks to a patch from Martin Atzmueller.
4763 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4764 by Tim Moore, ported to SBCL by Martin Atzmueller.
4765 * Martin Atzmueller fixed several filesystem-related problems,
4766 including bug #36, in part by porting CMU CL patches, which were
4767 written in part by Paul Werkowski.
4768 * More compiler warnings in src/runtime/ are gone, thanks to
4769 more patches from Martin Atzmueller.
4770 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4773 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4775 * DESCRIBE now works on CONDITION objects.
4776 * The debugger now handles errors which arise when trying to print
4777 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4779 * The build system now uses an additional file, customize-target-features.lisp,
4780 to allow local modifications to the target *FEATURES* list. (The point of
4781 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4782 debugging features enabled, without having to worry about propagating it
4783 into everyone's system when I do a "cvs update".) When no
4784 customize-target-features.lisp file exists, the target *FEATURES* list
4785 should be constructed the same way as before.
4786 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4787 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4788 warn about behavior which is undefined under the ANSI spec. Note: This
4789 is specified by ANSI, but it's not very popular with programmers.
4790 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4791 macro in the SBCL sources for an example of a workaround which you
4792 might use to make portable ANSI-standard code which does what you want.
4793 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4794 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4795 when it's used as a non-toplevel form. (This is required in order
4796 to implement the DEFCONSTANT-EQX macro.)
4797 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4798 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4799 funny things, due to bugs in the implementation of EVAL-WHEN
4800 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4801 non-toplevel DEFCONSTANTs are uncommon.)
4802 * The core file version number and fasl file version number have been
4803 incremented, because the old noncompliant DEFCONSTANT behavior involved
4804 calling functions which no longer exist, and because I also took the
4805 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4806 * fixed bug 1 (error handling before read-eval-print loop starts), and
4807 redid debugger restarts and related debugger commands somewhat while
4809 ** The QUIT debugger command is gone, since it did something
4810 rather different than the SB-EXT:QUIT command, and since it never
4811 worked properly outside the main toplevel read/eval/print loop.
4812 Invoking the new TOPLEVEL restart provides the same functionality.
4813 ** The GO debugger command is also gone, since you can just invoke
4814 the CONTINUE restart directly instead.
4815 ** The TOP debugger command is also gone, since it's redundant with the
4816 FRAME 0 command, and since it interfered with abbreviations for the
4818 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4820 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4821 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4822 some porting work by Martin Atzmueller)
4823 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4824 with a boatload of patches which he ported to SBCL. Now that those
4826 ** The system tries to make sure that its low-priority messages
4827 are prefixed by semicolons, to help people who like to use
4828 syntax highlighting in their ILISP buffer. (This patch
4829 was originally due to Raymond Toy.)
4830 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4831 patches originally written by Raymond Toy.
4832 ** The compiler understands coercion between single-value and
4833 multiple-VALUES type expressions better, getting rid of some very
4834 weird behavior, thanks to patches originally by Robert MacLachlan
4835 and Douglas Crosher.
4836 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4837 lambda lists, thanks to a patch originally by Pierre Mai.
4838 ** The system no longer bogusly warns about "abbreviated type
4840 ** The compiler gets less confused by inlining and RETURN-FROM,
4841 thanks to some patches originally by Tim Moore.
4842 ** The system no longer hangs when dumping circular lists to fasl
4843 files, thanks to a patch originally from Douglas Crosher.
4844 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4845 undefined function error.
4846 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4847 in structure slot initforms) for now, documented workaround instead:-|
4848 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4849 * fixed bug 5: added stubs for various Gray stream functions called
4850 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4851 installed, at least appropriate type errors are generated
4852 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4853 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4854 consistently in DEFMETHOD forms.
4855 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4856 it doesn't seem to affect SBCL after all
4857 * The C runtime system now builds with better optimization and many
4858 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4860 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4862 * The system is now under CVS at SourceForge (instead of the
4863 CVS repository on my home machine).
4864 * The new signal handling code has been tweaked to treat register
4865 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4866 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4867 downstream, e.g. in debug-int.lisp, has implicit dependencies
4868 on the unsignedness of integer representation of machine words,
4869 and that caused the system to bomb out with infinite regress
4870 when trying to recover from type errors involving signed values,
4871 e.g. (BUTLAST '(1 2 3) -1).)
4872 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4873 (This was one of the bugs Peter Van Eynde reported back in July.)
4874 * The system now uses code inspired by Colin Walters' O(N)
4875 implementation of MAP (from the cmucl-imp@cons.org mailing
4876 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4877 inline the MAP operation, and there is more than one
4878 sequence argument to the MAP call (so that it can't just
4879 do ETYPECASE once and for all based on the type of the
4880 single sequence argument). (The old non-inline implementation
4881 of the general M-argument sequence-of-length-N case required
4882 O(M*N*N) time when any of the sequence arguments were LISTs.)
4883 * The QUIT :UNIX-CODE keyword argument has been renamed to
4884 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4885 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4887 * Raymond Wiker's patches to port dynamic loading from Linux to
4888 FreeBSD have been added.
4889 * The BUGS file is now more nearly up to date, thanks in large part
4890 to Martin Atzmueller's review of it.
4891 * The debugger now flushes standard output streams before it begins
4892 its output ("debugger invoked" and so forth).
4893 * The core version number and fasl file version number have both
4894 been incremented, because of incompatible changes in the layout
4896 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4897 used to not be called for a saved Lisp image.)
4898 * Martin Atzmueller's version of a patch to fix a compiler crash,
4899 as posted on sbcl-devel 13 September 2000, has been installed.
4900 * Instead of installing Martin Atzmueller's patch for the
4901 compiler transform for SUBSEQ, I deleted the compiler transform,
4902 and transforms for some similar consing operations.
4903 * A bug in signal handling which kept TRACE from working on OpenBSD
4905 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4906 away by constant folding
4907 * The system now defines its address space constants in one place
4908 (in the Lisp sources), and propagates them automatically elsewhere
4909 (through GENESIS and the sbcl.h file). Therefore, patching the
4910 address map is less unnecessarily tedious and error-prone. The
4911 Lisp names of address space constants have also been systematized.
4912 * CVS tags like dollar-Header-dollar have been removed from
4913 the sources, because they have never saved me trouble and
4914 they've been source of trouble working with patches and other
4915 diff-related operations.
4916 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4917 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4919 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4921 * The system has been ported to OpenBSD.
4922 * The system now compiles with a simple "sh make.sh" on the systems
4923 that it's supported on. I.e., now you no longer need to tweak
4924 text in the target-features.lisp-expr and symlinks in src/runtime/
4925 by hand, the make.sh takes care of it for you.
4926 * The system is no longer so grossly inefficient when compiling code
4927 involving vectors implemented as general (not simple) vectors (VECTOR T),
4928 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4929 which dares to use the various sequence functions on non-simple
4930 vectors, takes less of a performance hit.
4931 * There is now a primitive type predicate VECTOR-T-P
4932 to test for the (VECTOR T) type, so that e.g.
4933 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4934 can now be compiled with some semblance of efficiency. (The old code
4935 turned the type declaration into a full call to %TYPEP at runtime!)
4936 * AREF on (VECTOR T) is still not fast, since it's still compiled
4937 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4938 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4939 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4940 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4941 without making full calls to SUBTYPEP at runtime.
4942 (Some analogous efficiency issues for non-simple vectors specialized to
4943 element types other than T, or for non-simple multidimensional arrays,
4944 have not been addressed. They could almost certainly be handled the
4945 same way if anyone is motivated to do so.)
4946 * The changes in array handling break binary compatibility, so
4947 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4948 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4949 instead of (VALUES T T).
4950 * By following the instructions that Dan Barlow posted to sbcl-devel
4951 on 2 July 2000, I was able to enable primitive dynamic object
4952 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4953 functionality is not implemented (since it calls ld to resolve
4954 library references automatically, requiring RUN-PROGRAM for its
4955 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4956 try to resolve library references) is now supported.
4957 * The system now flushes the standard output streams when it terminates,
4958 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4959 them at several other probably-convenient times, e.g. in each pass of
4960 the toplevel read-eval-print loop, and after evaluating a form given
4961 as an "--eval" command-line option. (These changes were motivated by a
4962 discussion of stream flushing issues on cmucl-imp in August 2000.)
4963 * The source transform for TYPEP of array types no longer assumes
4964 that an array whose element type is a not-yet-defined type
4965 is implemented as an array of T, but instead punts, so that the
4966 type will be interpreted at runtime.
4967 * There is now some support for cross-compiling in make.sh: each of
4968 the phases of make.sh has its own script. (This should be transparent
4969 to people doing ordinary, non-cross-compile builds.)
4970 * Since my laptop doesn't have hundreds of megabytes of memory like
4971 my desktop machine, I became more motivated to do some items on
4972 my to-do list in order to reduce the size of the system a little:
4973 ** Arrange for various needed-only-at-cold-init things to be
4974 uninterned after cold init. To support this, those things have
4975 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4976 symbols with such names are now uninterned after cold init).
4977 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4978 fasl files for cold load.
4979 ** Remove the old compiler structure pooling code (which used to
4980 be conditional on the target feature :SB-ALLOC) completely.
4981 ** Redo the representation of some data in cold init to be more compact.
4982 (I also looked into supporting byte compiled code at bootstrap time,
4983 which would probably reduce the size of the system a lot, but that
4984 looked too complicated, so I punted for now.)
4985 * The maximum signal nesting depth in the src/runtime/ support code has
4986 been reduced from 4096 to 256. (I don't know any reason for the very
4987 large old value. If the new smaller value turns out to break something,
4988 I'll probably just bump it back up.)
4989 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4991 * Many, many bugs reported by Peter Van Eynde have been added to
4992 the BUGS list; some have even been fixed.
4993 * While enabling dynamic object file loading, I tried to make the
4994 code easier to understand, renaming various functions and variables
4995 with less ambiguous names, and changing some function calling
4996 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4997 * While trying to figure out how to do the OpenBSD port, I tried to
4998 clean up some of the code in src/runtime/. In particular, I dropped
4999 support for non-POSIX signal handling, added various comments,
5000 tweaked the code to reduce the number of compilation warnings, and
5001 renamed some files to increase consistency.
5002 * To support the new automatic configuration functionality in make.sh,
5003 the source file target-features.lisp-expr has been replaced with the
5004 source file base-target-features.lisp-expr and the machine-generated
5005 file local-target-features.lisp-expr.
5006 * fixed a stupid quoting error in make.sh so that using CMU CL
5007 "lisp -batch" as cross-compilation host works again
5009 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
5011 * DESCRIBE no longer tries to call itself recursively to describe
5012 bound/fbound values, so that it no longer fails on symbols which are
5013 bound to themselves (like keywords, T, and NIL).
5014 * DESCRIBE now works on generic functions.
5015 * The printer now prints less-screwed-up representations of closures
5016 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
5017 * A private symbol is used instead of the :EMPTY keyword previously
5018 used to mark empty slots in hash tables. Thus
5019 (DEFVAR *HT* (MAKE-HASH-TABLE))
5020 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
5021 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
5022 now does what ANSI says that it should. (You can still get
5023 similar noncompliant behavior if bang on the hash table
5024 implementation with all the symbols you get back from
5025 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
5026 This breaks binary compatibility, since tests for equality to
5027 :EMPTY are wired into things like the macroexpansion of
5028 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
5030 * There's now a minimal placeholder implementation for CL:STEP,
5031 as required by ANSI.
5032 * An obscure bug in the interaction of the normal compiler, the byte
5033 compiler, inlining, and structure predicates has been patched
5034 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
5035 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
5037 * Missing ordinary arguments in a macro call are now detected even
5038 when the macro lambda list contains &KEY or &REST.
5039 * The debugger no longer complains about encountering the top of the
5040 stack when you type "FRAME 0" to explicitly instruct it to go to
5041 the top of the stack. And it now prints the frame you request even
5042 if it's the current frame (instead of saying "You are here.").
5043 * As specified by ANSI, the system now always prints keywords
5044 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
5046 * The default initial SIZE of HASH-TABLEs is now smaller.
5047 * Type information from CLOS class dispatch is now propagated
5048 into DEFMETHOD bodies, so that e.g.
5049 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5051 is now basically equivalent to
5052 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5053 (DECLARE (TYPE SINGLE-FLOAT X))
5055 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
5056 operation, without having to do run-time type dispatch.
5057 * The macroexpansion of DEFMETHOD has been tweaked so that it has
5058 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
5059 * Since I don't seem to be making big file reorganizations very often
5060 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
5061 is overflowing my ability to conveniently back them up), I've finally
5062 checked the system into CVS. (The CVS repository is on my home system,
5063 not at SourceForge -- putting it on SourceForge might come later.)
5064 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
5065 high-level GC-NOTIFY-FOO functions send their output. (There's
5066 still very little control of where low-level verbose GC functions
5067 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
5068 less than it used to -- the GC-NOTIFY-FOO functions are now under
5069 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
5070 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
5071 in only one place in the source code, and propagates it automatically
5072 everywhere that it's needed. Thus e.g. when I bump the version from
5073 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
5074 * The C source files now include boilerplate legalese and documentation
5075 at the head of each file (just as the Lisp source files already did).
5076 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
5077 to his page will be replaced with a link to his new CLiki service.
5079 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
5081 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
5082 * The build process now looks for GNU make under the default name "gmake",
5083 instead of "make" as it used to. If GNU make is not available as "gmake"
5084 on your system, you can change this default behavior by setting the
5085 GNUMAKE environment variable.
5086 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
5087 can build without error under CMU CL.
5089 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
5091 * There is now a partial SBCL user manual (with some new text and some
5092 text cribbed from the CMU CL manual).
5093 * The beginnings of a profiler have been added (starting with the
5094 CMU CL profiler and simplifying and cleaning up). Eventually the
5095 main interface should be through the TRACE macro, but for now,
5096 it's still accessed through vaguely CMU-CL-style functions and macros
5097 exported from the package SB-PROFILE.
5098 * Some problems left over from porting CMU CL to the new
5099 cross-compilation bootstrap process have been cleaned up:
5100 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
5101 instead of SB!XC:DEFMACRO, compounded by an oversight on my
5102 part when getting rid of the compiler *BACKEND* stuff.)
5103 ** The value of *NULL-TYPE* was screwed up, because it was
5104 being initialized before the type system knew the final
5105 definition of the 'NULL type. This screwed up several key
5106 optimizations in the compiler, causing inefficiency in all sorts
5107 of places. (I found it because I wanted to understand why
5108 GET-INTERNAL-RUN-TIME was consing.)
5109 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
5110 PROCLAIM FTYPE information. Unfortunately this broke binary
5111 compatibility again, since now the forms output by DEFGENERIC
5112 to refer to functions which didn't exist in 0.6.3.
5113 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
5114 can use the new (as of 0.6.3) transform for SOME into MAP into
5116 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
5117 values to (INTEGER 0 1000000), so that the time code will no longer
5118 occasionally get blown up by Linux returning 1000000 microseconds
5119 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
5120 its output conform to the ANSI spec. (Alas, this makes its output
5121 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
5122 * A full call to MAP NIL with a single sequence argument no longer conses.
5123 * fixes to problems pointed out by Martin Atzmueller:
5124 * The manual page no longer talks about multiprocessing as though
5125 it were currently supported.
5126 * The ILISP support patches have been removed from the distribution,
5127 because as of version 5.10.1, ILISP now supports SBCL without us
5128 having to maintain patches.
5129 * added a modified version of Raymond Toy's recent CMU CL patch for
5130 EQUALP comparison of HASH-TABLE
5132 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
5134 * The system still can't cross-compile itself with
5135 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
5136 entails), but at least it can compile more of itself that way
5137 than it used to be able to, and various buglets which were uncovered
5138 by trying to cross-compile itself that way have now been fixed.
5139 * This release breaks binary compatibility again. This time
5140 at least I've incremented the FASL file format version to 2, so that the
5141 problem can be detected reliably instead of just causing weird errors.
5142 * various new style warnings:
5143 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
5144 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
5145 ** using the lexical binding of a variable named in the *FOO* style
5146 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
5147 as specified by ANSI.
5148 * *RANDOM-STATE* is no longer automatically initialized from
5149 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
5150 default behavior of the system is to repeat its behavior every time
5151 it's run. If you'd like to change this behavior, you can always
5152 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
5153 old convention there was no comparably easy way to get the system to
5154 repeat its behavior every time it was run.
5155 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
5156 FTYPE declarations has been removed, in favor of their ANSI
5157 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
5158 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
5159 the types of their sequence arguments can be determined at compile time.
5160 This is done through a new open code expansion for MAP which eliminates
5161 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
5162 argument types can be determined at compile time.
5163 * The optimizer now transforms COERCE into an identity operation when it
5164 can prove that the coerced object is already of the correct type. (This
5165 can be a win for machine generated code, including the output of other
5166 optimization transforms, such as the MAP transform above.)
5167 * Credit information has been moved from source file headers into CREDITS.
5168 * Source file headers have been made more standard.
5169 * The CASE macro now compiles without complaining even when it has
5172 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
5174 * (Note that the way that the PCL macroexpansions were rewritten
5175 to accommodate the change in DEFGENERIC below breaks binary
5176 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
5177 not run under sbcl-0.6.2. Once we get out of alpha releases,
5178 i.e. hit release 1.0.0, we'll probably try to maintain binary
5179 compatibility between maintenance releases, e.g. between sbcl-1.4.3
5180 and sbcl-1.4.4. Until then, however, it might be fairly common
5181 for maintenance releases to break binary compatibility.)
5182 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
5184 * The reporting of unhandled signals has been changed to print some
5185 explanatory text as well as the report form. (Previously only
5186 the report form was printed.)
5187 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
5188 it defines, so that the compiler no longer issues undefined function
5189 warnings for compiled-but-not-yet-loaded generic functions.
5190 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
5191 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
5192 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
5193 section of the ANSI spec are supported.
5194 * The "" nickname for the "KEYWORD" package has been removed.
5195 The reader still handles symbol tokens which begin with a package marker
5196 as keywords, but it doesn't expose its mechanism for doing so in the
5197 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
5198 * The system now issues STYLE-WARNINGs for contradictory TYPE
5199 proclamations. (Warnings for contradictory FTYPE proclamations would
5200 be nice too, but those can't be done usefully unless the type system
5201 is made smarter about FUNCTION types.)
5202 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
5203 systematized, so that "*target-*.lisp is supposed to exist only on the
5204 target and imply that there's a related file which exists on the
5205 host, and *host-*.lisp is supposed to exist only on the host and imply
5206 that there's a related file which exists on the target. This involves a
5207 lot of renaming. Hopefully the acute confusion caused by the renaming
5208 will be justified by the reduction in chronic confusion..
5209 ** runtime-type.lisp -> early-target-type.lisp
5210 ** target-type.lisp -> late-target-type.lisp
5211 ** early-host-format.lisp -> early-format.lisp
5212 ** late-host-format.lisp -> late-format.lisp
5213 ** host-error.lisp -> misc-error.lisp
5214 ** early-error.lisp -> early-target-error.lisp
5215 ** late-error.lisp -> late-target-error.lisp
5216 ** host-defboot.lisp -> early-defboot.lisp
5217 ** code/misc.lisp -> code/target-misc.lisp
5218 ** code/host-misc.lisp -> code/misc.lisp
5219 ** code/numbers.lisp -> code/target-numbers.lisp
5220 ** code/early-numbers.lisp -> numbers.lisp
5221 ** early-host-type.lisp -> early-type.lisp
5222 ** late-host-type.lisp -> late-type.lisp
5223 ** host-typep.lisp -> typep.lisp
5224 ** load.lisp -> target-load.lisp
5225 ** host-load.lisp -> load.lisp
5226 ** host-disassem.lisp -> disassem.lisp
5227 ** host-insts.lisp -> insts.lisp
5228 ** byte-comp.lisp -> target-byte-comp.lisp
5229 ** host-byte-comp.lisp -> byte-comp.lisp
5230 ** host-signal.lisp -> signal.lisp
5231 ** host-defstruct.lisp -> defstruct.lisp
5232 ** late-target-type.lisp -> deftypes-for-target.lisp
5233 Furthermore, several other previously target-only files foo.lisp (e.g.
5234 hash-table.lisp and random.lisp) have been split into a target-and-host
5235 foo.lisp file and a target-only target-foo.lisp file, with their key type
5236 definitions in the target-and-host part, so that the cross-compiler will
5237 know more about target types.
5238 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5239 gone away. In their place are various *BACKEND-FOO* variables
5240 corresponding to the slots of the old structure.
5241 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5242 into the target SBCL has been fixed.
5243 * The chill.lisp system for loading cold code into a running SBCL
5245 * Support for the CMU CL "scavenger hook" extension has been removed.
5246 (It was undocumented and unused in the CMU CL sources that SBCL was
5247 derived from, and stale in sbcl-0.6.1.)
5248 * Various errors in the cross-compiler type system were detected
5249 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5250 (enabling various consistency checks). Many of them were fixed,
5251 but some hard problems remain, so the compiler is back to
5252 running without *TYPE-SYSTEM-INITIALIZED* for now.
5253 * As part of the cross-compiler type system cleanup, I implemented
5254 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5255 * I have started adding UNCROSS calls throughout the type system
5256 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5257 on cross-compiler input files will be able to go away, and various
5259 * CONSTANTP now returns true for quoted forms (as explicitly required
5262 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5264 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5265 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5266 (MAKE-LIST -1). In the longer term, it ought to become true
5267 that declarations are assertions even at SAFETY 1. For now, it's not
5268 quite true even at SAFETY 3, but it's at least more nearly true..
5269 (Note that this change seems to increases the size of the system by
5270 O(5%) and to decrease the speed of the compiler by 20% or more.)
5271 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5272 for the problem of printing dozens of lines of distracting information
5273 about low-level system machinery as part of the top stack frame
5274 on entry to the debugger when an undefined function was called.
5275 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5276 is not reset to COMMON-LISP-USER.
5277 * Compilation of stuff related to dyncount.lisp has been made conditional
5278 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5279 smaller. The various dyncount-related symbols have been moved into
5280 a new "SB-DYNCOUNT" package.
5281 * tty-inspect.lisp has been renamed to inspect.lisp.
5282 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5283 feature has gone away. (When we eventually port to other flavors of
5284 libc and/or Unix, we'll try to make the differences between flavors
5285 invisible at the user level.)
5286 * Various other *FEATURES* tags, and/or their associated conditionals,
5287 have been removed if obsolescent, or given better documentation, or
5288 sometimes given more-mnemonic names.
5290 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5292 * tidied up "make.sh" script
5293 * tidied up system directory structure
5294 * better "clean.sh" behavior
5295 * added doc/FOR-CMUCL-DEVELOPERS
5296 * many many small tweaks to output format, e.g. removing possibly-confusing
5297 trailing #\. character in DESCRIBE-INSTANCE
5298 * (EQUALP #\A 'A) no longer signals an error.
5299 * new hashing code, including EQUALP hashing
5300 * tidied up Lisp initialization and toplevel
5301 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5302 * command line argument processing
5303 * added POSIX-GETENV function to deal with Unix-ish environment variables
5304 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5305 terminating SBCL on EOF
5306 * non-verbose GC by default
5307 * There is no more "sbcl" shell script; the sbcl file is now the C
5308 runtime executable (just like CMU CL).
5309 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5311 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5312 symbol and package information as Lisp native symbol and package objects
5313 instead of strings naming symbols and strings naming packages. This way,
5314 whenever packages are renamed (as in warm init), debug information is
5315 transformed along with everything else.
5316 * tweaked the optimization policy declarations which control the building
5317 of SBCL itself. Now, among other things, the system no longer saves
5318 source location debugging information. (This helps two problems at once
5319 by reducing SBCL size and by keeping SBCL from trying to look for its
5320 sources -- which may not exist -- when reporting errors.)
5321 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5322 debugging and testing purposes
5323 * cleaned up printing, making the printer call PRINT-OBJECT for
5324 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5325 methods, giving nearly-ANSI behavior
5326 * converted almost all special variables to use *FOO* naming convention
5327 * deleted PARSE-TIME functionality, since it can be done portably
5328 * moved some files out of cold init into warm init
5329 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5331 * regularized formatting of source files
5332 * added an install.sh script
5333 * fixed ridiculous memory usage of cross-compiler by making
5334 compiler/alloc.lisp not try to do pooling unless it can hook
5335 itself into the GC of the cross-compilation host. Now the system
5336 builds nicely on my old laptop.
5337 * added :SB-ALLOC in target-features.lisp-expr
5338 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5339 was not implemented)
5340 * re-did condition handling and note reporting in the compiler. Notes
5341 are no longer handled by signalling conditions. Style warnings
5342 and warnings are handled more correctly and reported in such a way
5343 that it's easy to find one or the other in your output (so that you
5344 can e.g. figure out which of many problems caused COMPILE-FILE to
5346 * changed the severity of several compiler warnings from full WARNING
5347 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5348 compiler note reporting so that it doesn't use the condition system
5349 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5350 COMPILE-FILE command)
5351 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5352 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5353 no longer supported, since their implementation was deeply intertwingled
5354 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5355 * removed bogus "support" for compiler macros named (SETF FOO), and
5356 removed the compiler macro for SETF INFO (but only after making a fool
5357 of myself on the cmucl-imp mailing list by posting a bogus patch for
5358 DEFINE-COMPILER-MACRO..)
5359 * Compiled files containing forms which have side effects on the Lisp
5360 reader (such as DEFPACKAGE forms) are now handled more correctly.
5361 (Compiler queuing of top level lambdas has been suppressed by setting
5362 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5363 * deleted various currently-unused source files, e.g. gengc.lisp. They
5364 may be added back at some point e.g. when porting to other architectures,
5365 but until they are it's distracting to distribute them and to try to
5367 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5368 there were so many of them they're just distractions, and UNCROSS is
5369 known to be able to handle the current sources
5370 * moved PROFILE functionality into TRACE, so that it will be clear
5371 how the wrapping and unwrapping of functions when you profile them
5372 interacts with the wrapping and unwrapping of functions when you
5373 trace them. (Actually, the functionality isn't there yet, but at least
5374 the interface specification is there. Hopefully, the functionality will
5375 arrive with some future maintenance release.)
5376 * removed host-oops.lisp
5377 * changed signature of QUIT function to allow UNIX-CODE argument
5378 * fixed READ-SEQUENCE bug
5379 * tweaked verbose GC output so that it looks more like the progress
5380 output that ANSI specifies for functions like LOAD
5381 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5382 * added <http://sbcl.sourceforge.com> to the banner information printed by